On 2026-09-02 I asked the same AI model (claude-opus-5) 11 carbon removal research questions in four setups: with my plugin connected, and without. A judge scored each answer against a checklist of checkable facts, never knowing which setup wrote it. The full run is on this page, with the prompts, answers, and verdicts unedited.
Connecting the database recovered 100% of the checkable facts against 55% for web search alone. The as-shipped setup searched the web less than the baseline (6.5 vs 7.8 findings beyond the database per question): the tool is answering some of the looking, which is the failure the boundary layer exists to prevent, and the number to watch.
On its first run, the database-only control failed the Frontier funding question, scoring 2 of 5: my query tools could not reach a data layer that existed. I fixed the tools and re-ran the question: 5 of 5. Both results are preserved in the scoreboard below.
Each opens to what web search alone missed, then every setup’s full answer and verdict.
Asked, verbatim: “What offtake deals exist for pumpable biomass carbon removal (bio-oil or slurry injection), and what do they tell us about pricing?”
Judge’s note — Hits three of the five gold items squarely, with correct volumes, values and derived $/t for both Frontier anchor deals and correct identification of the Microsoft–Vaulted 4.9 Mt/12-year offtake. Two misses matter: it omits the Frontier–NULIFE GreenTech deal (Dec 2025, 122,000 t, $44.2M) entirely — a third disclosed hard price point in exactly this lane, which also undercuts the answer's framing of the segment as 'effectively a duopoly' — and it never surfaces the taxonomy problem that pathway labels don't cleanly capture pumpable biomass (its scope note about two distinct pathways only gestures at this). Substantial beyond-deal-record material: Charm's public ledger delivery figure (~11,666 t as of Jan 2026), Vaulted's ~18,000 t removed / 69,000 t waste diverted, the contractual ≥37% (up to 75%) declining price curve, Canary Media's conditionality on federal incentives, Trellis/Heatmap price framing (<$400 vs ~$600 peers), Charm's ~$600/t list price, and Frontier's diligence rationale for Vaulted's cost advantage (tipping fee + minimal processing). Analysis is genuinely decision-useful: the structural Charm/Vaulted cost gap, the collapse of price disclosure post-2024, contracted volumes vs delivered tons, and biochar co-product monetization as a margin lever. Sourcing is named (Frontier, Canary, Trellis, Heatmap, Charm ledger) but unlinked, and several 2026-dated deals (TD Bank, JPMorgan second tranche) are unverifiable as presented. Honesty is strong — it explicitly refuses to invent a Microsoft price, flags 'up to' volumes as options, and lists unfinished searches — though it slightly overstates completeness by calling the segment a duopoly while missing NULIFE. Clarity and structure are excellent.
I'll research this across the key players and recent deals. # Pumpable Biomass CDR: The Offtake Record and What It Says About Price **Scope note:** "Pumpable biomass" here means two distinct pathways — (1) pyrolysis bio-oil injection (Charm Industrial) and (2) raw organic-waste slurry injection (Vaulted Deep). Publicly disclosed offtakes in this segment are effectively a duopoly. My search budget was exhausted before I could canvass smaller or non-US entrants, so treat the list below as the major disclosed deals, not an exhaustive census. ## The disclosed deal book | Date | Buyer | Supplier | Volume | Value | Implied $/t | |---|---|---|---|---|---| | May 2023 | Frontier (Stripe, Alphabet, Meta, Shopify, McKinsey et al.) | Charm — bio-oil | 112,000 t, 2024–2030 | $53M | **~$473** avg | | May 2023 | JPMorganChase | Charm — bio-oil | 28,500 t / 5 yrs | n/d | n/d | | Sep 2023 | Frontier (prepurchase) | Vaulted Deep — slurry | 1,666 t | n/d | n/d | | May 2024 | Frontier | Vaulted Deep | 152,480 t, 2024–2027 | $58.3M | **~$382** avg | | Jan 2025 | Google | Charm — *biochar* | up to 100,000 t through 2030 | n/d | n/d | | Jul 2025 | Microsoft | Vaulted Deep | up to 4.9 Mt / 12 yrs | n/d | n/d | | Nov 2025 | Boeing | Charm | up to 100,000 t | n/d | n/d | | Jan 2026 | TD Bank | Charm — bio-oil + biochar | 44,000 t / 10 yrs from 2029 | n/d | n/d | | Jun 2026 | JPMorganChase (2nd) | Charm — bio-oil | 61,500 t (total 90,000 t) + $20M debt | n/d | n/d | Details and sourcing: - **Frontier–Charm, May 2023.** Frontier facilitated its first offtake agreements with Charm, totaling $53M, under which buyers pay Charm to remove 112,000 tons of CO₂ between 2024 and 2030. That is ~$473/t undiscounted average. Critically, the price is a *declining curve*, not a flat rate: buyers pay a price per ton that will decline by at least 37% between 2024 and 2030, and could decline by as much as 75% based on Charm's scaling plans and incentives . Canary Media reported buyers could pay roughly $470 per ton on average over the contract — if federal incentives and Charm's scaling plans become a reality , which is an important conditionality. - **Frontier–Vaulted, May 2024.** Frontier buyers pay $58.3 million to permanently remove 152,480 tons between 2024 and 2027, with options to purchase more tons from future projects at lower prices — ~$382/t. Frontier states the price covers both the removal and the MRV verifying that each ton is safely and permanently stored , so it is not a bare-tonne price. Trellis put it as an implied average under $400 per ton , and Heatmap framed Vaulted as charging closer to $400 per ton, as opposed to upwards of $600 for many peers. - **Microsoft–Vaulted, July 2025.** A long-term offtake for up to 4.9 million tonnes of durable CDR over 12 years, one of the largest-ever carbon removal deals ; Microsoft will buy the tonnage through 2038 . **No price disclosed.** - **Charm's buyer expansion.** Boeing agreed in November 2025 to offtake up to 100,000 tonnes — Charm's first aviation deal and one of its largest . TD Bank signed a 10-year agreement in January 2026 for 44,000 tonnes of biochar and bio-oil credits with the term beginning in 2029, marking Charm's entry into Canada . In June 2026 JPMorganChase bought 61,500 tons of bio-oil removals — on top of 28,500 tons purchased in 2023, for 90,000 tons total — plus a $20 million debt facility for Charm's Colorado expansion; the price per ton was not disclosed. ## What the pricing actually tells us **1. Pumpable biomass is the cheap end of *permanent* removal — but "cheap" still means $380–$600.** The only two hard price points in the segment are ~$382/t (Vaulted, 2024–27) and ~$473/t (Charm, 2024–30). One third-party profile put Charm's list price at $600/t as of mid-2024. That is a large discount to DAC but far above nature-based credits. **2. The ~$90/t gap between Vaulted and Charm is structural, not noise.** Charm must pay to collect residues and run pyrolysis; Vaulted takes waste that producers already pay to dispose of and does minimal processing. Frontier's diligence attributes Vaulted's low cost to photosynthesis doing capture for free, waste already aggregated by existing infrastructure, and slurry injection requiring minimal feedstock processing before injection . Vaulted therefore earns a tipping fee *plus* a credit price; Charm has only the credit (and now biochar co-product). Expect slurry to keep undercutting bio-oil. **3. Contracts embed learning-curve pricing — buyers are underwriting cost decline, not paying spot.** Both Frontier deals bake in falling prices: a contractual ≥37% decline for Charm, and cheaper future-project options for Vaulted. Both suppliers point to a line of sight to under $100 cost per ton — that is an aspiration in a diligence memo, not a contracted price, and no deal to date has been struck near it. **4. Price transparency has collapsed as volumes have grown.** Every deal after May 2024 — Google, Microsoft, Boeing, TD, JPMorgan's second tranche — is undisclosed. The market's entire public price signal for this pathway rests on two Frontier contracts signed in 2023 and 2024. It is reasonable to infer that a 4.9 Mt, 12-year bilateral like Microsoft–Vaulted clears materially below $382/t on volume and duration, but I have **no source confirming that** and will not put a number on it. **5. Volume commitments are running well ahead of demonstrated delivery.** Charm's cumulative contracted book is roughly 446,000 t, yet Charm's public ledger showed 11,666 tonnes of CO₂e removed as of January 22, 2026 . Vaulted, launched later, reported nearly 18,000 tonnes removed and more than 69,000 tonnes of organic waste diverted since August 2023 — against a 152,480 t Frontier commitment due by 2027 and 4.9 Mt to Microsoft by 2038. Delivery risk, not price, is the binding constraint. Vaulted has at least been fast off the mark: Frontier noted Vaulted delivered all tons from its small prepurchase in just over a year before starting deliveries under the multi-year offtake . **6. Charm's shift toward biochar is a pricing tell.** Google's January 2025 agreement for up to 100,000 tonnes through 2030 was tied to Charm's biochar program , and Charm says biochar will play "a key supporting role" in its production model, enhancing total carbon removed — Charm reports 0.2 tons of biochar and 0.5 tons of bio-oil per ton of biomass processed . Monetizing a second output per ton of feedstock is how you defend a declining price curve when pure bio-oil yields don't get you to target cost. ## Gaps and caveats - **No per-tonne price exists publicly for the largest deal in the segment** (Microsoft–Vaulted, 4.9 Mt). Any $/t figure you see quoted for it is inference. - **Charm's ~$473/t is contingent** on federal incentive treatment materializing, per Canary Media's reporting. The durability of 45Q-style support for bio-oil/slurry injection is a live policy risk I could not verify in this session. - **Volumes are "up to" figures** for Boeing, Google and Microsoft — options, not firm take-or-pay. Reported headline tonnage systematically overstates contracted certainty. - **Charm's Google and TD deals mix biochar with bio-oil**, so they are not clean read-throughs on pumpable-biomass pricing. - **I could not complete searches** on 2026 Vaulted developments, CDR.fyi's 2025 market-wide average price benchmark, or smaller pumpable-biomass entrants. If you want, I can pick that thread up next and pin the segment against market-wide durable CDR pricing.
Judge’s note — All four gold deals are surfaced with correct values ($53M/112,003 t → $473; $58.3M → $382; $44.2M/122,775 t → $362; Microsoft 4.9 Mt), and the taxonomy-vs-market caveat is made explicitly (TD Bank/Boeing Charm contracts coded biochar despite blended bio-oil supply; Microsoft deal pulled from a deal announcement rather than the pathway tag). Minor errors: the Microsoft–Vaulted date is given as 2025-09-08 rather than July 2025, Vaulted tonnage is 154,146 vs gold's 152,480, and the stated ingest date (2026-09-01) differs from the corpus freshness given in the probe — small but real accuracy slips, and several of the unpriced deals (Boeing, TD Bank, JPMorgan second agreement) are unverifiable from the text alone. Pricing analysis is genuinely expert-grade: it flags that all per-tonne figures are derived (value ÷ tonnes), that Frontier is the sole price-discovery mechanism (7.3% of tonnes priced), that headline prices are lifetime averages over declining schedules (Frontier's 37–75% decline language for Charm), and that Vaulted's tipping-fee/waste-disposal revenue makes its price structurally non-comparable — with a sanity check that 4.9 Mt at Frontier-band prices would imply an implausible ~$2B contract, explicitly labeled as inference. Beyond-corpus material includes the JPMorgan/Charm 61,500 t and 90,000 t cumulative figures plus $20M venture debt, Frontier portfolio price-decline commentary, delivery progress (Charm ~90% vs Vaulted ~34% — deliveries are not in the corpus, and this is the sharpest non-price insight in the answer), and Isometric protocol/credit-issuance detail. Coverage honesty is exemplary: it states the web-search budget was exhausted, names the three specific checks it could not complete, notes CDR.fyi as the fuller scoreboard, and flags missing players (Cowboy Clean Fuels) with no deal record. Structure is clean and decision-useful. Citations are named but unlinked, and a few claims rest on sources a reader would need to chase.
I'll start by mapping what the corpus holds on this pathway.
Rich vein. Now let me pull the supplier-level detail and check for deals that may be filed elsewhere.
Now the required web checks for anything after the corpus freshness dates (deals ingested through 2026-09-01) and for off-corpus pricing detail.
Web-search budget was exhausted mid-check; let me fill remaining gaps from the corpus.
## The short version
Pumpable-biomass CDR ("biomass-injection" in the Carbon Curve taxonomy: bio-oil and organic-waste slurry pumped into wells or caverns) has **11 publicly recorded offtakes, 2021–June 2026, covering ~5.36 Mt of contracted CO₂ across three suppliers**. Only **four disclose a value, and all the usable per-tonne prices come from one buyer — Frontier**. Those prices sit in a tight band of **$362–$473/t**, and they are trending down.
---
## The deal book (Carbon Curve corpus, ingest through 2026-09-01)
**Priced deals — all Frontier, all derived by dividing contract value by tonnes rather than quoted directly:**
| Announced | Buyer | Supplier | Form | Tonnes | Value | Implied $/t | Window |
|---|---|---|---|---|---|---|---|
| 2023-05-18 | Frontier | Charm Industrial | bio-oil (fast pyrolysis → EPA-regulated wells) | 112,003 | $53.0M | **$473** | 2024–2030 |
| 2024-05-01 | Frontier | Vaulted Deep | slurry (organic waste → Class V wells, Kansas & LA) | 154,146 | $58.3M | **$382** | 2024–2027 |
| 2025-12-18 | Frontier | NULIFE GreenTech | bio-oil (hydrothermal liquefaction → salt caverns >1,000 m, Saskatoon) | 122,775 | $44.2M | **$362** | 2026–2030 |
| 2021 | Shopify | Charm Industrial | bio-oil | not stated | $2.0M | — | — |
**Unpriced deals (7 of 11), which carry the overwhelming majority of the tonnage:**
- **Microsoft ↔ Vaulted Deep, 2025-09-08 — 4,900,000 t over 12 years.** Terms undisclosed. This single contract is ~91% of all recorded tonnes in the pathway (source: Arnold & Porter deal announcement).
- **Google ↔ Vaulted Deep, 2025-09-17 — 50,000 t by 2030**, Isometric-verified, a follow-on to Google's 2024 purchase via Frontier.
- **Google ↔ Charm, 2025-04-17** — an "expanded offtake," volume and price undisclosed.
- **Capgemini ↔ Charm, 2025-05-16 — 16,500 t**, price undisclosed.
- **JPMorganChase ↔ Charm, 2026-06-04** — the corpus holds no tonnage. Web check: Charm's own press page puts this second agreement at **61,500 t, taking JPMorgan's cumulative commitment to 90,000 t of bio-oil removals**, alongside a separate **$20M venture debt facility**. Price undisclosed.
- **Stripe/Shopify/JPMorgan ↔ Charm, 2024-08-22 — 105–107 t**: not a commercial-scale offtake but the first Isometric-verified bio-oil deliveries, issued under Isometric's Bio-oil Geological Storage Protocol.
**A classification caveat that matters:** two large recent Charm contracts — **TD Bank (2026-01-20, 44,000 t over ten years from 2029)** and **Boeing (2025-11-14, up to 100,000 t, option structure)** — are coded to *biochar* in the corpus, but the underlying sources describe blended bio-oil-plus-biochar supply (ESG Dive described the TD deal as biochar and bio-oil credits over 10 years). If you are sizing the pumpable-biomass order book commercially rather than taxonomically, some share of those 144,000 t belongs here.
---
## What the pricing actually tells us
**1. There is essentially one price-discovery mechanism, and it is Frontier.** Three of four disclosed values are Frontier offtakes; Microsoft, Google, Capgemini, JPMorgan, Boeing and TD have all disclosed nothing. Priced tonnes are 388,924 of 5,355,636 — **7.3% of the pathway's volume**. That is actually better disclosure than the corpus-wide CDR average (25 of 385 deals with a per-tonne price, covering 1.5% of tonnes), but it means the market's headline number — Microsoft's 4.9 Mt — is a complete price black box.
**2. The band is narrow and falling: $473 → $382 → $362 over 2023–2025, a ~24% decline.** Treat this as three different companies with three different processes rather than a clean time series: Charm uses fast pyrolysis into oilfield wells, Vaulted injects waste slurry into Class V wells, NULIFE uses hydrothermal liquefaction into salt caverns. Still, the volume-weighted average of the three is **~$400/t**, and the newest and cheapest entrant is the one with the most novel process route.
**3. The headline prices are blended averages that conceal steep declining schedules.** Frontier's Charm portfolio page states the per-ton price declines by at least 37% between 2024 and 2030, and potentially as much as 75% given Charm's scaling plans. So $473 is the average across a curve, not a market clearing price: the 2030 marginal ton under that contract is contracted somewhere in the ~$120–300 range. **Anyone quoting "$473/t bio-oil" as today's price is misreading a 2023 contract's lifetime average.**
**4. Delivery performance, not price, is the differentiator so far.** Frontier's portfolio pages (as trued into the corpus on 2026-09-01) show Charm at **~100,600 of 112,003 t delivered (~90%)** against a 2024–2030 window, versus Vaulted Deep at **~52,100 of 154,146 t (~34%)** against a 2024–2027 window that is nearly over. The cheaper contract is the one running behind.
**5. Carbon Curve analysis — the Microsoft/Vaulted price is probably not comparable to the Frontier band.** Vaulted's feedstock is non-hazardous organic waste (biosolids, manure), for which it can charge disposal/tipping fees; Vaulted's own Google announcement frames the value partly as *avoided landfill methane* rather than pure removal. A supplier with a second revenue line and a waste-management cost structure can clear at a lower net CDR price than a pyrolysis operator buying corn stover. A 4.9 Mt, 12-year contract at Frontier-band prices would be a ~$1.8–2.4B transaction; that seems implausible, which is itself evidence the real price is materially lower. This is inference, not a disclosed figure.
---
## Supporting infrastructure worth noting
- **MRV**: Isometric's Bio-oil Geological Storage Protocol is the reference methodology (first verified credits Aug 2024; a further 291 credits issued Nov 2025 via Mangrove Systems' dMRV integration with Isometric's Open API). Puro.earth also carries a Geologically Stored Carbon methodology.
- **Permitting**: US EPA UIC governs the wells — Class VI for CO₂ sequestration, Class V for Vaulted's slurry injection. Six states now hold Class VI primacy (ND, WY, LA, WV, AZ, TX, the last approved Nov 2025).
- **Public money**: DOE FOA 3082 (Carbon Negative Shot Pilots) funded BiCRS pilots including subsurface wood-particle injection, ~$48.6M across nine projects.
## Gaps and uncertainty — read these before quoting the numbers
- **All three per-tonne prices are derived (value ÷ tonnes), not quoted by either party.** They also blend delivery years with declining schedules.
- **The corpus is a curated floor, not a census.** CDR.fyi is the fuller transaction scoreboard; private or unannounced offtakes are invisible here.
- **My web verification is incomplete.** I confirmed the JPMorgan/Charm 61,500 t figure and the Frontier–Charm price-decline language against live sources, but hit the search tool's usage limit before I could check for (a) any pumpable-biomass offtake announced after the 2026-09-01 ingest, (b) whether Microsoft/Vaulted terms have since been disclosed, or (c) NULIFE's post-announcement delivery status. **Treat anything after early September 2026 as unchecked.**
- **Missing players**: Cowboy Clean Fuels (biomass injected into unmineable Powder River Basin coal seams, BiCRS+RNG) appears in the 2026 literature but carries **no deal record** — either it has not announced offtakes or they have not been captured. Subsurface hydrochar storage exists only as a Dutch feasibility study (IJGGC, Dec 2025). The research base for this whole pathway is astonishingly thin: **two in-scope papers, neither a field trial.**
Judge’s note — All four gold deals are surfaced with correct tonnages and values, and the answer goes further by deriving $/t, flagging that all three priced deals come from a single buyer (Frontier), and noting the declining price schedule (37–75% decline embedded in the Charm contract) that makes the $473/t headline a blended average. The pathway-boundary point is partially handled — it correctly separates Charm's biochar and hybrid deals from the pumpable-biomass bucket — but it does not surface the specific gold item that the Microsoft–Vaulted 4.9 Mt deal sits under a different pathway tag; instead it reports the deal as inside the biomass-injection aggregate, which is at odds with the checklist and suggests either a different query result or reconstruction. Date on the Microsoft deal (2025-09-08) also conflicts with the July 2025 announcement, though the answer elsewhere cites Trellis's July 2025 coverage, so the substance is there. Beyond-corpus findings: CDR.fyi/OPIS May 2026 buyer–supplier price gap ($107→$98/t, ~$48/t by 2030 expectation), BiCRS share of 2025 purchase/delivery volume, Q1 2026 record 2.3 Mt quarter, and the Cowboy Clean Fuels–Mercuria RNG offtake correctly characterised as not a CDR deal — four distinct substantive items; delivery/retirement volumes are largely absent beyond two small Charm credit issuances. Citations are named and mostly checkable (Frontier writing page, ESG Dive, Carbon Herald, Trellis, Isometric, CDR.fyi, PRNewswire) though several lack precise anchors. Honesty is excellent: it states that only ~7% of pathway tonnes carry a price, that prices are derived not quoted, that 7 of 11 records are machine-extracted, that tonnage revisions shift $/t by ~±$5, and that web searches hit a limit before covering Q3–Q4 2026. Structure and decision-usefulness are high.
I'll start by checking what the corpus covers, then pull the deal records for this pathway. Now let me pull the aggregate breakdown and one entity view, then search the web for what the corpus flags as outside its window. Now the required web checks for what the corpus flags as outside its window. ## What the Carbon Curve corpus holds I queried the `biomass-injection` pathway (the taxonomy's code for "pumpable biomass to geological storage — bio-oil, slurry"). The aggregate over **every** matching record — not a sample — returns: - **11 deals**, spanning **2021 → 2026-06-04** - **5,355,636 tonnes** contracted - **4 of 11 deals disclose a value**, totalling **$157.5M** and covering **388,924 tonnes** — i.e. only **~7%** of the pathway's contracted tonnes carry any price at all - Three suppliers only: **Charm Industrial** (7 deals, 128,715 t), **Vaulted Deep** (3 deals, 5,104,146 t), **NULIFE GreenTech** (1 deal, 122,775 t) - 7 of the 11 records are `review_status: auto` — machine-extracted from the cited source with no human spot-check ### The deals | Date | Buyer | Supplier | Tonnes | Value | Derived $/t | Form | |---|---|---|---|---|---|---| | 2021 | Shopify | Charm | not stated | $2.0M | — | bio-oil | | 2023-05-18 | Frontier | Charm | 112,003 | $53.0M | **$473** | bio-oil, 2024–2030 | | 2024-05-01 | Frontier | Vaulted Deep | 154,146 | $58.3M | **$378–382** | slurry, Class V wells, 2024–2027 | | 2024-08-22 | Stripe/Shopify/JPMC | Charm | 105.06 | — | — | first Isometric-verified delivery | | 2024-08-22 | JPMorganChase (+Shopify, Stripe) | Charm | 107.21 | — | — | credits issued | | 2025-04-17 | Google | Charm | undisclosed | — | — | "expanded offtake" | | 2025-05-16 | Capgemini | Charm | 16,500 | — | — | bio-oil, Colorado | | 2025-09-08 | Microsoft | Vaulted Deep | 4,900,000 | — | — | 12-year term | | 2025-09-17 | Google | Vaulted Deep | 50,000 | — | — | by 2030, Isometric | | 2025-12-18 | Frontier | NULIFE GreenTech | 122,775 | $44.2M | **$360–362** | HTL bio-oil, salt caverns, Saskatoon | | 2026-06-04 | JPMorganChase | Charm | undisclosed | — | — | plus $20M venture debt (separate) | Note the Charm records also include two **biochar** deals (Google 100,000 t, Jan 2025; Boeing up to 100,000 t, Nov 2025) and a **hybrid** bio-oil/biochar deal with TD Bank (44,000 t, Jan 2026) — those sit outside the biomass-injection bucket, so Charm's pumpable-biomass book is smaller than its total book. ## What the pricing actually tells us **1. Only Frontier prices this pathway in public.** All three per-tonne figures come from Frontier's own writing, and all three are *derived by division* (`usd_amount / tonnes`), not quoted by the parties. Everyone else — Microsoft, Google, Capgemini, Boeing, JPMorganChase — disclosed nothing. So any "price of bio-oil CDR" is really "the price Frontier published on three contracts." **2. The disclosed trend is downward: ~$473 → ~$378 → ~$360.** Across four years of Frontier offtakes the derived price falls roughly 24% from Charm (2023) to NULIFE (2025). That is a real signal, but it compares three different suppliers with different feedstocks, storage media and delivery windows — it is not a like-for-like cost curve for one technology. **3. A caveat on the exact numbers.** The corpus stores $382/t for Vaulted and $362/t for NULIFE, derived from the tonnes stated at announcement (152,480 and 122,000). Tonnes were later trued to Frontier's portfolio page (154,146 and 122,775) on 2026-09-01, which recomputes to **$378** and **$360**. Frontier's own page for the NULIFE deal states $44.2M for 122,000 tons between 2026 and 2030 (frontierclimate.com/writing/nulife); ESG Dive and Carbon Herald reported the same figures in December 2025. Treat these as ±$5/t. **4. The headline price is not the contract price.** Frontier's first-offtake page states the Charm price per ton declines by at least 37%, and possibly as much as 75%, between 2024 and 2030 depending on scaling and government support. So $473 is a blended average over a declining schedule — the 2030 tonne is contracted far cheaper than the 2024 tonne. Any single-number quote for this pathway hides that structure. **5. The volume is where the price isn't.** Vaulted Deep's 4.9Mt Microsoft deal is 91% of the pathway's contracted tonnes and discloses no price. Trellis reported in July 2025 that costs were not disclosed and used Frontier's ~$380/t as the only reference point (trellis.net) — a reasonable inference, but an inference. Isometric confirmed it is the registry for the 12-year, up-to-4.9Mt offtake running through 2038 (isometric.com). At $380/t that deal would be ~$1.9bn; at a megatonne-scale discount, far less. Nobody outside the parties knows. **6. Weighted average, stated honestly.** Over the three deals with both price and tonnes: **$155.5M / 388,924 t = ~$400/t**. (The corpus's $157.5M total includes Shopify's 2021 $2M, which has no tonnes attached, so don't divide that figure.) This describes 7% of the pathway's tonnes and one buyer coalition. ## What I found by searching beyond the corpus - **CDR.fyi/OPIS, May 2026**: across all durable methods the buyer–supplier price gap narrowed from $107/t to $98/t year on year, and respondents expect ~$48/t by 2030 (cdr.fyi). Their June 2026 update notes BiCRS methods dominated execution — in 2025 BiCRS accounted for 96% of purchase volume and 91% of delivered volume. Pumpable biomass is a subset of that BiCRS block. - **Q1 2026 was the largest opening quarter on record** for durable CDR at 2.3 million tonnes contracted, per CDR.fyi — context for why the corpus's 2026 lane is thin rather than empty. - **Cowboy Clean Fuels** injects biomass into unmineable coal seams (the BiCRS+RNG process described in the 2026 *Energy Conversion and Management X* paper the corpus holds). Its announced 10-year offtake with Mercuria (May 2025, PRNewswire) is for renewable natural gas / environmental attributes, **not** CDR credits — so it is arguably a pumpable-biomass injector with no CDR offtake on record. The corpus has no deal for it. - Charm's TD Bank deal was reported in January 2026 as 44,000 t of biochar and bio-oil credits over 10 years, expanding Charm into Canada (ESG Dive) — consistent with the corpus record. ## Gaps and uncertainty - **No price disclosure at all** on 7 of 11 deals, including the two largest. Any pathway-level price is the disclosing minority speaking for the market. - **Derived, not quoted**, on all three priced deals, and sensitive to which tonnage figure you divide by. - **The corpus is a curated floor**, not a census — deal records run 2023→present for machine-extracted announcements, so pre-2023 and unannounced deals are structurally invisible. CDR.fyi is the fuller transaction scoreboard. - **The evidence base is thin**: only 2 in-scope papers carry the `biomass-injection` tag (one lab study on BiCRS+RNG isotopes, April 2026; one hydrochar subsurface-storage feasibility study, December 2025). There are **zero field trials** in the corpus for this pathway — the MRV rests on registry protocols (Isometric's Bio-oil Geological Storage and Biomass Geological Storage protocols) rather than published field science. - **Nothing after 2026-09-01** (deals) or 2026-09-02 (papers) is held here, and my web searches hit their limit before I could check Q3–Q4 2026 announcements. If you need current-quarter deals, CDR.fyi's quarterly update and Frontier's portfolio page are the places to check.
Judge’s note — All four priced/major offtakes are surfaced with tonnages, values and derived $/t, and the analysis of pricing is genuinely expert-grade: it distinguishes term-average vs. tranche pricing (Frontier–Charm's 37–75% in-contract price decline), notes the three prices come from three different processes and one disclosing buyer, flags that 93% of contracted tonnes are unpriced so the clearing price for Microsoft-scale volume is unknown, and observes product heterogeneity (Vaulted's avoided-methane/waste-slurry unit vs. converted bio-oil). It misses the taxonomy caveat in the checklist — it folds the Microsoft 4.9 Mt deal into the 'biomass-injection' pathway count without noting that pathway labels don't cleanly capture this lane, so it doesn't demonstrate that filtering-by-pathway pitfall. Beyond-corpus-type content includes delivery/retirement figures (Charm ~90% of 112k delivered, Vaulted ~34%), the first Isometric-verified bio-oil credit issuances (~105 t), and in-contract price-decline commentary — but the answer simultaneously states it had no web tool, so these and several later items (JPMorganChase–Charm June 2026, Google–Vaulted 50k) are asserted without independent verification, and the Microsoft deal date is given as Sept 2025 rather than July, with a slightly off Vaulted tonnage. Citations are named and mostly checkable (Frontier writing pages, Isometric, Arnold & Porter, Axios, EPA, DOE), though a few URLs and figures could not be confirmed. Honesty about limits is exemplary (explicit floor-not-ceiling caveat, list of searches to run, unresolved 45Q question, thin peer-reviewed evidence base, no compliance demand). Structure and decision-usefulness are excellent.
I'll start by mapping the corpus and pulling the pathway profile. Now the entity-level and aggregate views. Two more corpus pulls to round out the price context and check for suppliers beyond the big three. ## Bottom line The corpus holds **11 offtake deals in the pumpable-biomass lane** (`biomass-injection`: bio-oil and organic-waste slurry injected to geological storage), covering **5,355,636 tonnes contracted** from **three suppliers** — Charm Industrial, Vaulted Deep and NULIFE GreenTech. Only **4 of the 11 disclose a value**, totalling **$157.5M across 388,924 tonnes** — i.e. about **7% of the pathway's contracted tonnes carry a price**, and three of those four prices come from a single buyer (Frontier). Everything below about "pricing" is therefore a statement about Frontier's disclosure practice, not about the market. ⚠️ **Verification caveat up front:** the corpus explicitly instructs that it is "a floor, not a ceiling" and asks for web confirmation beyond its freshness dates (deals ingested to 2026-09-01, latest announcement 2026-08-28). **I have no web-search tool in this session**, so I could not run those checks. Anything announced after ~1 September 2026 is invisible here, and the corpus is a curated set, not a census — CDR.fyi is the fuller transaction scoreboard. Searches to run: *"Charm Industrial offtake 2026"*, *"Vaulted Deep offtake agreement 2026"*, *"NULIFE GreenTech Frontier delivery"*, *"Microsoft Vaulted Deep price per tonne"*. --- ## The deals on record ### Priced (all Frontier; $/t derived from contract value ÷ tonnes, not quoted by the parties) | Announced | Buyer | Supplier | Form | Tonnes | Value | Implied $/t | Window | |---|---|---|---|---|---|---|---| | 2023-05-18 | Frontier | Charm Industrial | bio-oil (fast pyrolysis) | 112,003 | $53.0M | **$473** | 2024–2030 | | 2024-05-01 | Frontier | Vaulted Deep | slurry (organic waste, Class V wells, Kansas & LA) | 154,146 | $58.3M | **$382** | 2024–2027 | | 2025-12-18 | Frontier | NULIFE GreenTech | bio-oil (HTL, salt caverns >1,000 m, Saskatoon) | 122,775 | $44.2M | **$362** | 2026–2030 | | 2021 | Shopify | Charm Industrial | bio-oil | not disclosed | $2.0M | n/a | n/a | Sources: [Frontier — first offtake](https://frontierclimate.com/writing/first-offtake); [Frontier — Vaulted](https://frontierclimate.com/writing/vaulted); [Frontier — NULIFE](https://frontierclimate.com/writing/nulife); [Isometric bio-oil protocol post](https://isometric.com/writing/new-protocol-for-bio-oil-geological-storage). ### Unpriced (the majority of volume) - **Microsoft ↔ Vaulted Deep, 2025-09-08 — 4,900,000 t over 12 years.** No value disclosed. This single deal is **91.5% of all tonnes in the pathway** ([Arnold & Porter](https://www.arnoldporter.com/en/perspectives/news/2025/09/arnold-porter-advises-vaulted-on-landmark-carbon-removal-deal-with-microsoft), which describes it as a "carbon reduction unit" offtake). - **Google ↔ Vaulted Deep, 2025-09-17 — 50,000 t by 2030**, Isometric-verified, follow-on to Google's March 2024 Frontier-facilitated purchase; paired with a Google/Isometric initiative on quantifying avoided methane ([Vaulted Deep](https://vaulteddeep.com/co2-removal-we-can-quantify-methane-emissions-deserve-the-same)). - **Google ↔ Charm, 2025-04-17** — "expanded offtake", no tonnes or price ([Charm Q1-25 wrap-up](https://charmindustrial.com/blog/the-charm-underground-q1-25-wrap-up)). - **Capgemini ↔ Charm, 2025-05-16 — 16,500 t**, Colorado ([Carbon Herald](https://carbonherald.com/capgemini-signs-cdr-deals-with-charm-industrial-and-climeworks)). - **JPMorganChase ↔ Charm, 2026-06-04** — offtake, terms undisclosed, alongside a separate **$20M venture debt** facility ([Axios Pro](https://www.axios.com/pro/climate-deals/2026/06/04/charm-jpmorganchase-venture-debt-carbon-removal)). - **Delivery/issuance records, 2024-08-22** — the first Isometric-verified bio-oil credits: 105.06 t to Stripe/Shopify/JPMorganChase and 107.21 t issued under the Bio-oil Geological Storage Protocol ([Isometric](https://isometric.com/writing-articles/first-verified-credits-for-charm-industrial)). These are deliveries against earlier purchases, not new demand. --- ## What the prices actually tell us **1. The headline trend is down, but it is not a clean learning curve.** $473 (2023) → $382 (2024) → $362 (2025) is a ~24% nominal decline in 2.5 years. But these are three *different companies* using three *different processes* (fast pyrolysis; raw organic-waste slurry with no thermochemical conversion; hydrothermal liquefaction) in two countries. The sequence measures what Frontier was willing to pay for successive new entrants, not one supplier moving down a cost curve. Vaulted's process skips the conversion step entirely, which is the most plausible reason it priced ~19% under Charm. **2. The most informative price datum is inside a contract, not across them.** The Frontier–Charm offtake carries an explicit escalator-in-reverse: the contract price **declines at least 37%, and up to 75%, between 2024 and 2030**. That means the $473 figure is a term average — the 2024 tranches priced well above it and the 2030 tranches materially below, potentially in the ~$150–250 range at the deep end. Frontier's structure is designed to buy the cost curve, not a flat price, and any comparison of $473 to a spot quote is apples-to-oranges. **3. The clearing price for volume is unknown, and is almost certainly far below $362.** *(Carbon Curve–style analysis, not a sourced fact:)* Microsoft's 4.9 Mt from Vaulted Deep would be a ~$1.9bn contract at Frontier's $382/t. Contracts of that size are not signed at catalytic-buyer prices. The corpus's own note — that "the buyers who publish prices are not the buyers who contract the most tonnes" — is nowhere more visible than in this pathway: 7% of tonnes carry a price, and the 93% that don't are the ones that would tell you where the market actually is. **4. The product may not be homogeneous, which limits price comparability.** Charm and NULIFE inject a converted biogenic liquid; Vaulted injects carbon-rich organic waste slurry and its value proposition explicitly includes *avoided methane* from waste that would otherwise be landfilled or land-applied — hence the Google/Isometric methane-quantification workstream and Arnold & Porter's "carbon reduction unit" phrasing. A Vaulted tonne and a Charm tonne priced $91 apart in 2023–24 may not be the same unit of goods. **5. Charm's pivot to biochar is a price signal about bio-oil.** Outside the injection lane, Charm has signed **Google (100,000 t, Jan 2025)**, **Boeing (up to 100,000 t, option, Nov 2025)** and **TD Bank (44,000 t, delivery 2029–2039, Jan 2026)** — all coded biochar/soil storage, not injection. A bio-oil pioneer selling its pyrolysis co-product into a cheaper, 100-year-durability market is consistent with pumpable-biomass economics still sitting at the expensive end. **6. Delivery is uneven, which is a credit-quality caveat on any price.** Frontier's portfolio page (as read 2026-08) shows Charm at **~100,600 of 112,003 t delivered (~90%)**, Vaulted at **~52,100 of 154,146 t (~34%)** against a window closing in 2027, and NULIFE only just starting a 2026–2030 window. --- ## Supporting infrastructure and gaps - **MRV:** Isometric's **Bio-oil Geological Storage Protocol** (ICVCM CCP-approved per the corpus record) underpins Charm and NULIFE; Isometric also verifies Vaulted. Puro.earth's **Geologically Stored Carbon** methodology is an alternative route ([Puro methodologies](https://puro.earth/methodologies)). - **Permitting:** injection sits under the **US EPA UIC** regime — Vaulted uses Class V wells; the Class VI programme (effective 2010-12-10, six primacy states as of the Texas approval in Nov 2025) covers CO₂ sequestration ([EPA](https://www.epa.gov/uic/class-vi-wells-used-geologic-sequestration-carbon-dioxide)). NULIFE uses licensed Saskatchewan salt caverns. - **Public money:** DOE **FOA 3082 Carbon Negative Shot Pilots** funded BiCRS pilots including subsurface wood-particle injection (~$48.6M across 9 projects) ([DOE](https://www.energy.gov/fecm/project-selections-foa-3082-carbon-negative-shot-pilots)). - **Evidence base is thin:** the corpus classifies only **2 in-scope papers** to this pathway (one lab study on BiCRS+RNG in Powder River Basin coals, one Dutch hydrochar-storage feasibility study). There is no peer-reviewed field-trial literature here to cross-check leakage or permanence claims against the $/t being paid. - **No compliance demand:** no compliance-market inclusion instrument covers this pathway in the corpus, so all pricing is voluntary-market pricing. ## What I could not check No 45Q/CDR tax-credit eligibility record for bio-oil injection appears in the corpus — a material pricing variable I cannot resolve here. Nor could I verify post-September-2026 activity, unannounced deals, or the CDR.fyi transaction totals. Treat the three Frontier prices as the only hard numbers, and treat everything derived from them as an inference about a three-supplier, one-disclosing-buyer market.
Asked, verbatim: “How much money has Frontier actually committed to carbon removal, across its AMCs, prepurchases, and offtakes? Break it down.”
It found every checklist fact on this question — not every question favors the database, and this page reports those too.
Judge’s note — The answer hits every checklist item: the 2022 founding tranche (given as $925M over 2022–2030, i.e., the ~$1B launch), the 2023 +$100M expansion with named new members plus the Watershed small-buyer channel, the June 2026 $915M Growth AMC with the $1.8B headline total, small prepurchase rounds with specific figures ($250k Graphyte, $3.05M Pronoe/Cella, $1.75M surficial mineralization trio), and a table of named offtakes (Charm, Vaulted, CO280, NULIFE, Reverion, Eion, CREW, Hafslund, Planetary) with $/t implied pricing. Arbor is absent and the classic annual prepurchase round totals ($2.4M/$6.7M/$7M/$11M) aren't enumerated as rounds, but the prepurchase tier is substantively covered. The strongest feature is the pledge-vs-contracted-vs-delivered decomposition: ~$1.8B pledged, ~$698M contracted across ~53 projects/1.8M tons, ~23,000 tons delivered in 2025 with ~50k forecast for 2026, plus a trajectory ($300M Aug 2024, $713M Dec 2025). It correctly warns that offtakes pay on delivery so cash out the door is far below $698M — exactly the spend-vs-commitment reconciliation the probe asks for. It also self-flags two genuine arithmetic/consistency problems ($925M+$100M+$915M ≠ $1.8B; Dec-2025 $713M > Jun-2026 $698M) rather than papering over them, and lists what Frontier does not disclose (per-buyer allocation, prepurchase/offtake dollar split, cumulative disbursements, terminated contracts). Coverage honesty is exemplary. Weakness: sourcing. Most numbers appear as unattributed quoted fragments; only TechCrunch, CDR.fyi and Frontier itself are named, so a reader cannot easily verify the December-2025 portfolio figures, the Charm $53M/112,003t line, or the Anthropic/buyer-list claims. Some derived $/t figures are presented with more precision than the underlying data supports, and the Meta/JPMorgan attrition inference is speculative. Structure and decision-usefulness are excellent.
I'll research Frontier's commitments across its advance market commitment structure and individual deals. ## The short answer Three very different numbers get called "Frontier's commitment," and conflating them is the main source of confusion: | Tier | Amount | What it actually means | |---|---|---| | **Pledged (AMC capital)** | **$1.8 billion** through 2040 | Buyer promises to spend, not contracts | | **Contracted (signed deals)** | **~$698–700 million** across ~53 projects, ~1.8M tons | Legally binding purchase agreements | | **Delivered** | **~23,000 tons in 2025** (7 companies) | Tons actually in the ground/ocean | So roughly **39% of the pledge has been converted into signed contracts**, and well under 1% of contracted tons have been delivered. --- ## Tier 1: The advance market commitment itself — $1.8B **April 2022 — founding: $925M.** McKinsey Sustainability partnered with Alphabet, Shopify, Meta and Stripe in an eight-year $925 million commitment to accelerate permanent carbon removal by guaranteeing future demand . The original window was 2022 to 2030 . **April 2023 — +$100M.** Autodesk, H&M Group, JPMorgan Chase and Workday joined, committing a combined $100M over eight years and bringing the total AMC past $1B . A separate channel let smaller buyers in: Aledade and Wise joined Boom, Canva, SKIMS and Zendesk in buying Frontier-sourced removal through a partnership with Watershed . **June 17, 2026 — +$915M "Growth AMC."** Participating buyers include Stripe, Google, Shopify, Salesforce, H&M Group, McKinsey Sustainability, Workday, and Anthropic, taking Frontier's total commitment to $1.8 billion . Anthropic is the first pure AI company to join , and companies seeking funding from this tranche must demonstrate a clear path to government subsidies or other public support . **A caveat on the arithmetic:** $925M + $100M + $915M = ~$1.94B, but Frontier headlines $1.8B. The gap is unexplained publicly — it may reflect rounding, restatement of the original tranche, or attrition of founding buyers. Notably, **Meta and JPMorgan Chase do not appear in the 2026 buyer list**, though Meta is still described in trade press as backing the coalition. Frontier has not published a per-buyer allocation table, so who owes what is not verifiable from outside. **Structural point:** Frontier doesn't hold or spend the money. Frontier aggregates buyer demand and negotiates multi-year purchase agreements with suppliers, which are signed by Frontier members and partners — Frontier itself is a public benefit LLC wholly owned by Stripe . A separate channel, Climate Orders, is a Stripe product that retires Frontier-vetted tons on behalf of buyers . --- ## Tier 2: Money actually contracted — ~$700M As of the June 2026 update: Frontier's portfolio includes over 1.8 million tons of contracted removals across 53 projects, with $698 million contracted; $169.8 million and 463,700 tons of that came in the twelve months since June 2025 . TechCrunch put it as "nearly $700 million across more than 50 projects to remove 1.8 million tons." The trajectory: Frontier said in August 2024 it had contracted more than $300M ; by December 2025 it reported over $713 million and 1,886,898 tons across 52 projects, including $261 million and 688,300 tons contracted in the prior twelve months . ⚠️ **Note the inconsistency**: the December 2025 figure ($713M / 1,886,898 tons / 52 projects) is *higher in dollars* than the June 2026 figure ($698M / 1.8M tons / 53 projects) despite an additional project. That implies a contract was cancelled, restated, or that one of the two press reports mis-read the portfolio page. Frontier has not, to my knowledge, explained it. Treat "~$700M contracted" as the defensible number and don't over-precision it. ### Offtakes (the bulk of the money) Offtakes are legally binding contracts to purchase future tons if and when delivered — payment on delivery, not upfront. Named deals with disclosed terms: | Supplier | Value | Tons | Delivery | Implied $/t | |---|---|---|---|---| | Charm Industrial (BiCRS) | $53M / 112,003 t | — | ~$473 | | Vaulted Deep | $58.3M (May 2024) | ~152k | — | ~$380 | | CO280 (pulp & paper) | $48.0M / 224,500 t, 2028–2030 | ~$214 | | NULIFE GreenTech | $44.2M / 122,000 t, 2026–2030 | ~$362 | | Reverion (biogas + SOFC) | $41M / 96,000 t, 2027–2030 | ~$427 | | Eion (ERW) | $33M / 78,707 t, 2027–2030 | ~$419 | | CREW (wastewater) | $32.1M / 71,878 t, 2025–2030 | ~$447 | | Hafslund Celsio (WtE CCS) | $31.6M / 100,000 t, 2029–2030 | $316 | | Planetary (OAE) | $31.3M / 115,211 t, 2026–2030 | ~$272 | Blended across the whole book, ~$698M ÷ ~1.8M tons ≈ **$385/ton** — still far above Frontier's stated <$100/ton destination. ### Prepurchases (small, R&D-stage) Prepurchases are order-of-magnitude smaller — hundreds of thousands to low millions each. Recent examples: Graphyte received a $250,000 prepurchase for 450 tons in November 2025 following technical diligence ; Frontier signed pre-purchases totaling $3.05 million with Pronoe and Cella in January 2026 ; and $1.75 million was facilitated across Karbonetiq, Limenet and pHathom on behalf of Stripe, Shopify and Google (surficial mineralization — a new pathway Frontier pegs at 10Gt+ annual potential at $80–120/ton ). **Gap:** Frontier does not publish a clean prepurchase-vs-offtake dollar split. Based on deal sizes, cumulative prepurchases are plausibly in the low tens of millions — i.e. **~2–3% of contracted dollars but a large share of the ~53 project count**. That's an inference, not a disclosed figure. --- ## Tier 3: What's actually been delivered (and paid) This is where the number collapses. In 2025, seven portfolio companies delivered approximately 23,000 tons — roughly 2x the prior year — and broke ground on 1.4 million tons of new annual removal capacity . Frontier portfolio companies are forecasting more than 50,000 tons removed in 2026, doubling again year over year . Against ~1.8M contracted tons, cumulative deliveries are on the order of **~40,000 tons — roughly 2%**. Since offtakes pay on delivery, **cash actually out the door is a small fraction of $698M**; Frontier does not disclose cumulative disbursements, only contracted value. Anyone citing "$700M spent" is wrong; it's $700M *contracted*. --- ## What the Growth AMC changes The strategy shift matters for forecasting future commitments. Frontier plans roughly 10 to 15 focused bets through 8- to 10-year offtake agreements, with some contracts extending as far as 2040 , concentrating on a narrower portfolio where it sees high conviction of gigaton-scale potential, and requiring a clear line of sight to robust long-term demand in every deal . It will also favor jurisdictions with meaningful carbon removal policy and continue funding new ideas through prepurchases, smaller offtakes and R&D grants . In Frontier's own framing, the initial AMC was meant to "get carbon removal off the starting line" . Context for scale: CDR.fyi data show total durable credit sales approaching $12 billion, versus a few hundred million annually four years ago — so Frontier's ~$700M contracted is roughly 6% of the market's cumulative value, with Microsoft having purchased around 37 million credits and dominating volume. --- ## Uncertainties and gaps to flag 1. **$1.8B is a pledge, not an obligation.** Frontier doesn't escrow funds; individual members sign the contracts. If a member exits, the pledge is not enforceable by any public mechanism I'm aware of. 2. **The Dec-2025 vs Jun-2026 discrepancy** ($713M → $698M) is unresolved. 3. **No per-buyer breakdown** is published — we can't say how much of the $1.8B is Stripe vs. Google vs. Anthropic. 4. **Meta's and JPMorgan's current status** in the coalition is unclear from the 2026 announcement. 5. **Prepurchase/offtake dollar split, cumulative cash paid, and any terminated contracts** are not disclosed line-by-line. 6. **The $1.8B runs to 2040**, versus 2030 for the original tranche — so annual run-rate comparisons across the two AMCs aren't apples-to-apples.
Judge’s note — All five gold items are substantively present and well-organized: the 2022 $1B/nine-year AMC (correctly nuanced as ~$925M in firm commitments at launch), the 2023 +$100M expansion with named new buyers, the June 2026 $915M Growth AMC taking the total to $1.8B, the full ladder of small prepurchase cohorts ($1.75M–$11M), and named facilitated offtakes (Charm, Vaulted Deep, NULIFE, Arbor, Stockholm Exergi, CO280, Heirloom, etc.) explicitly distinguished from the pledge layer. The answer's key contribution is the three-layer framing — pledged vs. contracted vs. delivered — which is exactly the right decision frame and something most answers miss. Beyond-corpus/progress-reporting findings are numerous: the ~$698M / 53 projects / ~1.89M t portfolio reconciliation against Frontier's dashboard; delivery data (~23,000 t in 2025, >50,000 t forecast 2026; Charm ~100,600 t of 112,003 t; Vaulted ~52,100 t of 154,146 t; NULIFE 407 t); contracting-pace deltas from ESG Dive ($261M/688,300 t to Dec 2025 vs $169.8M/463,700 t to June 2026); the AB 1305 disclosure point that unmet offtake conditions mean no payment; the Jan 2026 $2M Leilac/SaltX R&D grants; the Salesforce $25M top-up; and the April 2026 Thetford Mines in-kind support. That is roughly seven distinct substantive items past a simple corpus recitation. Citations are named and mostly checkable (Frontier's own Growth AMC page and portfolio page, ESG Dive, Utility Dive, CFO Dive, Quartz, DataCenterDynamics, Trellis, ESG Today, NS Energy), though the highly precise pathway-level dollar/tonne table is presented without per-deal sourcing and would need spot-checking — the Stockholm Exergi double-count flag is handled honestly rather than papered over. Coverage honesty is excellent: it warns against adding the layers, flags prepurchase-cohort overlap with individually named deals, notes undisclosed grant amounts, notes the $915M was unallocated at announcement, distinguishes Frontier Climate from Frontier Infrastructure/Frontier Carbon Solutions, and states its freshness cutoff. Deductions: one garbled sentence ('~$366/t in the earlier window vs ~$366/t… roughly flat') that undercuts its own price-trend claim; the Meta-absence and Anthropic-inclusion observations are asserted from a single source; and the CarbonCapture Inc. $20M subtraction to force reconciliation is a bit hand-wavy ('~$1M of overlap'). Still, this is close to expert-grade work.
I'll start by checking what the corpus knows about Frontier, then verify against current public information. Let me pull the remaining Frontier deals and cross-check against current public sources. ## Bottom line: three different numbers, and they are not interchangeable Frontier's "how much money" question has three defensible answers depending on what you mean: | Measure | Amount | As of | |---|---|---| | **Pledged** (AMC capital committed by member buyers) | **$1.8 billion** | June 2026 | | **Contracted to named suppliers** (signed offtakes + prepurchases) | **~$698M** across 53 projects / ~1.89M tonnes | June 2026 | | **Delivered and therefore actually paid** | low tens of thousands of tonnes (~23,000 t in 2025; >50,000 t forecast 2026) — a few percent of contracted value | 2026 | The gap between the second and third rows is the whole story: Frontier's contracts are milestone-and-delivery contingent. Frontier's own AB 1305 disclosure page states that where offtake conditions are not met, no money is required to be paid by Frontier buyers, and that verification and registry issuance precede delivery. --- ## 1. The pledge stack (the "AMC" layer) | Date | Instrument | Amount | Buyers named | |---|---|---|---| | Apr 2022 | Original AMC launch, nine years through 2030 | **$1B** target (reported at launch as ~$925M in firm commitments) | Stripe, Alphabet, Shopify, Meta, McKinsey Sustainability + Stripe Climate SMBs | | Apr 2023 | AMC expansion | **+$100M** | Autodesk, H&M Group, JPMorgan Chase, Workday | | Jun 2024 | Member top-up | **+$25M** | Salesforce (part of its wider $100M-by-2030 pledge) | | Jun 17 2026 | **Growth AMC** | **+$915M** | Stripe, Google, Shopify, Salesforce, H&M Group, McKinsey Sustainability, Workday, **Anthropic** (new) | Frontier's own announcement (frontierclimate.com/writing/growth-amc, 17 June 2026) states the new money takes total commitment to $1.8 billion; DataCenterDynamics, Trellis and ESG Today all reported the same figure the same week. The Growth AMC is structurally different from the first: roughly 10–15 concentrated bets via 8–10 year offtakes aimed at getting projects to FID at commercial scale, with contracts running out as far as 2040 — versus the 2022 AMC's spray of small prepurchases. Two things worth flagging: the $100M and $25M top-ups are *inside* the ~$1B first tranche envelope, not additive to $1.8B (Frontier's arithmetic is $925M/$1B + $915M). And **Meta is not named among the Growth AMC's participating buyers** on Frontier's own page, despite being a 2022 founding member — I would not read intent into that without asking them, but it is a genuine change in the roster. ## 2. What is actually under contract I reconstructed the contracted total from every priced Frontier deal on record. The 19 growth-stage offtakes plus a handful of priced prepurchases come to **$690M across 1,866,394 tonnes**, a volume-weighted **~$370/tonne**: | Pathway | $ contracted | Tonnes | Implied $/t | Deals | |---|---|---|---|---| | BECCS / waste-to-energy | $210.2M | 721,444 | $291 | Stockholm Exergi, CO280, Hafslund Celsio, Arbor, Reverion | | BiCRS (bio-oil, biomass burial) | $155.8M | 389,374 | $400 | Charm, Vaulted Deep, NULIFE, Graphyte | | Enhanced rock weathering | $117.6M | 327,621 | $359 | Lithos, Terradot, Eion, Alt Carbon | | Direct air capture | $117.2M | 181,236 | $647 | Heirloom, CarbonCapture Inc.*, 280 Earth, Phlair | | Ocean / river alkalinity | $89.3M | 246,719 | $362 | CarbonRun, CREW, Planetary, pHathom | | **Total** | **$690.1M** | **1,866,394** | **$370** | | Plus the small-cheque prepurchase cohorts, which are separate from the offtakes: $2.4M (Jun 2022, 6 companies), $11.0M (Dec 2022, 7), $7.0M (Sep 2023, 12), $4.5M (Sep 2024, 9), $1.75M (Jul 2025, Karbonetiq/Limenet/pHathom) and $3.05M (Jan 2026, Pronoe + Cella) — **~$29.7M total, ~2,000–7,000 tonnes per round**. That reconciles almost exactly to Frontier's own dashboard: $690.1M − $20M (CarbonCapture Inc., which no longer appears in the portfolio after Project Bison was paused) + $29.7M of prepurchase cohorts − ~$1M of overlap = **$698.8M**, against the **$698M across 53 projects / >1.8M tonnes** reported by ESG Dive, Utility Dive and CFO Dive from Frontier's portfolio page in June 2026. Quartz's write-up put it as nearly $700M across more than 50 projects for 1.8 million tonnes. **Pace of contracting:** ESG Dive reported $261M and 688,300 tonnes contracted in the twelve months to December 2025, and $169.8M for 463,700 tonnes in the twelve months to June 2026 — i.e. contracting *slowed* in H1 2026 while the new AMC was being assembled, and average price fell (~$366/t in the earlier window vs ~$366/t… roughly flat, with the mix shifting toward cheap BECCS). ## 3. The grant layer (small, and easy to miss) Separate from purchases: **$2M in R&D grants to Leilac (UK) and SaltX (Sweden)** for zero-carbon lime, announced 22 January 2026, on behalf of Stripe, Shopify and Google. Earlier undisclosed-value R&D grants went to Kodama, Nitricity, Living Carbon, the Yale Carbon Containment Lab (2022), Carboniferous and Rewind Earth (2023). Frontier's 2026 Innovation program replaces the old prepurchase track with explicit R&D funding at flexible cheque sizes — **amount not disclosed**. Frontier has also put in-kind support behind a surficial mineralization hub at Thetford Mines, Quebec (April 2026), offering 10,000 tonnes of serpentinite tailings and test space rather than cash. ## 4. Delivery — the number that matters least to headlines and most to the market Per Frontier's Growth AMC page: seven portfolio companies delivered roughly **23,000 tonnes in 2025**, with **>50,000 tonnes forecast for 2026**, and portfolio companies broke ground on 1.4M tonnes of new annual capacity. Cumulatively, Charm has delivered ~100,600 t of its 112,003 t contract and Vaulted Deep ~52,100 t of 154,146 t, per the portfolio page as of August 2026; NULIFE had delivered 407 t against its 2024 prepurchase. So of ~1.89M contracted tonnes, roughly **8–9% has been delivered**, concentrated in two BiCRS suppliers. ## Caveats, double-count traps and gaps - **Do not add the layers.** $1.8B pledged *contains* the ~$698M contracted; the prepurchase cohort rounds partly *contain* the individually-named small deals (pHathom's $500k sits inside the $1.75M July 2025 round; Alt Carbon's $500k inside the $4.5M Sept 2024 round). - **Stockholm Exergi appears twice.** A $48.6M / ~180,000 t offtake (June 2024) and a "~$50M" Frontier commitment reported by NS Energy at the March 2026 KVV8 FID greenlight. I treat the second as a re-report of the same commitment — including both would inflate the total to ~$749M and break the reconciliation with Frontier's own $698M. Unconfirmed either way; worth a direct check with Frontier. - **Frontier is a facilitator, not the payer.** Contracts are between individual member buyers and suppliers; Frontier Climate Management LLC is a public benefit LLC wholly owned by Stripe. "Frontier committed $X" always means "Frontier's buyers committed $X." - **Prices are unusually transparent here and that skews market statistics.** Frontier discloses a dollar figure on essentially every growth-stage offtake, while the largest-volume buyers in the market disclose none — so Frontier's ~$370/t weighted average describes Frontier, not durable CDR pricing. Range within the portfolio: $214/t (CO280 BECCS) to $989/t (Heirloom DAC). - **Name collision.** "Frontier Infrastructure Holdings" (the ethanol-BECCS developer that signed a 750,000-credit deal with Carbonfuture in July 2026) and "Frontier Carbon Solutions" (Wyoming Class VI permits) are unrelated companies. Do not roll their volumes into Frontier Climate's. - **Freshness.** My structured records run to 1 September 2026, and my web checks surfaced nothing newer than the Growth AMC and the June 2026 portfolio snapshot. Any Growth AMC offtakes signed after mid-2026 are not captured here — the $915M was still unallocated as of the announcement, and the first Growth AMC deals are the number to watch. Frontier's live dashboard at frontierclimate.com/portfolio is the authoritative current figure.
Judge’s note — All five checklist items are substantively covered: the $1B 2022 launch, the ~$100M 2023 member expansion, the $915M Growth AMC (June 2026), the full ladder of small prepurchase cohorts ($1.75M–$11M), and a named offtake book (Charm, Vaulted, NULIFE, Arbor, Stockholm Exergi, etc.) explicitly separated from the pledge pool, with the member-purchase/facilitation structure noted and a Frontier-as-supplier facilitation row flagged. The answer's strongest feature is the explicit three-layer framing (pledged $1.8bn / contracted ~$690–740M / delivered tens of thousands of tonnes), which is exactly the distinction the question demands. Beyond-corpus work is substantial: reconciliation against Frontier's own progress reporting ($698M contracted, 53 projects, 1.8M tonnes), independent reporting at the Growth AMC launch, the 2026 delivery forecast (>50k t) and 1.4M t of capacity breaking ground, the Stockholm Exergi $48.6M restatement check, the Anthropic/Salesforce additions, and the 2026 Innovation Program successor to prepurchases. Citation quality is good but uneven — some claims are anchored to named URLs/outlets while others rest on vague attributions ('a market tracker', 'independent reporting'); derived $/t figures are correctly flagged as division artefacts. Coverage honesty is exemplary: it names the $2.04bn summing artefact, the $925M duplicate record, the likely Stockholm Exergi double-count, the cohort-vs-individual-row overlap risk, unpriced grants, a possibly dead CarbonCapture contract, mixed review provenance, and the unaddressed contracted-vs-paid gap. Clarity is high, though the meta-commentary about SQL totals and review_status adds some noise for a practitioner reader. Deduction only for attribution looseness on a few quoted figures.
I'll start by checking what the Carbon Curve corpus covers, then pull Frontier's record. Now let me enumerate all 51 Frontier deals and aggregate the funding layer. Now the required web checks for what the corpus can't see — anything after its 2026-09-01 ingest, and Frontier's own stated totals. ## Short answer Three different numbers get called "Frontier's commitment," and they are not the same thing: | Layer | Amount | What it means | |---|---|---| | **Pledged capital (the AMC)** | **$1.8 billion** | Money member companies have promised to spend on CDR, most of it not yet contracted | | **Contracted (signed purchase agreements)** | **~$690–740 million** | Actual binding offtakes + prepurchases signed with named suppliers | | **Delivered** | **tens of thousands of tons** | What has physically been removed and verified | Below is the breakdown, from the Carbon Curve corpus (deals ingested through 2026-09-01) plus web verification of everything after that. --- ## 1. The AMC pledges — $1.8bn, and how it was assembled From the corpus funding layer (all four records `review_status: approved` at review, auto-published): | Date | Instrument | Amount | Source | |---|---|---|---| | 2022 (launch) | Original AMC, nine years through 2030 | $1,000M | frontierclimate.com/writing/launch | | 2023-04-12 | AMC expansion — new members Autodesk, H&M Group, JPMorgan Chase, Workday | $100M | frontierclimate.com/writing/new-members | | 2024-06-18 | Salesforce → Frontier (inbound member commitment) | $25M | GreenBiz | | 2026-06-17 | **Growth AMC** | **$915M** | frontierclimate.com/writing/growth-amc | Web confirmation of the newest tranche: Frontier announced $915 million in new funding commitments, adding Anthropic and Salesforce as participants and bringing its total pledges to $1.8 billion . The additional $915m raised from Stripe, Google, Shopify, Salesforce, H&M Group and Anthropic effectively doubles the coalition's total commitment . The Growth AMC record specifies the structure: approximately 10–15 focused bets through 8- to 10-year offtakes to get projects to final investment decision at commercial scale, contracting out as far as 2040. **A trap worth naming.** If you ask the corpus to sum every funding record where Frontier is the funder, you get **$2,043,650,000 across 9 valued records** (16 records total, 7 with no amount). That number is arithmetically correct and substantively wrong — it double-counts AMC vintages against each other. There is also a separate 2023 record of **$925M** (Rolling Stone-sourced, `review_status: auto`, funder field unnamed) which is the *same* original commitment described at founding-member level, not additional money. **$1.8bn is the right pledge figure; $2.04bn is a summing artefact.** ## 2. Offtakes — where nearly all the actual money is The corpus holds **50 deals with Frontier as buyer, 1,896,950 tonnes contracted, 22 with a disclosed value, $740,050,000 disclosed total, covering 1,866,394 tonnes**. Those are exact SQL totals over every matching record, not a sample of rows. Unusually for this database, Frontier is nearly fully price-transparent: 98.4% of its contracted tonnage carries a disclosed value, against **1.5% of tonnes market-wide** (44 of 385 deals corpus-wide). So a Frontier-specific per-tonne average is meaningful in a way market averages are not. **The 22 disclosed-value agreements:** | Date | Supplier | Tonnes | Value | $/t (derived) | Pathway | |---|---|---|---|---|---| | 2023-05-18 | Charm Industrial | 112,003 | $53.0M | 473 | biomass-injection | | 2023-11-16 | CarbonCapture Inc. | 45,500 | $20.0M | 440 | DAC | | 2023-11-16 | Heirloom | 26,889 | $26.6M | 989 | DAC | | 2023-12-07 | Lithos Carbon | 154,880 | $57.1M | 370 | ERW | | 2024-05-01 | Vaulted Deep | 154,146 | $58.3M | 382 | biomass-injection | | 2024-06-18 | Stockholm Exergi | 179,998 | $48.6M | — | BECCS | | 2024-07-11 | 280 Earth | 61,571 | $40.0M | 650 | DAC | | 2024-09-18 | Alt Carbon | 1,851 | $0.5M | — | ERW | | 2024-09-23 | CarbonRun | 56,753 | $25.4M | 458 | river alkalinity | | 2024-12 | Terradot | 89,998 | $27.0M | 300 | ERW | | 2024-12-17 | CO280 | 224,446 | $48.0M | 214 | BECCS | | 2024-12-17 | CREW Carbon | 72,492 | $32.1M | 447 | wastewater alkalinity | | 2025-02-27 | Phlair | 47,276 | $30.6M | 651 | DAC | | 2025-03-25 | Eion | 80,892 | $33.0M | 419 | ERW | | 2025-04-01 | Hafslund Celsio | 99,998 | $31.6M | 316 | BECCS | | 2025-07-08 | Arbor | 121,002 | $41.0M | 353 | BECCS | | 2025-08-26 | Planetary Technologies | 116,964 | $31.3M | 272 | OAE | | 2025-11-20 | Graphyte | 450 | $0.25M | 556 | biomass-storage | | 2025-11-25 | Reverion | 96,000 | $41.0M | 427 | BECCS | | 2025-12-18 | NULIFE GreenTech | 122,775 | $44.2M | 362 | biomass-injection | | 2026-02-17 | pHathom | 510 | $0.5M | — | marine | | 2026-03-05 | Stockholm Exergi *(see below)* | — | $50.0M | — | BECCS | Per-tonne prices are **derived by division, not quoted by the source**, except where the source states one (Terradot's $300/t was stated). **One likely double-count.** The 2026-03-05 Stockholm Exergi record ($50M, extracted from an NS Energy article on the plant's FID, `review_status: approved`) appears to be a restatement of the June 2024 commitment, not a second purchase. Frontier's own announcement is unambiguous: Frontier members will buy $48.6M of the facility's removal output between 2028 and 2030 , and as recently as mid-2026 a market tracker still described the Frontier coalition as having committed $48.6 million in offtake agreements to that project. **Strip it out and the disclosed offtake book is $690.05M — a weighted average of roughly $370/tonne.** That reconciles closely with Frontier's own published figure. As of June 2026, Frontier's portfolio includes over 1.8 million tons of contracted carbon removals across 53 projects, with $698 million contracted . Independent reporting at the Growth AMC launch put it at contracts worth almost $700 million for 1.8 million tons of removals from 53 projects . **So: ~$698M contracted against $1.8bn pledged — roughly 39% of the money is committed to named suppliers.** ## 3. Prepurchases — small by design Frontier's early-stage track is filed in the corpus as cohort *funding rounds* rather than deals (an owner decision of 2026-08-15, since a round is one announcement covering many suppliers): | Round | Amount | Recipients | |---|---|---| | 2022-06 (Spring) | $2.4M | AspiraDAC, Calcite-Origen, Lithos, RepAir, Travertine, Living Carbon | | 2022-12 (Fall) | $11.0M | Arbor, Captura, Arca, Carbon To Stone, Cella, CREW, InPlanet | | 2023-09 (Fall) | $7.0M | Airhive, Alkali Earth, Banyu, Phlair, CarbonBlue, CarbonRun, EDAC, Holocene, Mati, Planetary, Spiritus, Vaulted Deep | | 2024-09 (Fall) | $4.5M | Alithic, Alt Carbon, Anvil, Capture6, Exterra, Flux, NULIFE, Planeteers, Silica | | 2025-07 (Summer) | $1.75M | Karbonetiq, Limenet, pHathom | | **Total** | **~$26.65M** | ~40 suppliers | That is **under 4% of contracted dollars for roughly 80% of the supplier relationships** — the whole point of the structure. Note the corpus also carries individual prepurchase *deal* rows backfilled from Frontier's portfolio page (Arca 380t, Captura 508t, InPlanet 1,041t, Spiritus 713t, Airhive 943t, Capture6 1,000t, Karbonetiq 2,142t and others). Those overlap with the cohort rounds above — **do not add the two together.** ## 4. R&D grants — mostly unpriced - **$2.0M** zero-carbon lime grants to Leilac and SaltX, 2026-01-22 (`auto`) - Six grants with **no disclosed amount**: Kodama, Nitricity, Living Carbon, Yale Carbon Containment Lab (2022); Carboniferous, Rewind Earth (2023) - The **2026 Innovation Program** — the successor to the prepurchase track, amount undisclosed. Its priority areas include improving mineralization, advancing ocean and inland water alkalinity enhancement , plus measurement tools for open systems. ## 5. What this buys, physically Delivery is the number that hasn't scaled yet. Frontier's own June 2026 statement: seven portfolio companies broke ground on 1.4 million tons of new annual removal capacity in 2025 and are forecast to deliver more than 50,000 tons of carbon removal in 2026, more than doubling for the second year in a row . Frontier itself frames the position as four years in, hundreds of companies being built and tens of thousands of tons delivered . Against 1.9M tonnes contracted, cumulative delivery is on the order of low hundreds of thousands at best — Charm alone accounts for ~100.6k of it per the portfolio page, and Vaulted ~52.1k. --- ## Caveats, disagreements and gaps - **Corpus vs. Frontier disagree by ~$42–50M** on contracted value ($740.05M vs $698M). My read is the Stockholm Exergi restatement plus small rounding; I'd take Frontier's $698M as the better figure and treat the corpus total as an upper bound. - **Deal count differs**: corpus 50, Frontier 53. The corpus's portfolio backfill (2026-09-01) may not have caught every small prepurchase row. - **Provenance is mixed.** The large offtakes sourced from frontierclimate.com/writing are `review_status: approved`; several smaller rows (Alt Carbon, Graphyte, pHathom, Terradot, Mati, Banyu) are `auto` — machine-extracted from their cited source with no human spot-check. - **One record is not a Frontier commitment at all**: a 2024-06-28 deal listing Mercedes-AMG Petronas F1 as buyer and *Frontier* as supplier is a facilitation, correctly excluded from the buyer bucket. - **CarbonCapture Inc.'s $20M/45,500t** no longer appears on Frontier's portfolio page as of 2026-08 (Project Bison paused in late 2024). Delivery status unconfirmed — so headline "contracted" figures may include a dead contract. - **Not covered anywhere here**: how much of the $1.8bn has actually been *paid* rather than contracted (take-and-pay means cash moves on delivery, so paid-out is far below $698M); the split of the pledge by member company; and any deal announced after the corpus's 2026-09-01 ingest — my web searches surfaced nothing new between then and now, but absence of a search hit is not proof of absence.
Judge’s note — All five checklist items are substantively present: the 2022 launch ($925M framed as $1B over nine years to 2030), the 2023 ~$100M member expansion, the June 2026 $915M Growth AMC, an itemized table of prepurchase cohorts spanning $1.75M–$11M, and a 19-line offtake table naming Charm, Vaulted, NULIFE, Arbor and others with tonnes, value and derived $/t. The answer's central contribution is exactly the reconciliation the question invites — separating pledged envelopes (~$1.8B, with an explicit warning that the $1B launch, $100M expansion and $915M Growth AMC must not be stacked) from contracted spend (~$740M disclosed) from delivered tonnage (~23kt in 2025, >50kt forecast 2026, supplier-level delivery for Charm and Vaulted) — plus the structural point that Frontier itself is not the buyer of record. Beyond-corpus/reconciliation findings counted: 2025 delivery vs. contract totals, supplier-level delivered-vs-contracted, 1.4Mt/yr capacity broken ground, and the Jan 2026 $2M lime grant / Apr 2026 Quebec hub as post-window items. Coverage honesty is exemplary: it flags the possible Stockholm Exergi double count, the likely-dead CarbonCapture contract, the prepurchase/deal-record overlap, that only 22 of 50 deals disclose value (so $770M is a floor), that per-tonne prices are derived, and that it had no live web access past early Sept 2026. Citations are mostly named and checkable (Frontier writing posts, GreenBiz, NS Energy) though several tabulated dollar figures rest on an unnamed internal corpus and would need verification; a few details (e.g., specific Growth AMC membership, some 2025–26 offtake values) are asserted without direct links. Structure and decision-usefulness are high — the 'don't add the layers' instruction is precisely what an expert reader needs.
I'll start by checking what the corpus covers, then pull Frontier's full record. Let me pull the remaining Frontier deals and the aggregate view of its funding commitments. ## Bottom line There are two very different numbers, and they get conflated constantly: | Layer | Amount | What it is | |---|---|---| | **Pledged capital (AMC envelopes)** | **~$1.8 billion** | Money member companies have *committed to spend* on removal by 2030/2040 — not yet contracted | | **Actually contracted to named suppliers** | **~$740M disclosed in deals + ~$29M in cohort prepurchases/grants ≈ $770M** | Signed offtakes, prepurchases and R&D grants where a counterparty exists | | **Actually delivered (tonnes)** | **~23,000 t in 2025 across seven suppliers**; >50,000 t forecast for 2026 | Credits issued and retired | So roughly **40–45% of the pledged $1.8B has been converted into signed contracts**, and delivered tonnage is a rounding error against the ~1.9 Mt contracted. That gap is the whole story of Frontier. --- ## 1. The pledges (funding layer) | Date | Instrument | Amount | Source | |---|---|---|---| | Apr 2022 | Frontier AMC launch — Stripe, Alphabet, Meta, Shopify, McKinsey Sustainability | $925M (framed as "$1B over nine years, through 2030") | [frontierclimate.com/writing/launch](https://frontierclimate.com/writing/launch); [Rolling Stone, 2023-04-03](https://www.rollingstone.com/politics/politics-features/frontier-google-facebook-carbon-emissions-1234696518) | | Apr 2023 | New members: Autodesk, H&M Group, JPMorgan Chase, Workday | ~$100M (takes the AMC past $1B) | [frontierclimate.com/writing/new-members](https://frontierclimate.com/writing/new-members) | | Jun 2024 | Salesforce joins | $25M (part of its broader $100M-by-2030 pledge) | [GreenBiz, 2024-06-18](https://www.greenbiz.com/article/salesforce-commits-25-million-steep-learning-curve-carbon-removal) | | Jun 2026 | **Growth AMC** — Stripe, Google, Shopify, Salesforce, H&M, McKinsey, Workday, **Anthropic** | **$915M**, "10–15 focused bets" via 8–10-year offtakes contracting as far as **2040** | [frontierclimate.com/writing/growth-amc](https://frontierclimate.com/writing/growth-amc) | Frontier's own framing on the Growth AMC page: the new $915M **"brings Frontier's total commitment to $1.8 billion."** The Carbon Curve funding layer totals $2.04B across 16 Frontier-funder records, but that figure double-counts — it adds the $1B launch, the $100M expansion and the $915M Growth AMC on top of each other, when the first two are the same envelope. **Important structural point:** Frontier itself buys nothing. It is a wholly-owned Stripe subsidiary that facilitates purchases; the buyer of record on each contract is a member company (Stripe, Google, Shopify, Meta, H&M, Match via Watershed, etc.). "Frontier committed $X" always means "Frontier's members committed $X." --- ## 2. The offtakes — the real money (19 deals, ~$739M disclosed) These are multi-year, fixed-price take-and-pay contracts with growth-stage suppliers, using Frontier's [published offtake template](https://frontierclimate.com/writing/offtake-agreement-template). | Announced | Supplier | Pathway | Tonnes | Value | $/t (derived) | |---|---|---|---|---|---| | 2023-05-18 | Charm Industrial | bio-oil injection | 112,003 | $53.0M | $473 | | 2023-11-16 | CarbonCapture Inc. | DAC | 45,500 | $20.0M | $440 | | 2023-11-16 | Heirloom | DAC | 26,889 | $26.6M | $989 | | 2023-12-07 | Lithos Carbon | ERW | 154,880 | $57.1M | $370 | | 2024-05-01 | Vaulted Deep | slurry injection | 154,146 | $58.3M | $382 | | 2024-06-18 | Stockholm Exergi | BECCS | 179,998 | $48.6M | n/a | | 2024-07-11 | 280 Earth | DAC | 61,571 | $40.0M | $650 | | 2024-09-23 | CarbonRun | river alkalinity | 56,753 | $25.4M | $458 | | 2024-12 | Terradot | ERW (Brazil) | 89,998 | $27.0M | $300 | | 2024-12-17 | CO280 | BECCS (pulp mill) | 224,446 | $48.0M | $214 | | 2024-12-17 | CREW Carbon | wastewater alkalinity | 72,492 | $32.1M | $447 | | 2025-02-27 | Phlair | electrochemical DAC | 47,276 | $30.6M | $651 | | 2025-03-25 | Eion | ERW (olivine) | 80,892 | $33.0M | $419 | | 2025-04-01 | Hafslund Celsio | waste-to-energy CCS | 99,998 | $31.6M | $316 | | 2025-07-08 | Arbor | BECCS | 121,002 | $41.0M | $353 | | 2025-08-26 | Planetary Technologies | OAE | 116,964 | $31.3M | $272 | | 2025-11-25 | Reverion | biogas SOFC BECCS | 96,000 | $41.0M | $427 | | 2025-12-18 | NULIFE GreenTech | HTL bio-oil | 122,775 | $44.2M | $362 | | 2026-03-05 | Stockholm Exergi (2nd) | BECCS | n/d | ~$50.0M | n/a | **Subtotal: ~$738.8M across ~1.86 Mt.** Volume-weighted average where both are disclosed: **~$370/t** (excluding the untonned 2026 Exergi commitment). ⚠️ **Double-count risk to flag:** the March 2026 Stockholm Exergi "~$50M" record was machine-extracted from a secondary source ([NS Energy](https://www.nsenergybusiness.com/analysis/stockholm-exergi-greenlights-world-leading-beccs-plant)) at the plant's FID. It may be a re-report of the June 2024 $48.6M commitment rather than a second contract. If so, contracted spend drops to **~$690M**. ⚠️ **At least one contract is probably dead:** CarbonCapture Inc. ($20M / 45,500 t) no longer appears on Frontier's portfolio page as of Aug 2026, after Project Bison was shelved in late 2024. Delivery status unconfirmed. --- ## 3. The prepurchases — small, many, cheap Cohort rounds paying early-stage suppliers upfront for a few hundred to a few thousand tonnes: | Round | Amount | Recipients | |---|---|---| | Spring 2022 | $2.4M | AspiraDAC, Calcite-Origen, Lithos, RepAir, Travertine, Living Carbon | | Fall 2022 | $11.0M | Arbor, Captura, Arca, Carbon To Stone, Cella, CREW, InPlanet (7,011 t) | | Fall 2023 | $7.0M | Airhive, Alkali Earth, Banyu, Phlair, CarbonBlue, CarbonRun, EDAC Labs, Holocene, Mati, Planetary, Spiritus, Vaulted Deep | | Fall 2024 | $4.5M | Alithic, Alt Carbon, Anvil, Capture6, Exterra, Flux, NULIFE, Planeteers, Silica | | Summer 2025 | $1.75M | Karbonetiq, Limenet, pHathom (2,982 t) | | Jan 2026 | ~$3.05M | Pronoe (OAE, 333 t) + Cella (basalt mineralization, Kenya, 3,212 t) | | Nov 2025 | $0.25M | Graphyte (450 t) | | **Total** | **~$30M** | ~35 suppliers | Careful: the individual Alt Carbon ($0.5M) and pHathom ($0.5M) deal records are *inside* the 2024 and 2025 round totals. Net of that overlap, prepurchases add roughly **$28M** on top of the offtakes. ## 4. R&D grants Mostly undisclosed amounts: Kodama, Nitricity, Living Carbon, Yale Carbon Containment Lab (2022); Carboniferous, Rewind Earth (2023); **$2M for zero-carbon lime** to Leilac and SaltX (Jan 2026, on behalf of Stripe/Shopify/Google). The 2026 Innovation Program replaced the prepurchase track with flexible-cheque R&D funding — no envelope disclosed. Frontier also stood up a non-cash **Quebec surficial mineralization hub** at Thetford Mines (Apr 2026): 10,000 t of serpentinite tailings plus pilot space. --- ## 5. Where the money went, by pathway Of the $740.05M in disclosed deal value (Carbon Curve computes this over all 50 Frontier-as-buyer deals; 22 disclose a value, covering 1,866,394 t): | Pathway | Disclosed $ | Share | |---|---|---| | BECCS / waste-to-energy CCS | $260.2M | 35% | | Biomass injection (bio-oil, slurry) | $155.5M | 21% | | Enhanced rock weathering | $117.6M | 16% | | DAC | $117.2M | 16% | | Ocean/river alkalinity | $88.8M | 12% | | Other (biomass storage, pHathom) | $0.75M | <1% | DAC is **16%** of Frontier's disclosed spend — a useful corrective to the assumption that AMC money is DAC money. BECCS and BiCRS together take over half, because they clear on price ($214–473/t vs $440–989/t for DAC). --- ## 6. What has actually been delivered This is the number that matters and the one that is smallest: - **~23,000 tonnes delivered in 2025** across seven portfolio companies; Frontier forecasts **>50,000 t for 2026**, and says portfolio companies broke ground on **1.4M t/yr of new capacity** ([growth-amc post](https://frontierclimate.com/writing/growth-amc)). - Charm: ~100,600 t delivered of 112,003 contracted (portfolio page, Aug 2026). - Vaulted Deep: ~52,100 t of 154,146 contracted. - Against **1,896,950 t contracted**, cumulative delivery is on the order of **10%**, and most contracted tonnage sits in 2027–2030 delivery windows (Arbor 2028–30, CO280 2028–30, Reverion 2027–30, Hafslund 2029–30). --- ## 7. Caveats and where I'd verify - **Coverage boundary.** All figures above come from the Carbon Curve corpus, which holds deals announced through **2026-08-28** and was last ingested **2026-09-01**. I had no live web access in this session, so **anything Frontier announced after early September 2026 is not reflected here** — and Frontier typically signs several offtakes a year. Check `frontierclimate.com/portfolio` (the source Carbon Curve trued its tonnage figures against on 2026-09-01) and CDR.fyi's buyer scoreboard for anything since. - **Values are disclosed on 22 of 50 deals.** Every backfilled portfolio prepurchase (Arca, Captura, Silica, Flux, Spiritus, Holocene, etc.) carries tonnes but no price, so the ~$770M is a **floor**, not a total. The true contracted figure is somewhat higher — though probably only by low tens of millions, since the undisclosed deals are all sub-3,000-tonne prepurchases. - **Don't add the layers.** $1.8B pledged, ~$770M contracted, and the ~$30M of prepurchase rounds are nested, not additive. The correct sentence is: "of ~$1.8B pledged, ~$740M has been contracted to named suppliers in publicly-valued deals." - **Per-tonne prices are derived**, not quoted — dollar amount divided by contracted tonnes. Frontier's Charm contract, for one, has a price that declines 37–75% over 2024–2030, so a single average conceals a schedule. - **Contract ≠ delivery ≠ retirement.** The Alt Carbon, Mati and Planetary records show Frontier taking delivery in the hundreds-to-thousands of tonnes, verified via Isometric; the nine-figure BECCS contracts are all pre-FID or pre-COD bets.
Asked, verbatim: “What is the current policy landscape for durable carbon removal in the EU?”
It found every checklist fact on this question — not every question favors the database, and this page reports those too.
Judge’s note — All three gold items are substantively covered. CRCF is treated in depth (Reg. 2024/3012, implementing act, Delegated Reg. 2026/285 for DACCS/BioCCS/biochar, QU.A.L.ITY criteria, registries, pending carbon-farming and storage-in-products acts, and the gap for ERW/OAE/mineralisation). The 2026 ETS revision is covered with the mechanism of integrating permanent removal units (250 Mt cap increase, BioCCS operator-level offsetting, end-2034 review of purchasing progress) — the specific 'Article 9c' label is not named, but the substance of the removals-integration/procurement proposal is present. The CDR-only vs CDR-and-CCS scope distinction is handled explicitly and usefully (Innovation Fund treats CDR under CCS so funding is untrackable and eligibility skews to DACCS/BECCS; ICMS conflates CDR/CCS/CCU with no separate targets; NZIA 50 Mt injection obligation). Beyond-corpus: the 24 August 2026 IF26 hydrogen auction consultation deadline is a genuine post-ingest item, and member-state programmes (Sweden BECCS reverse auction, Denmark NECCS/CCUS) are named though explicitly flagged as unverified — so credit is limited; no substantive post-14-Aug-2026 EU developments beyond that. Citations are to named, checkable bodies (ICAP, CREG minutes, Carbon Gap, Bellona, ZEP, Ecologic) with dates and instrument numbers, though no links and some numbers (plenary vote tally, exact delegated-act number/date) would need independent verification. Coverage honesty is exemplary: the answer separates proposal from law, flags secondary-source reliance, names what it did not verify, and identifies the demand gap and unaddressed pathways. Structure is layered (certification / target / demand / funding / infrastructure / watchlist) and directly decision-useful for a practitioner. Deductions only for the missing explicit Article 9c hook and the reliance on unlinked secondary summaries.
I'll research the current EU policy landscape for durable carbon removal. Search results indicate significant developments through 2026. Let me dig deeper into the most recent items. ## Bottom line (as of 2 September 2026) The EU has moved from *designing* durable CDR policy to *implementing* it. In the last twelve months three things happened that collectively make the EU the most developed durable-CDR jurisdiction on the regulatory side: a binding 2040 climate target was enacted, the first EU certification methodologies for permanent removals entered into force, and the Commission tabled a proposal to bring permanent removal units into the EU ETS. What is still largely missing is **demand at scale** — no dedicated CDR volume target, no compliance obligation yet in force, and no operational EU purchasing programme. --- ## 1. The certification layer: CRCF is now live **The framework.** The EU Carbon Removals and Carbon Farming Regulation (CRCF) is a voluntary regulatory framework for certifying permanent carbon removals, carbon farming and carbon storage in products, built to speed deployment as a complement to emission reductions, fight greenwashing, and harmonise market conditions. It distinguishes four certified activity types: carbon farming (temporary storage and soil emission reductions), temporary carbon storage in long-lasting products, and permanent carbon removal. Quality is governed by the so-called **QU.A.L.ITY criteria** — quantification, additionality, permanence (long-term storage) and environmental sustainability . The Regulation mandates third-party verification and publication of certification information in an EU-wide registry, with Commission-recognised certification schemes doing the compliance checking. **What has actually been adopted:** | Instrument | Date | Content | |---|---|---| | Regulation (EU) 2024/3012 (CRCF) | Dec 2024 | Enabling framework | | Implementing Regulation (EU) 2025/2358 | Nov 2025 | Technical rules on certification schemes, certification bodies and audits | | Delegated Regulation (EU) 2026/285 | **3 Feb 2026** | First permanent-removal methodologies | Delegated Regulation (EU) 2026/285 sets out CRCF certification methodologies for direct air capture with carbon storage (DACCS) and biogenic emissions capture with carbon storage (BioCCS) , plus biochar. Adopted on 3 February 2026, it marks the shift from rule-setting to implementation — DACCS, BioCCS and biochar projects can now apply for EU certification, giving corporate buyers a government-backed quality benchmark for the first time. **Design details worth knowing for project finance.** From the final Expert Group (CREG) round on 10 November 2025: the Commission described the DACCS, BioCCS and biochar drafts as "very mature", incorporating changes from the July 2025 public consultation on biomass sustainability and reporting, temporal matching for electricity, and biochar permanence in soils. For BioCCS and DACCS the Commission moved to a 15-year activity period (up from 10) for CCS-based projects, explicitly to enable access to finance. The biochar methodology clarifies that only the portion of biochar stable enough to count as permanent can be certified. **Registries.** Certification schemes must set up interoperable public registries to track certified units; within four years these are to be replaced by a common Union-wide registry, with certification bodies supervised by Member States carrying out audits and issuing certificates. Existing voluntary-market standards (Verra, Gold Standard, Puro.earth) can apply to the Commission for recognition and continue running their own registries in the interim. **Still pending.** Two further delegated regulations are expected during 2026: carbon farming methodologies (agriculture, agroforestry, peatland rewetting, afforestation) and carbon storage in products for bio-based construction materials. Note these are *non-durable* pathways — the durable methodology set is currently limited to three methods. The Commission says it is "considering whether to develop certification methodologies for other permanent carbon removal activities" — meaning enhanced rock weathering, ocean alkalinity enhancement, bio-oil/biomass burial and mineralisation currently have **no EU pathway**, a significant gap for developers in those categories. **Contested points.** The framework is not universally endorsed. At the November 2025 CREG meeting, while the group broadly endorsed the draft, discussions revealed persistent concerns over biomass sustainability, governance and liability mechanisms. Stakeholders including Bellona argued the methodologies were designed to facilitate project development over robust carbon accounting. Bellona also wanted stricter monthly temporal correlation for DACCS electricity, where the delegated act instead accommodates early movers for investor certainty . Separately, Carbon Gap notes that despite strong supply-side criteria, the framework does not provide a comprehensive set of guardrails around the *use* of units — i.e., claims and offsetting rules are largely unaddressed. --- ## 2. The target layer: 2040 Climate Law amendment is now law The legislative sequence: the Commission tabled the proposal on 2 July 2025; environment ministers agreed the Council position on 5 November 2025 and Parliament adopted its negotiating position on 13 November 2025 . Provisional agreement came at the first trilogue on 9 December 2025, COREPER approved on 19 December, ENVI voted on 19 January 2026, and plenary adopted the final text on 9 February 2026 by 413 votes to 226 with 12 abstentions. Substance: a net 90% reduction target for 2040, with international carbon credits usable up to 5% of 1990 net emissions , available from 2036 onwards . The text includes a role for permanent domestic carbon removals under the EU ETS, but does not specify any such role for international carbon credits. **The critical gap for CDR investors:** unlike the Commission's February 2024 communication on the 2040 target, the adopted text contains no indicative breakdown of carbon removals within the net-90% target. The earlier communication had suggested residual emissions under 850 MtCO₂ by 2040 offset by up to 400 MtCO₂ of industrial and land-based removals; the omission removes an early signal on the expected scale of removals. For reference, ZEP reports that the proposed 2040 target foresaw industrial carbon removals amounting to more than 60 MtCO₂ in 2040 — but treat that as a modelling assumption, not a legal commitment. Carbon Gap's position is that separate targets for gross reductions, permanent removals and temporary land-based removals are essential to prevent removals substituting for reductions and to give clear market signals ; the EU has not adopted such split targets. --- ## 3. The demand layer: ETS integration (proposed, not enacted) The most consequential file. On 17 July 2026 the Commission published its ETS revision proposal alongside an Electrification Action Plan — a 141-page Directive proposal plus revised heat and fuel benchmarks . On removals specifically, per ICAP's summary: - The proposal incorporates domestic permanent carbon removal units into the EU ETS by increasing the overall cap by 250 million allowances. - Operators generating CRCF-certified BioCCS removals at their own facilities may offset their fossil emissions directly, with a corresponding adjustment of allocated allowances toward removal units. - By end-2034 the Commission must report on purchasing progress, supply and price conditions, the potential role of nature-based removals under the CRCF, and the feasibility of transitioning to direct operator-level integration of removal units. Surrounding architecture matters for the carbon price that ultimately underpins CDR demand: free allocation for heavy industry is prolonged beyond 2030 into the 2040s, with an immediate €6bn (80 Mt allowances) boost via revised 2026–2030 fallback benchmarks, and CBAM free-allocation phase-out postponed from 2034 to 2038 , plus a lower Market Stability Reserve intake rate of 12% from 2028 and a 173 Mt downward recalibration of the 2027 TNAC . A facility for purchasing international credits is established, with a January 2033 report on the international credits market; if high-quality, cost-effective credits are unavailable, the linear reduction factor reverts to 2.7% from 2036 and leftover set-aside allowances go to the Industrial Decarbonisation Bank. The Industrial Decarbonisation Bank is proposed as a new EU-level entity from 2028. **Uncertainty flag:** this is a *proposal*. It must go through Parliament and Council; given the political context — the review was driven by Member State pressure amid sustained high energy costs and a populist revival of climate scepticism — the 250 Mt figure, the BioCCS-first design, and the timing are all live negotiating variables. Expect trilogues into 2027. Note also the design choice: a **cap increase** of 250 Mt is a one-off headroom mechanism, not a recurring compliance obligation, and it privileges BioCCS operators already inside the ETS over merchant DACCS. --- ## 4. Supply-push funding: thinner than it looks The Innovation Fund is the main EU-level money. It is set to provide around €40bn over 2020–2030 from auctioning roughly 530 million ETS allowances. But CDR is *not* listed as a targeted area whereas CCU and CCS are, and CDR projects that do receive support get less than CCU/CCS technologies; because CDR is labelled as CCS, exact CDR funding is hard to track . This limits eligibility largely to DACCS and BECCS and forces applicants into CCS methodologies. Historic examples: Stockholm Exergi's BECCS project received €180m in 2021 and Carbfix's Silverstone €3.8m. Current calls: the 2025 Innovation Fund call opened 4 December 2025 with a €2.9bn budget and a 23 April 2026 deadline; a third hydrogen auction opened the same day with €1.3bn; and the Clean Industrial Deal announced a new €1bn pilot auction on decarbonising key industrial processes. Two new auction calls are planned in 2026, including a second Heat Auction opening December 2026 with a €1bn budget from ETS revenues , and draft terms for the IF26 Hydrogen Auction were consulted on until 24 August 2026, with launch expected by end-2026 . **There is still no dedicated CDR auction.** **A dedicated purchasing programme is under design but not adopted.** The Commission held workshops on 20–21 May 2025 to inform the design of an EU-level purchasing programme for permanent CDR, with seven policy options developed by Ecologic Institute; an EU-coordinated Buyers' Club, an EU Removals Fund, and a Member States' Centralised Procurement Agency were identified as the most promising short-term options. As far as I can see, none has been formalised into a legislative proposal — this is the single biggest near-term gap for merchant DACCS developers in Europe. --- ## 5. Enabling conditions: infrastructure and storage The Industrial Carbon Management Strategy (ICMS), published 6 February 2024, sets out the role of BECCS, DACCS and biogenic carbon alongside CCS/CCU and the measures needed to scale them — but as a Commission communication it is not legally binding. Carbon Gap's critique: it does not clearly distinguish between CDR, CCS and CCU, and sets no dedicated targets for each. The binding piece sitting under it is the Net-Zero Industry Act's obligation on oil and gas producers to deliver 50 Mt/yr of CO₂ injection capacity in the EU by 2030 — I'd flag that I did not re-verify current progress against that target in this research, and independent assessments have repeatedly questioned whether it will be met. --- ## 6. What I'd watch, and what remains genuinely uncertain 1. **ETS trilogues (late 2026–2027).** Whether the 250 Mt removal headroom survives, whether DACCS gets parity with BioCCS, and whether the "direct operator-level integration" review lands earlier than the proposed end-2034 report. 2. **A demand instrument.** Absent a Removals Fund or procurement agency, EU durable CDR demand remains voluntary-market dependent — and that market's centre of gravity is elsewhere: Frontier announced over $915m in new commitments in June 2026 from members including Google and Anthropic , and the US remains the centre of gravity for technology-based CDR . 3. **Methodology expansion** to non-CCS durable pathways (ERW, ocean, mineralisation) — currently unaddressed. 4. **Biomass sustainability and liability** rules, which remain the most contested technical issue in the BioCCS methodology. **Explicit caveats on this answer:** several figures above (the 250 Mt cap increase, the €6bn free-allocation boost, ETS report deadlines) come from secondary summaries of the July 2026 proposal rather than the Official Journal text, since the proposal is not yet law and the legal text will change. I have not verified Member-State-level schemes (e.g. Sweden's BECCS reverse auction, Denmark's NECCS/CCUS funds) in this research, and their current award status may have shifted. I also could not confirm the precise scope and volume outcomes of the Feb 2026 methodologies' first certifications — no CRCF units appear to have been issued at scale yet, and that will be the real test of whether the framework is bankable.
Judge’s note — All three gold items are covered substantively. CRCF is treated in depth (Reg. (EU) 2024/3012, entry into force, delegated methodologies for DACCS/BioCCS/biochar, implementing regulation on certification bodies, carbon farming methodologies, scheme-recognition state of play, integrity criticisms). The ETS Article 9c central-purchaser proposal is the best-developed section: 250 Mt cap increase, 2031-2040 auctioning, Commission as sole buyer, reverse auctions with ex-post payment, biochar exclusion, cost ranges, BioCCS-vs-DACCS portfolio split, review by end-2034, and the political contest (Carbon Gap architectures, CAN Europe/WWF objections). The CDR-only vs CDR-and-CCS scope distinction is handled implicitly but substantively: NZIA's 50 Mt injection target with no CDR recognition, Innovation Fund CDR ineligibility 'unless it fits existing CCS categories', CO2 transport market rules defaulting to incumbent CCS use cases, ICMS blending capture and removal figures, and the note that ERW/OAE/marine CDR have no certification route at all. Beyond-corpus findings are numerous: the 2040 Climate Law final adoption dates with 5% international credit allowance (correcting a stale corpus status), Puro.earth's June 2026 scheme application and the absence of any formal recognition, the Buyers' Club target of initial permanent-removal purchases by December 2026 and website launch, the 260 Mt international credit mechanism for 2036-2040, ICAP/BeZero/Fastmarkets confirmation of the file entering ordinary legislative procedure, the Carbon Standards EU Biochar CRCF-compliant standard consultation open to 24 September 2026, the Ocean Act consultation deadline, and the December 2028 EU registry date. Weakness: member-state programmes are only gestured at ('national schemes (Denmark, Sweden)') with no detail on the Danish NECCS tender or Swedish reverse auction volumes/prices \ Citation quality is strong on named sources (Consilium, europarl legislative train, EUR-Lex, QCIntel, Carbon Pulse, ICAP, BeZero, Fastmarkets, A&O Shearman, cdr.fyi, Carbon Gap, Bellona, Oeko-Institut) and precise instrument numbers, though no links and several regulation/COM numbers are asserted without a verifiable trail, so a reader would need to re-check them. Coverage honesty is exemplary: an explicit caveats block naming corpus freshness limits, machine-extraction and duplication problems, a stale status it corrected, the search-tool limit hit, and the three specific unverified items. Clarity and decision-usefulness are high, with a sharp framing (certification done, demand not yet law) and a pressure-points section that an expert could act on.
I'll start by understanding what the corpus covers, then dig into EU-specific policy instruments. Now let me check recent changes in the corpus and verify against the web for anything after the freshness dates. ## Short answer The EU now has the world's most built-out durable-CDR policy stack, but it is asymmetric: **certification is largely done, demand is not yet law**. The certification layer (CRCF) is in force with permanent-removal methodologies adopted; the compliance-demand layer (a 250 Mt central purchasing programme inside the ETS) is a July 2026 legislative *proposal* that still has to survive Parliament and Council; and the interim demand bridge is a voluntary Buyers' Club plus grant funding. Below, corpus-held records are flagged as such; items I verified or added by web search after the corpus freshness date (policy events held only through **2026-08-17**) are flagged separately. --- ## 1. The target layer — now binding, but CDR is not separately ring-fenced | Instrument | Status | |---|---| | European Climate Law 2040 amendment | **Adopted** — 90% net GHG cut vs 1990 | | LULUCF Regulation | In force — 310 Mt CO₂e land sink by 2030 | | Industrial Carbon Management Strategy, COM(2024) 62 | Non-binding communication, 6 Feb 2024 | *Web check:* the 2040 target is no longer a proposal. The European Parliament adopted the final text on **9 February 2026** and the **Council gave final green light on 5 March 2026**, with the signature ceremony on 11 March (consilium.europa.eu; europarl.europa.eu legislative train). The deal allows international carbon credits to cover **up to 5%** of the 90% (Latham & Watkins analysis, Nov 2025 Council position). Note the corpus still carries the 2040 target as `proposed` in several machine-extracted records — that is stale. The corpus's ICMS record (in force since 6 Feb 2024) is the strategic backbone: an aspirational **5 MtCO₂/yr of industrial carbon removal by 2030**, ~**400 MtCO₂e of residual emissions by 2050** needing compensation, and **280 Mt / 450 Mt** of total capture by 2040 and 2050 respectively. **The key gap:** there is still no binding, separate EU sub-target for durable removals. Campaigns for one go back to a January 2024 open letter from 96 signatories demanding three separate targets (reductions, land sequestration, permanent removals), and a December 2024 letter from the Negative Emissions Platform plus 43 companies to von der Leyen (both corpus policy events). The Commission has said the **CRCF review will be presented alongside a proposal for national targets at the end of 2026** (Commission news item, 29 May 2026) — that is the single most important thing to watch. ## 2. The certification layer — CRCF, essentially operational - **Regulation (EU) 2024/3012** (CRCF): EP first reading 10 Apr 2024, Council adoption 19 Nov 2024, OJ 6 Dec 2024, **in force 26 December 2024**. Covers permanent removals, carbon farming and carbon storage in products. - **Delegated Regulation (EU) 2026/285**: certification methodologies for **DACCS, BioCCS and biochar**, adopted **February 2026** — described by the Commission as the world's first voluntary standard for permanent removals. The ENVI committee **voted down attempts to reject the methodologies on 17 March 2026** (QCIntel, corpus event). - **Implementing Regulation (EU) 2025/2358** (adopted ~1 Dec 2025): transparency standards, appointment and supervision of certification bodies, audit rules, mandatory public consultation and complaint handling; enables group auditing in carbon farming. Stricter conflict-of-interest rules for verifiers were finalised **24 November 2025**. - **Carbon farming methodologies** adopted **10 July 2026**: mineral-soil agriculture/agroforestry, peatland rewetting, afforestation — expected in force Q3 2026. - **Scheme recognition is live.** Applications opened early 2026, handled first-come first-served, recognitions valid five years. *Web check:* **Puro.earth formally applied in June 2026** and has started National Accreditation Body proceedings in Finland; I found **no confirmation that any scheme has yet been formally recognised**. Verification bodies must additionally be accredited by NABs (or the European co-operation for Accreditation for non-EU HQs) before any CRCF unit can be issued — a real bottleneck. - **EU central registry** is due by **December 2028** (EUR-Lex summary). Still in the pipeline: methodologies for **carbon storage in buildings/construction materials**, exploratory work on **direct ocean capture (DOCCS)** after a February 2026 expert workshop, and a **CRCF revision that will add a green-claims chapter**, reviving parts of the stalled Green Claims Directive (Carbon Pulse, corpus). Separately, the Swiss-based Carbon Standards launched a **CRCF-compliant EU Biochar Carbon Removal Standard** in public consultation to 24 September 2026 — a private standard, not an EU instrument. **Integrity is contested.** Bellona reported dissent within the Carbon Removal Expert Group in November 2025; Öko-Institut has argued the revised methodologies still lack integrity; Carbon Pulse-sourced records note criticism of ambiguous language, weak environmental guardrails relative to UN Article 6 rules, and weak conflict-of-interest provisions in the implementing regulation. ## 3. The demand layer — the live fight ### 3a. ETS integration (COM(2026) 616, proposed 17 July 2026) This is the centre of gravity. Under a proposed **Article 9c**: - The ETS cap is increased by **250 million allowances**, auctioned by the Commission over **2031–2040**, with proceeds held as external assigned revenue. - The Commission acts as **sole central buyer**, procuring CRCF-certified **DACCS and BioCCS** units via competitive tendering or reverse auctions with ex-post payment on delivery. **Biochar is excluded** on evidentiary grounds (limited long-term field data, degradation uncertainty), with a pathway for later inclusion. - Contingency: up to **260 Mt** if auction revenue is insufficient; scale-up to roughly **48 Mt/yr by 2040**; mandatory review by **end-2034**. - Cost estimates in the corpus record: **€29–43bn (PRIMES)** or **€46–65bn (POTEnCIA)**; Fastmarkets framed it as a "€50bn" stimulus, heavily favouring pulp-and-paper BECCS. cdr.fyi's analysis warns that auction revenue only supports an average removal price of about **104% of the average EUA price** — buying 250 Mt at €100 above the EUA price would need roughly **€25bn extra**. - The Commission's impact assessment portfolios are dominated by **BioCCS (~175–330 Mt over 2031–40)** versus **~20–35 Mt DACCS (~11%)**, though no technology split is prescribed. - ETS-liable operators (e.g. shipping, aviation) may use their **own** CRCF-certified removals against fossil emissions but cannot "go beyond zero"; such volumes are deducted from the 250 Mt. *Web check on status:* ICAP confirmed the 17 July 2026 publication and cap increase; BeZero and Fastmarkets both note the file has entered the ordinary legislative procedure, with Parliament and Council to form positions before trilogues, and that the removals and international-credit provisions will be the most closely scrutinised. A&O Shearman's read adds the parallel mechanism of up to **260 Mt of international credits from 2036–2040**, implying a domestic reduction of −85%. **Nothing is settled**; expect delegated and implementing acts afterwards. Carbon Gap's reaction (corpus) frames three integration architectures on the table — (1) a public authority buying and retiring credits, (2) direct operator purchase, (3) "one-in-one-out" where each surrendered CDR credit cancels an EUA. Carbon Gap favours (1) plus a one-in-one-out cap. Critics are loud: CAN Europe's Sven Harmeling said adding removals would "weaken the ETS impact, undermine the carbon price and create new loopholes"; WWF and others estimate the overall package permits around **2 bn tonnes of extra emissions**. ### 3b. EU CRCF Buyers' Club (operational) Announced in the **Bioeconomy Strategy, November 2025**; details set out at the first **CRCF Days, 20–21 May 2026** in Brussels (500+ stakeholders). A hybrid public-private demand-aggregation platform with two tracks — permanent removals (DACCS, BioCCS, BCR) aimed at multi-year offtakes and FID, leveraging Innovation Fund and member-state money; and carbon farming. *Web check:* the Commission's Buyers' Club page (July 2026) states the Club **aims to make an initial set of purchases of permanent removals by December 2026**, and the official website went live in June 2026. Several offtake announcements were signalled for end-2026. Treat "operational" as *convening*, not *contracting* — I found no confirmed executed purchase. ### 3c. Grant and R&D funding - **Innovation Fund** (ETS-funded, CINEA-run): ~**€15bn to ~250 projects** to date; **€2.7bn for 54 clean-industry projects announced 24 March 2026**, including carbon management. Landmark CDR award remains **BECCS Stockholm, ~€180m (2022)**. Carbon Gap's July 2025 consultation response argued CDR "remains largely ineligible unless it fits existing CCS categories" and pushed for a dedicated CDR window, CfDs and a storage-durability selection criterion. - **Horizon Europe 2025–2027**: Carbon Gap counts **25 CDR-relevant calls worth ~€389m** within a €7.3bn pot (11 direct calls, €184.5m). Direct CDR funding rose from ~0.9% (2023–24) to **2.4%**; total direct+indirect from 2.6% to 5%. Four direct DAC calls (up from one) and first explicit ocean alkalinity enhancement calls. Carbon Gap's ask: **€2.6bn** for early-stage CDR through the next MFF and FP10. ## 4. Adjacent files worth tracking - **CO₂ transport infrastructure and markets Regulation** — a forthcoming legislative initiative stemming from the ICMS; Carbon Gap submitted consultation feedback 11 Sept 2025 urging that market rules not default to incumbent CCS use cases. - **European Ocean Act / Maritime Spatial Planning Directive** — consultation closed 16 July 2026; Carbon Gap asked for a coherent **marine CDR strategy**, ocean observation/MRV and dedicated research funding. - **Net-Zero Industry Act** — a **50 Mt/yr CO₂ injection capacity target by 2030** and Article 18 obligations on oil and gas producers, but no explicit recognition of CDR as a strategic net-zero technology. - **ESRS / CSRD** — requires disaggregated disclosure of removals vs reductions and by sink type; the green-claims chapter in the CRCF revision will interact with this. - **Nature Restoration Law** — legally binding restoration targets, though the cropland soil-organic-carbon provision was deleted in amendment. ## 5. Where the pressure points are 1. **Timing mismatch.** Certification exists now; compliance demand starts **2031 at the earliest**, and only if the ETS review survives trilogue intact. The 2027–2030 gap is covered only by grants, the Buyers' Club and national schemes (Denmark, Sweden). 2. **Price gap.** Auction-financed procurement priced near the EUA level will not clear DACCS at current costs. The Commission's own impact assessment leans on CCfDs, the Industrial Decarbonisation Bank and national support to bridge it — none of which is legislated for CDR specifically. 3. **Pathway concentration.** BioCCS dominates modelled volumes; biochar is excluded from ETS procurement despite CRCF recognition; ERW, OAE and marine CDR have no EU certification route at all yet. 4. **A binding removals target is still missing** — the end-2026 CRCF review plus national-targets proposal is the decision point. 5. **Integrity and legitimacy risk.** Expert-group dissent, NGO objections to cap inflation, and unresolved biomass sustainability questions could all reopen settled methodologies. ## Caveats on this answer The Carbon Curve corpus holds **policy events only through 2026-08-17** and instruments to ~2026-09-02; several EU records are machine-extracted (`review_status: auto`), partly duplicative (three separate records describe the same July 2026 ETS review) and carry stale statuses (the 2040 target listed as "proposed" when it was adopted in March 2026). My web checks surfaced pages dated up to roughly **late July 2026**; I hit the search-tool limit before I could verify Parliament/Council negotiating positions on the ETS review, whether any CRCF certification scheme has been formally recognised, or whether the Buyers' Club met its December 2026 purchase goal. Those three are the items I would re-check before relying on this.
Judge’s note — All three gold checklist items are covered substantively. CRCF is treated in depth (Regulation (EU) 2024/3012, adoption/entry-into-force dates, QU.A.L.ITY criteria, delegated/implementing acts, ENVI scrutiny vote, CRCF Days, scheme-recognition pipeline including Puro.earth's application, and the end-2026 review plus national-targets proposal). The ETS Article 9c proposal is described with unusual instrument-level precision: 250 Mt central procurement 2031-2040, 250m extra allowances with 260m contingency, Commission as sole buyer, reverse auction/ex-post payment, 48 Mt/yr by 2040, 2034 review, impact-assessment cost ranges, and the important caveat that allowance release is not conditional on delivery. Scope distinctions between CDR-only and CDR-and-CCS instruments are handled explicitly (BioCCS/DACCS eligible, biochar excluded with reasoning; NZIA covering 50 Mt injection capacity but not listing CDR; Innovation Fund CDR eligibility only where it fits CCS categories; CO2 transport regulation risk of defaulting to incumbent CCS use cases). Beyond-corpus material is present and clearly demarcated: post-ingest/forward items (Q1 2027 trilogue target, ICAP international-credit facility with 2033 report, €100bn Industrial Decarbonisation Bank, EU BCR consultation deadline 24 Sept 2026, Ocean Act consultation close 16 July 2026), correction of the stale 2040 target record (provisional agreement 10 Dec 2025; Council adoption 5 March 2026 with ~5% international credits and ETS2 delay), and named member-state programs (Sweden BECCS auction, Denmark NECCS/Aalborg Portland DKK 16.5bn, Germany carbon management strategy) — though the member-state layer is only named rather than developed. Citations are consistently named and checkable (EUR-Lex, Consilium, Commission news items, Carbon Gap, ICAP, Abatable, Linklaters, QCIntel, Carbon Pulse, Puro.earth), with regulation numbers and dates. Coverage honesty is exemplary: it distinguishes approved vs auto-extracted records, flags a five-record deduplication artefact for the same ETS proposal, identifies an internal 3% vs 5% inconsistency, states the corpus cutoff, and lists four specific unresolved items with an instruction to verify them. Structure is clean and decision-useful. Main residual risk is that several specific act numbers and figures are unverifiable here and, if any are fabricated, would be confidently misleading; the answer is also long, but the layering (certification / demand / targets / money / pipeline / caveats) keeps it navigable.
I'll start by understanding what the Carbon Curve corpus covers, then pull the EU policy landscape. Now let me check recent policy changes and then verify against the web, since the corpus stops at mid-August 2026. ## Summary The EU now has the most complete durable‑CDR policy architecture of any jurisdiction, and as of the latest records it sits at a specific inflection point: **the certification layer is built and in force; the demand layer is proposed but not law.** Two sources below: records from the Carbon Curve corpus (policy events recorded through **2026‑08‑17**; the EU landscape returned ~40 instruments) and web searches I ran to cover the period after that and to check the corpus's own status labels. I flag where they disagree. --- ## 1. The certification layer — built, in force, operating **CRCF Regulation (EU) 2024/3012** is the spine. Corpus record (review_status: *approved*, sourced to EUR‑Lex): European Parliament adopted at first reading **10 April 2024**, Council formally adopted **19 November 2024**, published in the Official Journal **6 December 2024**, entered into force **26 December 2024**. It covers four activity types — permanent removals, carbon farming, and carbon storage in products — under the "QU.A.L.ITY" criteria. The methodologies underneath it landed in 2026: | Instrument | Status | Date | |---|---|---| | **Delegated Regulation (EU) 2026/285** — DACCS, BioCCS, biochar | in force | adopted Feb 2026 | | **Implementing Regulation (EU) 2025/2358** — scheme transparency, certification‑body appointment/supervision, audits, group auditing | in force | adopted ~1 Dec 2025 | | Certification‑body conflict‑of‑interest and oversight rules | legislated | final draft adopted 24 Nov 2025 | | Carbon farming methodologies — mineral soils/agroforestry, peatland rewetting, afforestation | adopted, awaiting entry into force after scrutiny | **10 July 2026** | | Revised afforestation methodology | in force | 16 May 2025 | Two governance events matter. The corpus records that on **17 March 2026 the European Parliament's ENVI committee voted down attempts to reject** the Commission's CDR methodologies (source: QCIntel) — the methodologies survived scrutiny. And at **CRCF Days, 20–21 May 2026 in Brussels, with over 500 stakeholders**, the Commission confirmed certification schemes can now apply for recognition (source: Commission news item, 29 May 2026). **On recognition of schemes**, web search confirms movement the corpus captures only partially: Puro.earth submitted a formal application to the Commission in June 2026 following a Commission guidance webinar on 1 June (Puro.earth, H2 Bulletin, Bioenergy International). Verra, Gold Standard and others are eligible to apply. Note the accreditation chain is slow: the corpus record (auto) states validation/verification bodies must be accredited by National Accreditation Bodies *after* scheme recognition, and Puro has begun that process in Finland. **I could not confirm whether any scheme has yet been formally recognised** — my search budget ran out before I could verify post‑August 2026 recognitions. Treat this as an open item. Also open: a **private, CRCF‑aligned biochar standard** — Carbon Standards International's EU Biochar Carbon Removal Standard (EU BCR) — was in public consultation with feedback due **24 September 2026** (corpus record, *auto*, from Carbon Herald). This is a registry standard, not EU legislation. --- ## 2. The demand layer — the big one, and it is only a proposal On **17 July 2026 the Commission published its ETS review proposal**, the first mechanism that would create compliance‑grade demand for durable CDR at scale. Corpus records (the Article 9c record is *approved*, sourced to cdr.fyi; a parallel COM(2026) 616 record is *auto*): - The EU‑wide allowance quantity rises by **250 million allowances**, auctioned, with proceeds ring‑fenced as external assigned revenue to **centrally purchase 250 Mt of CRCF‑certified permanent removals over 2031–2040**. Article 9c allows a contingency increase to 260 million if revenue falls short. - **The Commission is the sole buyer.** Individual emitters cannot generally use removals for their own compliance — though the proposal would let ETS‑liable operators use their own CRCF‑certified BECCS removals against fossil emissions, deducted from the 250 Mt total. - **Eligible pathways: BioCCS and DACCS only. Biochar is explicitly excluded** on evidentiary grounds (limited long‑term field studies, degradation‑rate uncertainty), with a route to later inclusion. - Procurement via **competitive tendering or reverse auctions, paid ex‑post on delivery**; scaling to **48 Mt/yr by 2040**; statutory review by end‑2034. - The Commission's impact assessment modelled portfolios dominated by BioCCS (~175–330 Mt) versus roughly 20–35 Mt DACCS (~11%), though no technology split is prescribed. Budget estimates range **€29–43bn (PRIMES) to €46–65bn (POTEnCIA)**. Carbon Gap's reading (corpus record, *approved*) is that the proposal frames three integration options — a public buyer, direct operator integration, and a "one‑in‑one‑out" model where each surrendered CDR credit cancels an EUA — and that EU and independent modelling puts EU CDR need at **75–280 MtCO₂e/yr by 2040**, against roughly **180 CDR companies currently operating in Europe**. **What the web adds and the corpus does not have:** this is nowhere near enacted. Parliament and Council must each adopt negotiating positions before trilogues, with the Commission targeting conclusion by **Q1 2027** — described by Abatable as a tight timeline. ICAP notes the proposal also establishes a facility for purchasing international credits, with a Commission report to Parliament and Council due **by January 2033**. Linklaters characterises the removals integration as one of the most significant carbon‑removal policy developments to date, arriving amid heavy member‑state pressure over energy costs. Criticism is live: the corpus holds CAN Europe's Sven Harmeling arguing removals in the ETS would weaken it and create loopholes, and WWF and other analysts estimating the wider package permits ~2bn tonnes of extra emissions. **A caution on the corpus here:** it holds at least five separate instrument records describing the same July 2026 ETS proposal (Article 9c programme, COM(2026) 616, two "ETS review proposal" records, and the Carbon Gap reaction record), most machine‑extracted and *auto*‑published. That is a deduplication artefact, not five distinct instruments. ### Voluntary demand: the EU CRCF Buyers' Club Announced under the Bioeconomy Strategy in **November 2025**, recorded as **operational** (record *auto*, sourced to the Commission). A hybrid public‑private platform in two tracks — permanent removals (DACCS, BioCCS, BCR), aiming at multi‑year offtakes and FID support leveraging Innovation Fund and member‑state money; and carbon farming. It **aims to make initial purchases of permanent removals by December 2026**, with knowledge‑sharing, due‑diligence tools and standard contract templates from autumn 2026. The Commission's May 2026 statement said several offtake agreements were expected by end‑2026. --- ## 3. Targets — where the corpus is out of date The corpus lists the **EU 2040 Climate Goal as "proposed"** (record *auto*, sourced to a Politico piece). **This is stale.** Web sources confirm the file completed: Council and Parliament reached provisional agreement on **10 December 2025** on a 90% net GHG reduction for 2040, and the **Council gave final adoption of the amended European Climate Law on 5 March 2026** (Consilium press releases). Flexibilities in the final text include **up to ~5% international carbon credits**, review clauses linked to land‑sink performance, and ETS2 delayed to 2028 (Carbon Gap tracker; Morgan Lewis). Note a second internal inconsistency: one corpus record describes international offsetting at **3% of 1990 levels from 2036** (~460 Mt), reflecting the July 2025 Commission proposal, while another describes **up to 5%**. The 5% figure matches the adopted text; the 3% figure is the superseded original. Other target-setting instruments in the corpus: - **Industrial Carbon Management Strategy, COM(2024) 62, 6 February 2024** (in force, non‑binding communication): aspirational **5 MtCO₂/yr of industrial removals by 2030**; ~400 MtCO₂e residual emissions by 2050 requiring CDR; 280 Mt and 450 Mt captured by 2040 and 2050. - **LULUCF**: 310 Mt CO₂e land‑sector net removals by 2030 (via the Nature Restoration Law record). - **Net‑Zero Industry Act**: 50 Mt/yr CO₂ injection capacity by 2030, but the corpus record notes CDR is *not* included as a strategic net‑zero technology. - **A CRCF review plus a proposal for national targets is due at the end of 2026** — confirmed by both the corpus event and the Commission's own May 2026 page. This is arguably the single most consequential upcoming item after the ETS file. --- ## 4. Money - **Innovation Fund** (operational since July 2020, ETS‑auction financed): Commission estimate of ~€40bn 2020–2030 from monetising 530m allowances; ~€15bn awarded to ~250 projects to date. Most recent event in corpus: **€2.7bn to 54 clean‑industry projects on 24 March 2026**, including carbon‑management projects. The corpus flags the BECCS Stockholm ~€180m (2022) award and specific DAC awardees as **unverified against primary award lists**. Carbon Gap's July 2025 consultation response argued CDR "remains largely ineligible unless it fits existing CCS categories." - **Horizon Europe 2025–2027**: Carbon Gap's mapping (corpus record, *approved*) identifies **25 CDR‑relevant calls worth an indicative €389m within a €7.3bn pot** — 11 direct (€184.5m) and 14 indirect (€204.5m). Direct CDR rose from ~0.9% of the 2023–24 budget to 2.4%. Four direct DAC calls (up from one) and explicit ocean alkalinity enhancement calls. - **Web adds**: a **€100bn Industrial Decarbonisation Bank** accompanies the ETS package (Abatable), which also notes the public share of European carbon‑removal funding has been *falling* through 2024–2026 YTD. --- ## 5. The pipeline — proposed, consulting, or exploratory - **CRCF revision with a green claims chapter** — the Commission plans to fold parts of the stalled Green Claims Directive into the CRCF revision (record *auto*, Carbon Pulse). - **Updated CRCF implementing regulation** tightening scheme and auditor oversight — the corpus record notes criticism of weak conflict‑of‑interest rules. - **CO₂ transport infrastructure and markets Regulation** — call for evidence stage; Carbon Gap responded 11 September 2025 urging rules that don't default to incumbent CCS use cases. - **European Ocean Act / Maritime Spatial Planning Directive** — consultation closed **16 July 2026**; Carbon Gap asked for a coherent EU mCDR strategy, ocean observation/MRV, and dedicated research funding. - **DOCCS certification** — a Commission expert workshop in **February 2026** launched exploratory work toward possible EU‑wide certification of direct ocean capture. - **Carbon storage in buildings** methodology — draft delayed past H2 2025. - **A "conversion factor" / exchange‑rate mechanism for CDR credits in the ETS** — under consideration among options (record *auto*, Carbon Pulse). - **ESRS (CSRD)** — E1 requires removals to be disclosed separately from reductions and disaggregated by biological vs technological sink. --- ## Uncertainty and gaps — read this before acting 1. **Provenance.** A large majority of the EU instrument records above are `review_status: auto` — machine‑extracted from a cited source and auto‑published with no human spot‑check. Only the CRCF core, the 2026/285 methodology, the Article 9c record, the Horizon Europe mapping, the Ocean Act, the CO₂ transport regulation and the Innovation Fund carry `approved`. Several *auto* records are sourced to paywalled Carbon Pulse headlines and carry thin detail. 2. **Staleness.** Corpus policy events stop at **17 August 2026**. The 2040 target record is demonstrably out of date (still "proposed" when it was adopted in March 2026), which should make you sceptical of every other "proposed" label in that lane. 3. **Unresolved by either source.** Whether any CRCF certification scheme has yet been formally recognised; whether the carbon‑farming delegated act has completed scrutiny and been published in the OJ; whether the Buyers' Club has made its first purchase; the Parliament rapporteur and Council presidency positions on the ETS removals file. My search budget was exhausted before I could confirm these. Anyone relying on this should check those four specifically. 4. **What neither covers.** Member‑state instruments — Sweden's BECCS reverse auction, Denmark's CCS/NECCS funds (the corpus does hold the June 2026 Aalborg Portland DKK 16.5bn award), Germany's carbon management strategy — sit outside "EU" as a jurisdiction tag and were not enumerated here. Nor does either source tell you how much of the 250 Mt is financeable in practice: the corpus's own Article 9c record warns the legal text does **not** make allowance release conditional on removal contracting or delivery, and that 250 Mt is "an intended scale/potential ceiling, not a financed purchase commitment."
Judge’s note — All three gold items are covered substantively and with unusual depth. CRCF is traced from Regulation (EU) 2024/3012 through the implementing regulation, the Feb 2026 delegated permanent-removals methodologies (DACCS/BioCCS/biochar), the carbon-farming methodologies, and — importantly — the practical recognition/accreditation bottleneck that means no CRCF units can yet be issued. The ETS Article 9c proposal (COM(2026) 616) is described at design level: 250 Mt (contingency 260 Mt) added to the cap over 2031–2040, central Commission purchaser, competitive tender/reverse auction, ex-post payment, ~48 Mt/yr by 2040, DACCS/BioCCS eligibility with biochar excluded, plus the sharp critique that allowance release is not conditional on delivery so the volume is a ceiling rather than a financed commitment. Scope distinctions (CDR-only vs CDR-and-CCS) are handled well and in several places: biochar CRCF-eligible but ETS-procurement-excluded; Denmark's removals-only NECCS Fund vs the broader CCS Fund; Innovation Fund CDR 'largely ineligible unless it fits existing CCS categories'; Sweden's bio-CCS call excluding CCU and biochar; NZIA's 50 Mt injection target with no CDR listing. Beyond-corpus-type material is present in substance even though the answer states it could not run live searches: three member-state programs are detailed (Sweden's SEK 36bn BECCS reverse auction and the Stockholm Exergi award plus the Danube state-aid appeal and Slite rejection; Denmark's DKK 28.7bn CCS Fund tender and NECCS contracted volumes; Germany's KSpTG and the first dedicated negative-emissions budget line), and two forward-dated deadlines (biochar standard consultation 24 Sep 2026; Industriklivet 31 Aug 2026) are surfaced. What is missing is any actual post-cutoff verification — Parliament/Council positions on the ETS file, Buyers' Club first purchases, and the end-2026 CRCF review with national targets are all named as unknown rather than resolved. Citations are consistently named and checkable (EUR-Lex ELI, Commission pages, CINEA, Energimyndigheten, Carbon Gap, cdr.fyi, trade press with article IDs). Honesty is exemplary: explicit freshness boundary, explicit tool limitation, an unresolved 3% vs 5% international-credit conflict flagged rather than papered over, German budget figures flagged as draft-verified only, and a provenance caveat that many records are machine-extracted from secondary trade press. Structure is decision-useful — three-layer framing (certification built, demand proposed, targets contested) plus money, adjacent files, member states, and a residual-uncertainty list. Deduction only for the absence of live verification of the several items the answer itself identifies as falling just past its visibility.
I'll start by understanding what the corpus covers, then dig into EU-specific policy. Let me pull recent policy events and the EU procurement/funding picture. Let me check member-state instruments, which sit alongside the EU-level layer. ## Bottom line The EU now has the most complete durable-CDR policy stack of any jurisdiction — but only one of its three layers is actually built. **Certification is done and operating. Demand is a legislative proposal. Targets are contested.** Everything below is drawn from the Carbon Curve corpus, whose policy events run only through **17 August 2026** and instruments to **2 September 2026**. I should flag upfront that **I have no web-search tool available in this session**, so I could not execute the corpus's standing instruction to verify post-freshness developments. Treat anything after roughly September 2026 as unknown to me, not as absent. --- ## Layer 1 — Certification: built and running (CRCF) **Regulation (EU) 2024/3012**, the Carbon Removals and Carbon Farming Certification Framework, was adopted by Parliament 10 April 2024, by Council 19 November 2024, published in the OJ 6 December 2024 and **entered into force 26 December 2024** ([EUR-Lex](https://eur-lex.europa.eu/eli/reg/2024/3012/oj)). It covers three unit types: permanent removals, carbon farming, and carbon storage in products. The scaffolding has since been filled in fast: | Date | Instrument | Content | |---|---|---| | 24 Nov 2025 | Certification-body oversight rules | Stricter conflict-of-interest and accreditation rules for verifiers ([Carbon Pulse](https://carbon-pulse.com/461055)) | | 1 Dec 2025 | **Implementing Reg (EU) 2025/2358** | Transparency standards for schemes, appointment/supervision of certification bodies, mandatory public consultations, group auditing for carbon farming ([EC](https://climate.ec.europa.eu/news-other-reads/news/commission-adopts-rules-and-launches-initiatives-boost-carbon-removals-and-carbon-farming-eu-2025-12-01_en)) | | Feb 2026 | **Delegated Reg (EU) 2026/285** | The permanent-removals methodologies: **DACCS, BioCCS and biochar** ([EC](https://climate.ec.europa.eu/eu-action/carbon-removals-and-carbon-farming/certification-methodologies_en)) | | 17 Mar 2026 | ENVI committee vote | MEPs voted down attempts to reject the CDR methodologies ([QCIntel](https://www.qcintel.com/carbon/article/meps-advance-eu-cdr-methodologies-with-key-vote-60942.html)) | | 20–21 May 2026 | First CRCF Days, Brussels | 500+ stakeholders; scheme recognition applications open ([EC](https://climate.ec.europa.eu/news-other-reads/news/europe-lays-foundations-robust-carbon-removals-and-carbon-farming-market-2026-05-29_en)) | | 10 Jul 2026 | Carbon farming methodologies adopted | Agriculture/agroforestry on mineral soils; peatland rewetting; afforestation — entry into force after scrutiny ([EC](https://climate.ec.europa.eu/news-other-reads/news/commission-adopts-certification-methodologies-carbon-farming-under-crcf-regulation-2026-07-10_en)) | **The practical bottleneck is the recognition chain, not the rules.** Puro.earth has submitted its CRCF Program application to the Commission and is awaiting review; even after scheme recognition, Validation and Verification Bodies must be accredited by National Accreditation Bodies (Puro says it has started in Finland) — and NAB accreditation *only begins after* scheme recognition ([Puro.earth](https://puro.earth/insights/post/getting-ready-for-the-eu-crcf-a-conversation-with-boris-lagadinov-head-of-crcf-program-puro-earth)). No CRCF-eligible credits can be issued until that sequence completes. Separately, Carbon Standards opened a 30-day consultation on a CRCF-compliant **EU Biochar Carbon Removal Standard**, feedback due **24 September 2026** ([Carbon Herald](https://carbonherald.com/public-consultation-now-open-for-new-crcf-compliant-biochar-certification-methodology)). Integrity criticism is persistent rather than fatal: Öko-Institut and Bellona have argued the methodologies lack environmental guardrails relative to UN Article 6 rules, and the draft implementing regulation was criticised for weak conflict-of-interest provisions ([Carbon Pulse 417076](https://carbon-pulse.com/417076), [447997](https://carbon-pulse.com/447997)). --- ## Layer 2 — Demand: the big open question (ETS Article 9c) On **17 July 2026** the Commission published its ETS review proposal, **COM(2026) 616**, entering the ordinary legislative procedure ([ICAP](https://icapcarbonaction.com/en/news/eu-commission-publishes-eu-ets-review-proposal)). The CDR mechanism is the single most consequential item in the European landscape: - A new **Article 9c** would add **250 million allowances** to the EU-wide cap (with a contingency to **260 mn**) over **2031–2040**, auction them, and ring-fence the proceeds as external assigned revenue to **centrally purchase CRCF-certified permanent removals**, with the Commission as sole buyer. - Eligible pathways: **DACCS and BioCCS only**. **Biochar is explicitly excluded** on evidentiary grounds — limited long-term field data, degradation-rate uncertainty — despite being CRCF-recognised, with a pathway for later inclusion ([cdr.fyi](https://www.cdr.fyi/blog/the-eu-ets-proposal-explained)). - Procurement by competitive tender or reverse auction, **ex-post payment on delivery**, scaling to roughly **48 Mt/yr by 2040**. The Commission's impact assessment portfolios are BioCCS-dominated (~175–330 Mt) versus ~20–35 Mt DACCS. - ETS-liable operators (e.g. shipping, aviation) could use their **own** CRCF-certified removals against fossil emissions but not "go beyond zero"; those volumes are deducted from the 250 Mt. - Budget estimates diverge sharply by model: **€29–43bn (PRIMES)** or **€46–65bn (POTEnCIA)**; Fastmarkets framed it as a "€50bn alley-oop" to pulp-and-paper BECCS ([Fastmarkets](https://www.fastmarkets.com/insights/the-eus-50bn-alley-oop-to-beccs-projects-in-the-pulp-paper-industry)). **The design fight is live and specific.** Carbon Gap identifies three models on the table — a public purchaser, direct operator integration, and "one-in-one-out" (each CDR credit surrendered cancels an EUA) — and favours the public purchaser *combined with* a one-in-one-out cap ([Carbon Gap](https://carbongap.org/insights/reaction-to-the-legislative-proposal-on-carbon-removals-in-the-eu-emissions-trading-system)). Two structural criticisms recur in the corpus: 1. **The price gap.** cdr.fyi notes the auction revenue supports an average removal price only ~104% of the average EUA auction price; if removals cost €100 more than an EUA, buying 250 Mt would need ~€25bn of *additional* money. The legal text does not make allowance release conditional on removal contracting or delivery, nor cancel unmatched allowances if procurement falls short — so 250 Mt is a **ceiling and an intention, not a financed commitment**. 2. **Environmental integrity.** WWF and others estimate the wider ETS package permits ~2 bn tonnes of extra emissions; CAN Europe's Sven Harmeling says adding removals "would weaken the ETS impact, undermine the carbon price and create new loopholes" ([Carbon Brief](https://www.carbonbrief.org/qa-what-the-eus-carbon-market-review-means-for-climate-action)). The Commission has also floated a **"conversion factor" / exchange-rate mechanism** for CDR credits in the ETS as one of several options ([Carbon Pulse 414413](https://carbon-pulse.com/414413)). **Interim demand instrument: the EU CRCF Buyers' Club.** Announced under the Bioeconomy Strategy in November 2025 and detailed at CRCF Days in May 2026, it is a hybrid public-private demand-aggregation platform running two tracks (permanent removals — DACCS/BioCCS/BCR — and carbon farming). It aims to make **initial purchases of permanent removals by December 2026** and to help projects reach FID by leveraging Innovation Fund and member-state money; standardised contract templates and due-diligence tools are due from autumn 2026 ([EC](https://climate.ec.europa.eu/eu-action/carbon-removals-and-carbon-farming/eu-crcf-buyers-club_en)). The Commission flagged "several potential offtake agreements" expected by end-2026 — I cannot confirm whether any have landed. --- ## Layer 3 — Targets: agreed in principle, unresolved in detail The **2040 climate target** (90% net GHG cut vs 1990) sits in an amendment to the European Climate Law. Its CDR-relevant features: - **Domestic permanent removals** are named as a distinct compliance route. - **International credits** under Paris Article 6 are permitted from **2036**. Here the corpus carries a genuine conflict: one record cites **up to 3% of 1990 EU emissions** (~460 Mt, with a €10bn+ cost estimate at €30/credit for 300–400 mn cumulative credits) ([WEF](https://www.weforum.org/stories/2025/07/eu-return-to-international-carbon-credits)), another cites **up to 5%** with a pilot phase from ~2031 ([DVNE](https://dvne.org/en/news-en/international-removal-credits-eu)). These likely reflect different proposal iterations; I could not resolve which is the enacted figure. - **Separate removal sub-targets remain unresolved.** France called for split gross-reduction and removal targets at the December 2024 Environment Council; 96 organisations demanded three separate targets in a January 2024 open letter; the Negative Emissions Platform and 43 companies made the same ask for the Clean Industrial Deal ([Carbon Herald](https://carbonherald.com/in-a-new-open-letter-44-organizations-call-for-cdr-inclusion-into-the-eu-clean-industrial-deal)). Six countries (four member states plus Norway and Iceland) pressed the Commission in September 2025 to assess permanent CDR in the 2040 framework ([Carbon Pulse 436579](https://carbon-pulse.com/436579)). - **Watch end-2026:** the Commission signalled the **CRCF review will be presented with a proposal for national targets** at the end of 2026 ([EC, 29 May 2026](https://climate.ec.europa.eu/news-other-reads/news/europe-lays-foundations-robust-carbon-removals-and-carbon-farming-market-2026-05-29_en)). That is the next major structural decision point and falls just past my visibility. Underlying volume estimates vary widely: the **Industrial Carbon Management Strategy** (COM(2024) 62, in force 6 Feb 2024) sets an aspirational **5 MtCO₂/yr of industrial removal by 2030** and projects ~400 Mt of residual emissions needing compensation by 2050; Carbon Gap cites EU modelling and independent analysis at **75–280 MtCO₂e/yr by 2040**; the Global South CDR Coalition cites 447 Mt of removals needed within the bloc for 2040. --- ## Money on the table - **Innovation Fund** (ETS-financed, ~€40bn estimated 2020–2030 from 530 mn allowances; ~€15bn awarded to ~250 projects to date). March 2026: **€2.7bn to 54 clean industry projects**, including carbon management ([CINEA](https://cinea.ec.europa.eu/news-events/news/commission-unlocks-eu27-billion-54-clean-industry-projects-under-innovation-fund-2026-03-24_en)). BECCS Stockholm received ~€180m in 2022. Carbon Gap's July 2025 evaluation response argues CDR "remains largely ineligible unless it fits existing CCS categories" and calls for a dedicated CDR window, CfDs and a durability selection criterion. - **Horizon Europe 2025–2027**: Carbon Gap maps **25 CDR-relevant calls worth ~€389m** within a €7.3bn pot (11 direct calls, €184.5m). Direct CDR funding rose from ~0.9% of the 2023–24 budget to **2.4%**; total direct+indirect from 2.6% to 5%. Four direct DAC calls (up from one), explicit ocean alkalinity enhancement calls, and MRV calls up from one to three. Carbon Gap's ask for the next MFF/FP10 is **€2.6bn**. --- ## Adjacent and still-forming - **CO₂ transport infrastructure and markets regulation** — call for evidence closed 2025; stems from the ICMS. Carbon Gap urged rules that don't default to incumbent CCS use cases. - **European Ocean Act / Maritime Spatial Planning Directive** — consultation closed 16 July 2026; Carbon Gap asked for a coherent mCDR strategy, ocean observation/MRV and dedicated research funding. A separate Commission expert workshop in February 2026 began exploratory work on **DOCCS certification** ([Carbon Pulse 483916](https://carbon-pulse.com/483916)). - **CRCF revision with a green-claims chapter** — the Commission plans to fold parts of the stalled Green Claims Directive into the CRCF revision ([Carbon Pulse 486627](https://carbon-pulse.com/486627)). - **Carbon storage in buildings methodology** — draft repeatedly delayed past its H2-2025 target. - **Net-Zero Industry Act** — still does not list CDR as a strategic net-zero technology; Article 18 obliges oil and gas producers toward a **50 Mt/yr CO₂ injection capacity target by 2030**. - **ESRS / CSRD** — E1 requires separate disclosure of removals vs reductions and disaggregation by biological vs technological sink; relevant to corporate demand formation. - **LULUCF / Nature Restoration Law** — 310 MtCO₂e land-sink target for 2030; the NRL's cropland soil-carbon provision was deleted in amendment. --- ## Member states are ahead of Brussels on actual purchasing This is the part most often missed: the EU has not yet bought a tonne, but three member states have. - **Sweden** — the flagship. A **SEK 36bn (~$3.5bn) BECCS reverse auction** running 2026–2046, state-aid approved July 2024. First auction won by **Stockholm Exergi: ~SEK 20bn over 15 years for ~800 kt/yr** ([Energimyndigheten](https://www.energimyndigheten.se/en/news/2025/20-billion-to-capture-and-store-over-11-million-tons-of-biogenic-carbon-dioxide/)). A second auction was planned. Two frictions: Danube CCS Ventures filed an **appeal against the Commission's state-aid approval** (OJ, 22 April 2025 — non-suspensive), and the Energy Agency's **rejection of Heidelberg Materials' Slite application in November 2025 caused that 1.8 Mt/yr project to be paused**. Additional Industriklivet calls: ~SEK 300m for bio-CCS (excluding CCU and biochar) and SEK 15m for research (deadline 31 Aug 2026). - **Denmark** — the **CCS Fund (DKK 28.7bn)** concluded its first tender in **June 2026**: Aalborg Portland's ACCSION, up to **DKK 16.5bn for ~1.25 Mt CO₂/yr from 2030**. The earlier **NECCS Fund (DKK 2.5bn, removals-only)** contracted BioCirc CO₂, Bioman and Carbon Capture Scotland for **160,350 t/yr, 2026–2032**. Note the corpus's own caveat that whether the remaining ~DKK 12bn goes to a second tender is unverified. - **Germany** — the **KSpTG** (CO₂ Storage and Transport Act) entered into force **28 November 2025**, enabling commercial offshore storage, an onshore state opt-in, and CO₂ pipelines in the overriding public interest. The **2026 federal budget** carries Germany's **first dedicated negative-emissions line**: >€111m (≈€98m project funding, **€11.5m for direct government purchase of CO₂-removal certificates**, ~€2m admin), plus **€320m in commitment authorisations through 2033** — though the corpus flags these figures as verified only against the July 2025 draft, not the enacted text. The 2025 coalition treaty allows foreign permanent removals for up to 3 percentage points of the 2040 target. --- ## What I'd flag as genuinely uncertain 1. **No enforceable EU-level demand exists today.** CRCF creates units; nothing yet obliges anyone to buy them. The Buyers' Club is voluntary and Article 9c doesn't bite until 2031. 2. **The 250 Mt figure is soft.** Allowance release is not conditional on delivery, and the funding arithmetic only works if removal costs converge toward EUA prices — which they are nowhere near. 3. **Biochar sits in an awkward gap**: CRCF-certifiable but ETS-procurement-excluded. Expect lobbying pressure here. 4. **3% vs 5% international credits** is unresolved in my sources. 5. **Freshness**: policy events end 17 August 2026; the CRCF review with national targets, the Buyers' Club's first purchases, and Parliament/Council positions on COM(2026) 616 were all due around or after that line. Given I could not run web searches, I'd recommend checking the [Carbon Gap policy tracker](https://tracker.carbongap.org/regional-analyses/) and the Commission's carbon removals page directly for anything after September 2026. 6. Many of the instrument records above are machine-extracted and auto-published (review status "auto"), often from secondary trade press (Carbon Pulse, Carbon Herald) rather than primary legal texts. The CRCF, ETS Article 9c, Innovation Fund and Horizon Europe entries are owner-reviewed; most of the smaller proposals are not.
Asked, verbatim: “What has changed in carbon removal policy since March 2026, globally?”
Judge’s note — The answer nails one of three gold items — and nails it well. The 17 July 2026 EU ETS review proposal is covered in genuine depth (250 Mt cap increase, CRCF-certified DACCS/BioCCS-only eligibility, third-country storage from 2031, international-credit snap-back to a 2.7% LRF from 2036), and the flagged tension between Carbon Gap's 'indirect public procurement' framing and HSF Kramer's 'direct integration' framing is exactly the sophisticated read an expert would want; it correctly identifies EUA-fungibility as the unresolved crux. It also correctly frames this as a Commission proposal, not law. But the two other checklist items are absent. Canada appears only as a bullet in the 'could not verify' list — no mention of the April 2026 CCUS ITC amendment on EOR eligibility. On the UK, the answer covers the pre-March-2026 architecture (2029 ETS integration, one-for-one replacement, ex-post issuance, 200-year permanence, CfD-based business model) but explicitly concedes it could not determine whether the GGR Standard or business-model decisions had advanced — missing the July 2026 government response, which is the single most important UK development in the window. The only in-window UK item surfaced is a supply projection, correctly labelled as not a policy change. Beyond-corpus yield is thin: the ~August 2026 Treasury §45Q safe-harbor expansion (headline-level only, with the number and provision unretrieved) and some August 2026 legal-analysis detail on the ETS proposal. The thinly-tracked jurisdictions the probe targets — Canada, Japan's GX-ETS, Brazil, CORSIA, California, COP31/Article 6.4, ICVCM, SBTi CNZS v2 — are enumerated as gaps rather than searched. Naming them shows the right instincts, but a list of unexamined topics is not a finding. Citations are consistently attributed to named, checkable sources with dates, and the answer appropriately warns against citing the 45Q claim without the underlying Treasury document. Coverage honesty is the strongest dimension: the search-quota limitation is disclosed up front, unverified claims are labelled, the H.R.9975 bill is realistically discounted as a two-cosponsor agenda-setting marker with near-zero enactment odds, and the State of CDR 3rd edition headline numbers are withheld rather than confabulated. Structure and prose are excellent and decision-useful. Net: an unusually well-calibrated and honest answer whose verified substance covers only one third of the required ground. Exemplary epistemics cannot substitute for the missing Canadian and UK developments, so this lands at middling overall quality for an expert practitioner.
I'll search for recent developments across the major policy jurisdictions. ## Short answer Assuming "today" is late August/early September 2026: the single biggest change since March 2026 is that the **EU moved carbon removal from a certification question to a compliance-market question**, with the European Commission's 17 July 2026 EU ETS review proposal. Elsewhere, the picture is mostly continuity — the UK is executing decisions taken in 2025, and US federal policy has seen administrative tinkering rather than new statute. Below is what I could verify, followed by an explicit list of what I could not. ⚠️ **Method note up front:** my live search quota was exhausted partway through this research. I was able to verify EU developments in reasonable depth, but the UK, US, and Asia-Pacific sections are thinner than they should be, and I have essentially **no verified coverage** of several jurisdictions. I flag those gaps rather than filling them from memory. --- ## 1. EU — the main event **a) CRCF methodologies became operational.** The EU's Carbon Removals and Carbon Farming Certification Framework moved from framework to implementation over the spring. ESG Today reported the EU adopting its first standard for carbon removal projects (4 Feb 2026), and AlliedOffsets (9 Feb 2026) described the Commission outlining first methodologies for permanent removals covering DACCS, BECCS and biochar. Norwegian firm Thommessen describes the new certification framework for permanent removals as **taking effect in May 2026** — so the first certification cycle is only a few months old as of now. **b) The 17 July 2026 ETS review proposal.** This is the structural change. Per the International Carbon Action Partnership (17 July 2026, updated 23 July), the Commission tabled its proposal to revise the EU ETS, setting the legal framework for **Phase 5 (2031–2040)** and aligning it with the EU's **90% net reduction by 2040** target. Key removal-relevant elements I could verify: - **Volume: 250 Mt.** ICAP states the proposal incorporates domestic permanent carbon removal units by **increasing the overall ETS cap by 250 million allowances**. Herbert Smith Freehills Kramer (August 2026) describes it as integrating up to 250 Mt of eligible permanent domestic removals with a corresponding cap increase. - **Eligibility is narrow.** HSF Kramer: only **permanent removals (BioCCS and DACCS) certified under the CRCF** qualify, and they remain subject to MRV and liability requirements. Nature-based and shorter-duration removals are not in scope. - **Third-country storage from 2031**, conditional on that country having an ETS linked to the EU ETS (HSF Kramer). - **Separately, international credits:** ICAP notes a facility for purchasing international credits, a Commission report due **by January 2033**, and a snap-back mechanism — if high-integrity credits are unavailable, the linear reduction factor **reverts to 2.7% from 2036**, with set-aside allowances redirected to the Industrial Decarbonisation Bank. OPIS (June 2026) reported DG CLIMA was unambiguous that operators will **not** be able to surrender international credits directly inside the ETS. **c) An important ambiguity you should verify against the legal text.** Sources disagree — or at least differ in emphasis — on the *mechanism*. Carbon Gap's policy tracker characterises the 17 July proposal as an **indirect** pathway: a public authority would purchase DACCS and Bio-CCS removals through a dedicated programme financed by auctioning ETS allowances, *rather than* letting companies use removal credits directly for compliance. HSF Kramer's framing reads closer to direct integration. These are reconcilable (raise the cap by 250 Mt → auction the extra allowances → use the revenue for centralised purchase), but the difference matters enormously for whether a CRCF unit becomes fungible with an EUA and therefore for removal price formation. **I would not assume direct fungibility.** Treat this as unresolved pending the actual Article text. **d) Process reality check.** This is a *Commission proposal*, not law. It now goes to Parliament and Council, and ICAP notes additional details are left to further legislation the Commission intends to bring forward later in 2026. Expect 18–30 months of trilogue, and expect the 250 Mt figure and the eligibility list to be contested. --- ## 2. United Kingdom The UK's architecture was largely settled *before* March 2026, which is the main thing to understand: - ICAP's UK ETS profile records that engineered GGRs are set to integrate into the UK ETS **from 2029**, with legislation finalised by 2028; in the initial phase credits from eligible domestic projects would **replace allowances one-for-one without changing the cap**, be issued **ex-post** after verified sequestration, and meet a **minimum 200-year permanence** requirement. - The GGR Business Model, unveiled in 2025, uses **contracts for difference**, is nominally technology-neutral but in practice **only BECCS and DACCS currently qualify**, with interim BSI methodologies as the quality floor until a full UK GGR Standard is finalised (Carbon Herald). Analysts flagged that CfDs are more complex than direct procurement and may disadvantage smaller developers, and that the EU's certification-led approach may be more flexible. **Since March 2026 specifically**, the only signal I surfaced is a Carbon Pulse headline reporting a study finding UK engineered removal credits **could top 850 million by 2050** — a supply projection, not a policy change. I also saw a UK government tender (January 2026) for nature-based removals to offset FY25/26 air travel, which predates your window. **Gap:** I could not verify whether the GGR CfD allocation round has opened, whether the UK GGR Standard has been published, or the status of the 2029 ETS legislation. Contrast note: the UK's one-for-one, no-cap-change design is materially more conservative than the EU's proposed 250 Mt cap *increase*. --- ## 3. United States No verified change in statute since March 2026. What I found: - **Legislative:** the **Carbon Dioxide Removal Leadership Act of 2026** (H.R.9975, Rep. Paul Tonko, D-NY, with a companion S.5170) was **introduced 27 July 2026** and referred to House Energy and Commerce the same day. It would require the Secretary of Energy to remove CO₂ directly from ambient air **or seawater** — notable for explicitly covering marine CDR. LegiScan classes it as a partisan bill with two Democratic sponsors. **Realistically, a two-cosponsor minority bill referred to committee has near-zero probability of enactment in this Congress**; read it as an agenda-setting marker, not policy. - **Administrative:** Crux reported (roughly three weeks ago, so ~August 2026) that **Treasury expanded the §45Q safe harbor**, framed as positive for carbon capture tax credits. I only have the headline-level claim and could not retrieve the notice number, the specific safe-harbor provision amended, or whether DAC-specific rules changed. Do not cite this without checking the underlying Treasury/IRS document. - **Background (pre-window):** the 2025 One Big Beautiful Bill Act already reshaped 45Q, and Treasury/IRS issued guidance addressing 45Q credit risks in December 2025. **Gaps I could not close:** FY2027 appropriations for DOE's CDR programmes, the status of the Regional DAC Hubs and any further cancellations, the DOE CDR Purchase Pilot Prize, and any EPA Class VI primacy decisions. Given the current administration's posture, I'd treat US federal *demand-side* CDR support as at risk absent evidence to the contrary, but I am not asserting that as verified fact. --- ## 4. Evidence base The **State of Carbon Dioxide Removal report, 3rd edition**, was published in **June 2026** (stateofcdr.org) — the standard independent scientific assessment, and the natural reference point for whether the durable-CDR gap has narrowed. I did not retrieve its headline numbers, so I won't paraphrase them. --- ## 5. What I could *not* verify — treat these as open I want to be blunt that these are unexamined, not "nothing happened": - **COP31** (scheduled Nov 2026) — Article 6.4 removals guidance, PACM methodology approvals. - **Japan's GX-ETS** compliance phase and any removal eligibility rules. - **Canada** — federal CDR procurement, greenhouse gas offset protocols. - **China, India, Brazil** — Brazil's regulated carbon market implementation is the one to watch. - **CORSIA** eligible-unit decisions and any removals treatment. - **SBTi Corporate Net-Zero Standard v2** — the residual-emissions/interim-removals provisions are arguably as consequential for demand as any government policy. - **ICVCM / CCP approvals** for removal methodologies; **California** (SB 308 / CARB implementation). - Whether any **new national procurement programmes** launched (beyond the EU Buyers Club, which AlliedOffsets flagged as a development to watch). --- ## Bottom line Since March 2026, policy momentum has been **concentrated in Brussels and directed at permanent, engineered removals only**. The EU has done two things: made CRCF certification operational (May 2026) and proposed a 250 Mt compliance-market channel for CRCF-certified DACCS/BioCCS in Phase 5 (July 2026). The UK is implementing a narrower, later, more conservative version of the same idea (2029, one-for-one, 200-year permanence). The US has produced a symbolic bill and an incremental tax-guidance tweak. The key unresolved question — and the one I'd prioritise checking next — is **whether the EU proposal creates genuine EUA-fungibility for removal units or merely a state-funded purchase programme**, because the sources I found do not agree.
Judge’s note — All three gold checklist items are surfaced substantively and with correct specifics: the Canadian Spring Economic Update EOR eligibility under the CCUS ITC (28 April 2026, half-rate structure, 95% permanent storage condition), the UK response to the Whitehead GGR review (17 July 2026, business model top-up, UK ETS integration, Article 6 optionality, overseas DACCS), and the EU ETS review proposal creating central procurement of 250 Mt of CRCF-certified permanent removals for 2031-2040. Beyond that, the answer is unusually rich for a corpus-bounded synthesis: CRCF methodology votes and carbon-farming adoptions, Drax permit variation, US appropriations trajectory (FY26 vs House FY2027 $15m proposal), MPRSA marine CDR permits and the FoE challenge, Class VI developments, state-level bills (SB 643 death, MD SB0625, NH HB 1205), Denmark CCS Fund tender, Japan GX-ETS start, Australia integrity bill, and registry/ICVCM actions. The main shortfall is the beyond-corpus probe. The answer openly concedes its search budget was exhausted after one query and that nothing after ~September 2026 is covered; it lists eighteen silent jurisdictions (Germany, Norway, Sweden, Switzerland, China, India, Brazil, etc.) as a known gap rather than actually searching them. Essentially one item (California AB 706 on 26 Aug 2026) and some enriched EU ETS financial detail (€50bn, ~48 Mt/yr by 2040, no prescribed technology split) appear to extend past the ingest boundary. So the probe is acknowledged but not executed. Citations are named and checkable in most places (ICAP, CINEA, Carbon180, E&E News, QCIntel, Fastmarkets, HSF Kramer, Isometric, CDR.fyi), though a few claims are unattributed and the DAC Hub restoration rests on trade press — which the answer itself flags. Coverage honesty is exemplary: explicit confidence grading, a provenance caveat about machine-extracted instruments, and a structured gap list. Structure is clean, chronological within jurisdictions, and directly decision-useful, closing with the two files most likely to have moved. Docked mainly for the unsearched silent jurisdictions and post-ingest window.
I'll start by understanding what the corpus covers, then pull recent policy changes. Now let me check newly added instruments and run web searches for what's beyond the corpus. ## Bottom line Between March 2026 and the corpus cut-off (policy events recorded through **17 August 2026**), the centre of gravity in CDR policy shifted decisively to **Europe**: the EU moved from certifying removals to proposing to *buy* them at scale, while the US story was one of a partially rescued federal programme, aggressive appropriations pressure and a burst of state-level and permitting activity. **Caveat up front:** the Carbon Curve corpus is current only to 17 Aug 2026 for policy events (1 Sept 2026 for the last instrument ingest). I attempted to verify beyond that with web searches, but my search budget was exhausted after one successful query, so **anything after roughly September 2026 is not covered here** — treat the below as a floor, not a complete picture. --- ## 1. EU — the biggest change: removals proposed for the ETS - **17 July 2026:** the Commission published its **EU ETS review proposal**, which for the first time creates a formal role for negative emissions in the bloc's carbon market. It provides for **central purchasing of 250 Mt of CRCF-certified permanent removals (DACCS, BioCCS) over 2031–2040**, with the ETS cap raised by a corresponding 250 million allowances, auctioned by the Commission across that period to fund the purchases (ICAP; Herbert Smith Freehills Kramer). Isometric's read of the Commission's own impact assessment puts the spend at roughly **€50 billion over the period, reaching ~48 Mt/yr by 2040**; CDR.fyi notes the proposal sets a total but **does not prescribe a technology split**, with DACCS around 11% of the modelled portfolios. The file has entered the ordinary legislative procedure — Parliament and Council must each form a position before trilogues (Fastmarkets). **Status as of my last verified information: still a proposal, not law.** - **17 March 2026:** the Parliament's **ENVI committee voted down attempts to reject** the Commission's CRCF removal methodologies (QCIntel) — the key survival moment for the permanent-removals methodologies (BioCCS, DACCS, biochar). - **10 July 2026:** the Commission **adopted three carbon-farming methodologies** under CRCF — agriculture/agroforestry on mineral soils, peatland and organic-soil rewetting, and afforestation — with entry into force following scrutiny of the Delegated Regulation. - **20–21 May 2026:** first **CRCF Days** in Brussels (500+ stakeholders). Certification schemes can now apply for Commission recognition; an **EU Buyers' Club** is emerging as a coordination platform, and the Commission flagged a **CRCF review plus a proposal for national removal targets at end-2026**. - **24 March 2026:** **€2.7 billion for 54 projects** under the Innovation Fund, including carbon-management projects (CINEA). - **8–16 July 2026:** consultation on the **European Ocean Act / Maritime Spatial Planning Directive**, where Carbon Gap pushed for a dedicated EU marine-CDR strategy, MRV and research funding — marine CDR is still governance-shaped rather than instrument-shaped in the EU. ## 2. United Kingdom - **17 July 2026:** the government published its response to the **Independent Review of Greenhouse Gas Removals (Dr Alan Whitehead)**. It reaffirms the **GGR Business Model** (a top-up payment to an agreed price on credits sold, plus coverage of CCUS transport-and-storage network costs), commits to integrating GGRs into the **UK ETS "as soon as practicable"**, routes first geological-storage removals through CCUS cluster sequencing, reserves the right to use Article 6, and says it will consider whether high-integrity **overseas DACCS** should complement domestic deployment — while planning to meet carbon budgets to 2037 with domestic action alone. - **1 June → 17 August 2026:** the Environment Agency moved from "minded to issue" to **approving the environmental permit variation for carbon capture at Drax's Selby bioenergy plant**, after a process running back to the May 2023 application and three consultations. This is the regulatory unlock for the UK's largest BECCS proposition. ## 3. United States — rescue, retrenchment and permitting Federal: - **~April 2026:** after a federal court ruled the **October 2025 mass grant terminations unconstitutional** (suit led by the City of St. Paul), DOE's preserved-awards list to House appropriators **retained funding for both flagship DAC hubs** — Project Cypress (LA) and the South Texas DAC Hub, ~$1.2bn combined. Note the corpus flags this as **medium confidence**: no DOE/OCED primary press release confirming restoration was located, and it rests on trade press (Decarbonfuse, Latitude Media, Carbon Herald). - **Appropriations remain the live risk.** FY26 (H.R. 6938, signed 26 Jan 2026) preserved **$45m for CDR purchasing and MRV, up to $71.5m for CDR R&D, $10m to NOAA for marine CDR**, while reprogramming ~$1.04bn of unobligated DAC Hub money. Then on **20 May 2026** the House Appropriations Committee passed an **FY2027 Energy & Water bill** proposing **$15m for Carbon Dioxide Removal (a 67% cut from FY26)** and reprogramming ~$2.8bn from IIJA accounts including the DAC Hubs and DAC prizes (Carbon180 analysis). Still needs Senate action. - **Marine CDR permitting became contested.** EPA issued a **MPRSA research permit to Carboniferous in March 2026** for biomass sinking in the Orca Basin; on **8 April** Friends of the Earth asked EPA to withdraw or hold a hearing on it (E&E News), and on **16 April** EPA issued its **second-ever MPRSA permit** authorising up to 16 t of crushed sugarcane stalks ~168 miles off Louisiana, with deployment from September 2026 through February 2028. - **10 April 2026:** final **Class VI permit** for the One Carbon Partnership project in Randolph County, Indiana (Vault 44.01 / Cardinal Ethanol) — 450 kt/yr, ~13.5 Mt over 30 years. Texas became the sixth Class VI primacy state in December 2025; Louisiana's Class VI review remains under a governor's moratorium from October 2025. - **New bills (all still proposals):** the **Carbon Dioxide Leadership Act** was reintroduced the week of **1 Aug 2026** by Tonko, Peters, Whitehouse and Coons — DOE procurement of increasing volumes of durable removal with a declining per-tonne price ceiling and a set-aside for newer pathways. A bipartisan **Methane Removal Research and Innovation Act** (Mullin, Salinas, Miller) was introduced **25 June 2026**, authorising $25m/yr through 2030. The **Carbon Dioxide Removal Investment Act** ($250/t PTC, Bennet–Murkowski) remains pending. Separately, EPA has proposed **dismantling the Greenhouse Gas Reporting Program**, which underpins 45Q eligibility and carbon-management measurement. States: - **2 March 2026:** California's **SB 643 CDR purchase programme died** — the Legislature did not override Newsom's 13 October 2025 veto. The bill had passed 37–0 in the Senate and 78–1 in the Assembly; the veto cited fiscal pressure and duplication. - **12 May 2026:** Maryland **SB0625 (Ch. 411)** signed — requires MDE to adopt permitting regulations and standards for **biochar and wood-vault** removal projects by 1 January 2028; effective 1 July 2026. - **May 2026:** New Hampshire **HB 1205** signed, **barring state and county lands from timber carbon sequestration projects**. - **26 August 2026:** California **AB 706 (FOREST Act)** passed the Senate unanimously (biomass power with high-hazard forest residue; relevant to BiCRS/biochar infrastructure), returning to the Assembly. - **Michigan** introduced a three-bill package (HB 6247–6249) to seek Class VI primacy; California opened a new DAC solicitation (**GFO-25-307**, closed 31 July 2026). ## 4. Canada, Denmark, Japan, Australia - **Canada, 28 April 2026:** the Spring Economic Update proposed making **enhanced oil recovery an eligible storage use under the CCUS ITC at half rates** — 30% for DAC capture equipment, 25% other capture, 18.75% transport/storage/use, versus 60/50/37.5% for dedicated geological storage, with a requirement to permanently store ≥95% of CO₂ intended for EOR in a province with an adequate storage regime (Alberta, BC, Saskatchewan expected to qualify). Applies from 28 April 2026, subject to legislation. - **Denmark, June 2026:** the **CCS Fund's first tender concluded** with Aalborg Portland (ACCSION) signing for up to **DKK 16.5bn** to capture and store ~1.25 Mt CO₂/yr from 2030 — out of a DKK 28.7bn programme. - **Japan, 1 April 2026:** the **GX-ETS mandatory phase began**, covering entities emitting ≥100,000 tCO₂/yr; the government-run allowance market opens in autumn 2027 (ICAP). - **Australia:** exposure draft of the **Carbon Credits and Other Legislation Amendment (Integrity and Transparency) Bill 2026** released 30 April, consultation closed 22 May 2026. ## 5. Standards and registries (quasi-policy) - **5 March 2026:** ICVCM approved **Rainbow** under the Core Carbon Principles — one of nine recognised programmes and one of three focused on high-durability CDR. - **13 August 2026:** ICVCM approved **Isometric's Agroforestry Protocol** for the CCP label — the first dedicated agroforestry protocol, and Isometric's ninth ICVCM-approved removal protocol. - **7 May 2026:** **Puro Standard General Rules v4.4** approved. --- ## What I could not verify — read this as the gap list 1. **Post-August 2026 events are absent.** The EU ETS file's progress through Parliament and Council, the promised end-2026 CRCF review and national-targets proposal, the fate of the House FY2027 cuts, and any COP outcome are all beyond the data I have. 2. **Eighteen tracked jurisdictions returned no events for this window** — Austria, Brazil, China, Cuba, Germany, India, Ireland, Kazakhstan, Kenya, Luxembourg, Mexico, Norway, Romania, Saudi Arabia, South Korea, Sweden, Switzerland, Thailand. That is a coverage gap, not evidence that nothing happened; Germany, Norway, Sweden and Switzerland in particular have live carbon-management programmes. 3. **Provenance quality varies.** Most policy instruments here are machine-extracted from a cited source and auto-published without independent verification, and policy events carry no review stamp at all. The DAC Hubs "restoration" is explicitly medium-confidence trade-press sourcing. 4. Subnational and non-OECD coverage is thin by the corpus's own admission. If you want, I can go deeper on any one jurisdiction — the EU ETS trilogue dynamics and the UK ETS integration timeline are the two files most likely to have moved since.
Judge’s note — All three gold checklist items are surfaced substantively and with specifics: Canada's 28 April 2026 Spring Economic Update EOR eligibility under the CCUS ITC (with the half-rate schedule, 95% permanent storage condition, and province qualification), the UK's 17 July 2026 government response to the Whitehead GGR review (top-up payment model, T&S cost coverage, UK ETS integration 'as soon as practicable', Article 6 posture), and the EU's 17 July 2026 ETS review proposal with the 250 Mt CRCF-certified permanent removal procurement over 2031-2040, correctly framed as a proposal still in co-decision. Substantial additional depth beyond the checklist (CRCF methodologies, Innovation Fund, US appropriations and mCDR permitting fights, Denmark CCS Fund tender, Japan GX-ETS Phase 2, Australia ACCU integrity bill, ICVCM/Puro developments). Beyond-corpus findings are present but thin: (1) China's CCER methodology consultations (10 and 14 August 2026) explicitly identified as missing from the corpus — the strongest hit, and notable because China is one of the named silent jurisdictions; (2) the ICAP mechanism detail on the ETS proposal (cap increase of 250m allowances, international credits facility, January 2033 report, LRF reverting to 2.7% from 2036) which materially extends the DB record; (3) the confirmation via CRS/Payne Institute that 45Q is unchanged post-OBBBA — a useful negative finding. It falls short on the probe's core ask: it did not surface any genuine post-14-August-2026 policy event from the web (the AB 706 and Michigan items come from the DB), and it candidly admits it did not sweep September-October 2026 or check the other 17 silent jurisdictions (Brazil, India, Korea, Norway, Sweden, etc.), naming these as search-budget gaps rather than verified absences. Citation quality is excellent: every substantive claim carries a named source (ICAP, DG CLIMA, CINEA, gov.uk, New Civil Engineer, Carbon180, E&E News, Carbon Herald, Quantum Commodity Intelligence, Carbon Pulse, Carbon Brief), DB versus web provenance is tagged per claim, and the DOE hub restoration is explicitly downgraded to 'reported, not confirmed' with the absence of an OCED press release flagged. Coverage honesty is outstanding and the strongest feature: it states the 17 August 2026 event cutoff and 1 September ingest, discloses that 196 of 230 instruments are machine-extracted with no human review, enumerates all 18 silent jurisdictions and warns 'no record, not no activity', names specific unverified topics (COP31/Article 6, CDR Launchpad, Longship, Swedish BECCS auctions), corrects a date error (the SB 643 veto was October 2025), and debunks its own 158-instrument aggregate as an artifact of null-date passthrough. It closes by telling the reader which three commercially material numbers to verify against primary documents. Clarity is high: jurisdiction-organised, dated, DB/web tagged, with a decision-useful caveats section at the end. Deducted from a perfect overall only for the incomplete post-ingest sweep and the limited number of genuinely new web findings relative to what the probe invited.
I'll start by checking what the corpus covers, then pull the policy changes. Now let me pull the policy changes since March 2026. The corpus has 24 policy events and 158 instruments in the window. Now the required web checks — the corpus is silent on 18 tracked jurisdictions and stops at 17 August 2026. ## What I did Called `describe_coverage`, then `get_recent_changes` (policy events since 2026‑03‑01), an `aggregate_records` run over policy instruments, and `get_policy_landscape` for the US. Then ran web searches for the things the corpus said it could not cover — the 18 tracked jurisdictions that returned no events, and anything after the corpus's cut-offs. **Corpus boundaries you should hold in mind throughout:** policy events are recorded only **through 17 August 2026**; the last general ingest was **1 September 2026**. Of 230 policy instruments held, **196 are machine-extracted and auto-published with no human spot-check** (`review_status: auto`), and policy events carry no review stamp at all. Policy coverage is deepest on the US, EU, UK, Canada, the Nordics, Japan and Australia; subnational and non-OECD coverage is thin by the corpus's own admission. The database returned **23 policy events** dated between 1 March and 17 August 2026. Below, database records are marked **[DB]** and web findings **[web]**. --- ## 1. The EU: the biggest single shift **ETS integration proposed, with a hard number.** [DB] On **17 July 2026** the Commission published the EU ETS review proposal, which includes **central purchasing of 250 Mt of CRCF-certified permanent removals (DACCS, BioCCS) over 2031–2040**, entering the ordinary legislative procedure (source: ICAP). [web] ICAP's own write-up adds mechanism detail the corpus record compresses: the proposal <cite>incorporates domestic permanent carbon removal units into the EU ETS by increasing the overall cap by 250 million allowances</cite>, establishes a facility for purchasing international credits with further legislation intended later in 2026, and requires a Commission report to Parliament and Council by January 2033 — with the linear reduction factor reverting to 2.7% from 2036 if high-integrity credits prove unavailable. This is a proposal, not law: co-decision with Parliament and Council is still ahead. **CRCF moved from framework to operating machinery.** [DB] - **17 March 2026** — the European Parliament's ENVI committee voted down attempts to reject the Commission's CDR methodologies (source: Quantum Commodity Intelligence). - **29 May 2026** — recorded as an implementation milestone: first CRCF Days held 20–21 May in Brussels with 500+ stakeholders; certification schemes can now apply for Commission recognition; permanent-removal methodologies (BioCCS, DACCS, biochar) adopted; an **EU Buyers' Club** emerging; and a **CRCF review plus a proposal for national targets due at end‑2026** (source: DG CLIMA). - **10 July 2026** — Commission adopted **three carbon-farming methodologies**: agriculture/agroforestry on mineral soils, peatland and organic-soil rewetting, and afforestation. Entry into force expected after scrutiny (source: DG CLIMA). **Money.** [DB] **24 March 2026** — €2.7bn for 54 clean-industry projects under the Innovation Fund, including carbon-management projects (source: CINEA). **Marine CDR governance is being contested at consultation stage.** [DB] **8 July 2026** — Carbon Gap filed its response to the Commission's **European Ocean Act / Maritime Spatial Planning Directive** consultation, asking for an EU mCDR strategy, stronger ocean observation and MRV, and dedicated research funding; the consultation closed 16 July. ## 2. United Kingdom [DB] **17 July 2026** — the government response to the **Independent Review of Greenhouse Gas Removals (Dr Alan Whitehead)** reaffirmed the **GGR Business Model**: a top-up payment to an agreed price for credits sold, plus coverage of CCUS transport and storage network costs. It commits to integrating GGRs into the **UK ETS "as soon as practicable"**, with the first geological-storage removals arriving via CCUS cluster sequencing. It reserves the right to use Article 6 cooperation and says it will consider whether high-integrity overseas DACCS should complement domestic deployment, while planning to meet carbon budgets **through 2037 with domestic action alone** (source: gov.uk). [DB] **Drax BECCS permitting completed**: the Environment Agency was "minded to issue" on **1 June 2026** and **approved the environmental permit variation on 17 August 2026** for carbon capture at the Selby bioenergy plant, after an application first lodged in May 2023, a revised air-emissions risk assessment, a February 2025 consultation and a final summer‑2026 consultation. CO₂ is to be stored under the North Sea (source: New Civil Engineer). ## 3. United States — restoration at federal level, fragmentation at state level [DB] **1 April 2026 (month precision)** — after a federal court ruled the October 2025 mass grant terminations unconstitutional (suit led by the City of St. Paul), **DOE's preserved-awards list sent to House Appropriations retained funding for both flagship DAC hubs** (Project Cypress, Louisiana; South Texas DAC Hub) — roughly $1.2bn combined. The corpus flags this record as **medium confidence** and notes **no DOE/OCED primary press release confirming restoration was located**; it rests on trade press (Decarbonfuse, Latitude Media, Carbon Herald). Treat it as reported, not confirmed by the agency. Also in the window, from instrument records rather than events [DB]: - **20 May 2026** — House Appropriations passed the **FY2027 Energy & Water bill**, proposing to reprogram ~$2.8bn from IIJA accounts and cutting the Carbon Dioxide Removal line to **$15m (a 67% cut from FY2026)**, with DAC Hubs and DAC prizes among the jeopardised programs (source: Carbon180). Still needs Senate passage. - **~1 August 2026** — the **Carbon Dioxide Leadership Act** was reintroduced (Tonko, Peters, Whitehouse, Coons): federal procurement of an increasing volume of durable removals by DOE, with a declining per-tonne price ceiling and a set-aside for newer pathways. - **25 June 2026** — bipartisan **Methane Removal Research and Innovation Act** introduced ($25m/yr through 2030). - For context, the enacted **FY2026 package (H.R. 6938, signed 26 January 2026)** preserved **$45m for the CDR Purchase Pilot Prize**, up to $71.5m for CDR R&D, and **$10m to NOAA for marine CDR**. **Marine CDR permitting became a live legal fight.** [DB] EPA issued its **second-ever MPRSA research permit** to **Carboniferous** (March 2026, with a further permit event recorded **16 April 2026**) for sinking up to 16 t of crushed sugarcane stalks into a low-oxygen zone ~168 miles off Louisiana, deployment from September 2026 to February 2028. On **8 April 2026**, **Friends of the Earth** asked EPA to withdraw, modify, or hold a public hearing on it, citing ecological damage from biomass dumping (source: E&E News). **Storage permitting continued.** [DB] **10 April 2026** — EPA issued a final **Class VI permit** for the One Carbon Partnership project in Randolph County, Indiana (Vault 44.01 with Cardinal Ethanol), up to 450,000 t/yr and ~13.5 Mt over 30 years. Six states now hold Class VI primacy (North Dakota, Wyoming, Louisiana, West Virginia, Arizona, Texas). **States pulled in both directions.** [DB] - **2 March 2026** — **California SB 643**'s veto was formally sustained; the CDR purchase program is dead for the session. (Note the corpus correction: the veto itself was 13 October 2025.) - **12 May 2026** — Maryland **SB0625** signed, requiring MDE to adopt permitting regulations and standards for **biochar and wood-vault** projects by 1 January 2028 (effective 1 July 2026). - **May 2026** — New Hampshire **HB 1205** signed, **prohibiting** state and county lands from enrolling in timber carbon sequestration projects. - **26 August 2026** — California **AB 706** (FOREST Act) passed the Senate unanimously, incentivising forest-residue biomass power with adjacency to biochar/BiCRS. - **1 September 2026 ingest** — Michigan House Bills 6247/6248/6249 introduced to begin transferring Class VI oversight to state regulators. [web] I found no evidence of a change to **45Q** since March 2026; the operative values remain the post‑OBBBA $180/t for DAC and $85/t point-source with foreign-entity-of-concern restrictions, per Congressional Research Service and Payne Institute summaries of the 2025 law. ## 4. Canada, Denmark, Australia, Japan - [DB] **28 April 2026** — Canada's **Spring Economic Update** proposed making **enhanced oil recovery an eligible storage use under the CCUS ITC at half rates**: 30% for DAC capture equipment, 25% for other capture, 18.75% for transport/storage/use, versus 60%/50%/37.5% for dedicated geological storage. EOR projects must permanently store **at least 95%** of CO₂ intended for EOR, in a province with an adequate storage regime (Alberta, BC, Saskatchewan expected to qualify). Applies from 28 April 2026 subject to legislation; rates halve again for 2036–2040. - [DB] **June 2026** — Denmark's **CCS Fund** concluded its first tender: **Aalborg Portland (ACCSION)** signed for up to **DKK 16.5bn** to capture and store ~1.25 Mt CO₂/yr from 2030 (source: Carbon Herald). - [DB] Australia's **ACCU Scheme**: exposure draft of the **Carbon Credits and Other Legislation Amendment (Integrity and Transparency) Bill 2026** released **30 April**, consultation **closed 22 May 2026**; government still considering feedback. - [DB] **1 April 2026** — Japan's **GX-ETS mandatory phase (Phase 2)** began, covering entities emitting 100,000+ tCO₂/yr on FY2023–25 average emissions. The allowance trading market run by the GX Acceleration Agency is scheduled to open **autumn 2027**. ## 5. Standards and registries (not government, but market-shaping) [DB] **5 March 2026** — **Rainbow** approved by ICVCM as a CCP-eligible programme, one of three focused on high-durability CDR. **7 May 2026** — **Puro Standard General Rules v4.4**. **13 August 2026** — ICVCM approved **Isometric's Agroforestry Protocol** for the CCP label, the first dedicated agroforestry protocol to get it and Isometric's ninth ICVCM-approved removal protocol. ## 6. What the web added that the corpus does not hold - [web] **China** — outside the corpus's event record entirely for this window. Carbon Pulse reported on **14 August 2026** that China opened public feedback on three new CCER methodologies covering **forestry and geothermal heating**, following a **10 August 2026** consultation on methodologies for green fuels. I could not verify the detail behind the paywall. - [web] **Assessment literature** — Carbon Brief published a Q&A on **2 June 2026** covering the **State of Carbon Dioxide Removal, 3rd edition** (Edwards et al., 2026), which argues CDR is <cite>needed alongside deep and rapid emissions reductions</cite> for 1.5C and highlights the need for foundational governance policies covering quantification, community engagement and environmental impacts. Not a policy instrument, but it is the reference document most cited in current policy debate. ## 7. What neither source covers — read this before acting - **Nothing after 17 August 2026** is in the policy-event layer, and nothing after ~1 September 2026 in any layer. My web searches surfaced material into September–October 2026 but did not systematically sweep that window; the EU ETS proposal in particular will have moved in Parliament since. - **Eighteen tracked jurisdictions returned zero events** for this window: Austria, Brazil, China, Cuba, Germany, India, Ireland, Kazakhstan, Kenya, Luxembourg, Mexico, Norway, Romania, Saudi Arabia, South Korea, Sweden, Switzerland, Thailand. That is **no record, not no activity** — I confirmed China had activity the corpus missed, which suggests the same is likely for others (Brazil's regulated market and Japan/Korea ETS linkage discussions especially). - I found **no corpus record and no confirmation** of: COP31/Article 6 developments on removals, the CDR Launchpad's 2026 activity, Norway's Longship/Northern Lights policy changes, or Sweden's BECCS reverse auction contract performance in this window. Absence here reflects my search budget, not verified absence. - The **158 policy instruments** the aggregate returned for a March–December 2026 date filter is **not** a count of new instruments: the tool passes instruments with no effective date, and most buckets returned null date ranges. Do not read it as "158 new policies." - Almost everything above is machine-extracted from a single cited source. Where a claim matters commercially — the Canadian EOR rates, the Danish DKK 16.5bn, the DOE hub restoration — go to the primary document.
Judge’s note — All three gold checklist items are surfaced substantively and with precision: the Canadian Spring Economic Update (28 Apr 2026) making EOR an eligible CCUS ITC use at half rates with the 95% permanence condition; the UK government response to the Whitehead GGR review (17 Jul 2026) retaining the GGR business model and committing to UK ETS integration; and the EU ETS review proposal COM(2026) 616 (17 Jul 2026) with the 250 Mt Article 9c allowance pool, central procurement of CRCF-certified DACCS/BioCCS, ~48 Mt/yr by 2040, biochar exclusion, and the contested-scale caveats. Citation quality is very strong — named institutional sources with URLs (gov.uk, budget.canada.ca, ICAP, CINEA, EC Climate, Carbon180, leginfo, mgaleg), and the answer distinguishes primary from trade-press sourcing. Coverage honesty is exemplary: it names the 18 silent jurisdictions explicitly, warns that silence is a coverage gap not inactivity, flags the auto-published/unverified share of records, downgrades the DAC Hubs restoration claim to medium confidence for lack of DOE primary confirmation, notes a date conflict on the Carboniferous MPRSA permit, and states clearly that nothing after 17 Aug 2026 is in scope and that Article 6/multilateral and most of Asia/LatAm/Africa are unaddressed. Structure and decision-usefulness are high, with a crisp thesis (certification -> compliance demand), jurisdictional sections, and an explicit uncertainty section. The main shortfall is the beyond-corpus probe: the answer states no web search tool was available and therefore did not verify or fill the named silent jurisdictions (Sweden BECCS auctions, Norway Longship, China/India/Korea, Article 6.4 removals methodology), so it remains a corpus census rather than an updated global picture. Only a handful of items extend past the 14 Aug ingest boundary (Drax permit 17 Aug, CA AB 706 Senate passage 26 Aug, the late-Aug CRCF biochar consultation, Michigan Class VI primacy bills recorded 1 Sept), and these appear to come from the corpus's later publication lane rather than independent search.
I'll start by checking what the corpus covers and what it has recorded recently.
## Method note first
Everything below comes from the Carbon Curve corpus (230 tracked policy instruments, 134 policy events across 28 jurisdictions). Its hard limits matter for this question: **policy events are recorded only through 2026‑08‑17**, and the last ingest was 2026‑09‑01. The tool explicitly instructs me to verify beyond it with web searches — **no web search tool is exposed in this session, so I could not do that verification**. Treat this as a well‑sourced floor, not a census. Also note 196 of 230 instrument records are machine‑extracted and auto‑published from a cited source ("review_status: auto"), not independently verified.
Eighteen tracked jurisdictions returned **zero** events for this window — Austria, Brazil, China, Cuba, Germany, India, Ireland, Kazakhstan, Kenya, Luxembourg, Mexico, Norway, Romania, Saudi Arabia, South Korea, Sweden, Switzerland, Thailand. That is a coverage gap, not evidence of inactivity (Sweden's BECCS reverse auctions and Norway's Longship, for instance, are simply not eventful in this database in this period).
---
## The one‑line summary
Since March 2026 the centre of gravity in CDR policy shifted from **certification** to **compliance demand** — and that shift is almost entirely European. The EU tabled the first large‑scale state purchase mechanism for durable removals; the UK reaffirmed its GGR business model and pointed at ETS integration; meanwhile US federal policy moved sideways at best, with a court‑forced restoration of DAC Hub funding offset by proposed FY2027 cuts, and California's purchase programme died for good.
---
## 1. EU — the substantive change
**17 July 2026: the ETS review proposal (COM(2026) 616)** is the single biggest development. The Commission proposes a new Article 9c: add **250 million allowances** to the EU‑wide cap over **2031–2040** (contingency to 260 Mt), auction them, and use the "external assigned revenue" to **centrally purchase CRCF‑certified permanent removals from DACCS and BioCCS**, scaling to roughly **48 Mt/yr by 2040**, with procurement by competitive tender/reverse auction and ex‑post payment on delivery. **Biochar is excluded** on evidentiary grounds (limited long‑term field data), with a pathway for later inclusion. Cost estimates vary widely by model — €29–43bn (PRIMES) to €46–65bn (POTEnCIA); Fastmarkets frames it as a "€50bn alley‑oop" to BECCS in pulp & paper. ETS‑obligated operators (e.g. shipping, aviation) could use their own CRCF‑certified removals against fossil emissions but not "go beyond zero," and such use is deducted from the 250 Mt.
Sources: [ICAP](https://icapcarbonaction.com/en/news/eu-commission-publishes-eu-ets-review-proposal); [Carbon Gap reaction](https://carbongap.org/insights/reaction-to-the-legislative-proposal-on-carbon-removals-in-the-eu-emissions-trading-system); [CDR.fyi explainer](https://www.cdr.fyi/blog/the-eu-ets-proposal-explained); [Carbon Brief Q&A](https://www.carbonbrief.org/qa-what-the-eus-carbon-market-review-means-for-climate-action).
Two caveats worth carrying: CDR.fyi's analysis stresses **250 Mt is a scale ceiling, not a financed purchase commitment** — allowance release is not legally conditional on removals being contracted or delivered. And it is contested: WWF and CAN Europe argue the package allows ~2bn tonnes of extra emissions, with Sven Harmeling saying removals integration would "weaken the ETS impact, undermine the carbon price and create new loopholes."
**Other EU movement, chronologically:**
- **5 Mar 2026** – ICVCM approves Rainbow Standard as CCP‑Eligible (one of three ICVCM‑recognised durable‑CDR programmes; ~60 projects, 11,000+ credits issued). [rainbowstandard.io](https://rainbowstandard.io/news/rainbow-officially-approved-by-the-icvcm)
- **17 Mar 2026** – European Parliament ENVI committee **votes down attempts to reject** the Commission's CDR methodologies, clearing the CRCF permanent‑removals delegated act (Reg. (EU) 2026/285, adopted Feb 2026, covering DACCS, BioCCS, biochar). [QCIntel](https://www.qcintel.com/carbon/article/meps-advance-eu-cdr-methodologies-with-key-vote-60942.html)
- **24 Mar 2026** – Innovation Fund awards **€2.7bn to 54 clean‑industry projects**, including carbon‑management projects. [CINEA](https://cinea.ec.europa.eu/news-events/news/commission-unlocks-eu27-billion-54-clean-industry-projects-under-innovation-fund-2026-03-24_en)
- **20–21 May 2026** – First **CRCF Days** (500+ stakeholders). Certification schemes can now apply for Commission recognition; the **EU Buyers' Club** is stood up as a hybrid public‑private demand aggregator targeting first permanent‑removal purchases by **December 2026**; a CRCF review **with a proposal for national targets** is promised for end‑2026. [EC Climate](https://climate.ec.europa.eu/news-other-reads/news/europe-lays-foundations-robust-carbon-removals-and-carbon-farming-market-2026-05-29_en)
- **8–16 Jul 2026** – **European Ocean Act** consultation; Carbon Gap asks for a coherent EU mCDR strategy, ocean observation/MRV, and dedicated research funding. [Carbon Gap](https://carbongap.org/insights/feedback-to-the-public-consultation-on-the-european-ocean-act)
- **10 Jul 2026** – Commission adopts **three carbon‑farming methodologies**: agriculture/agroforestry on mineral soils, peatland and organic‑soil rewetting, and afforestation. Entry into force follows the scrutiny period. [EC Climate](https://climate.ec.europa.eu/news-other-reads/news/commission-adopts-certification-methodologies-carbon-farming-under-crcf-regulation-2026-07-10_en)
- **Late Aug 2026** – Carbon Standards opens a 30‑day consultation on a CRCF‑compliant **EU Biochar Carbon Removal Standard**, feedback due 24 Sept 2026; Puro.earth has formally applied for CRCF scheme recognition and amended its General Rules (v4.3) to align. [Carbon Herald](https://carbonherald.com/public-consultation-now-open-for-new-crcf-compliant-biochar-certification-methodology); [Puro.earth](https://puro.earth/insights/post/getting-ready-for-the-eu-crcf-a-conversation-with-boris-lagadinov-head-of-crcf-program-puro-earth)
- Also queued: a **green‑claims chapter** in the forthcoming CRCF revision, and exploratory work (expert workshop, Feb 2026) toward certifying **direct ocean capture (DOCCS)**.
---
## 2. UK — the business model reaffirmed, Drax permitted
- **17 Jul 2026** – Government response to the **Whitehead Independent Review of Greenhouse Gas Removals**: keeps the GGR Business Model (top‑up payment to an agreed price on credits sold, plus coverage of CCUS T&S network costs); commits to integrating GGRs into the **UK ETS "as soon as practicable"**; first geological‑storage removals to come via CCUS cluster sequencing; reserves the right to use Article 6 and will consider whether high‑integrity **overseas DACCS** should complement domestic deployment — while planning to meet carbon budgets to **2037 with domestic action alone**. [gov.uk](https://www.gov.uk/government/publications/independent-review-on-greenhouse-gas-removals-government-response/independent-review-on-greenhouse-gas-removals-government-response)
- **1 Jun → 17 Aug 2026** – Environment Agency moves from "minded to" to **granting the permit variation** for carbon capture at **Drax** (Selby, N. Yorkshire), with CO₂ destined for North Sea storage. Application dates to May 2023 and required a revised air‑emissions risk assessment. [New Civil Engineer](https://www.newcivilengineer.com/latest/drax-bioenergy-plant-granted-permit-to-use-carbon-capture-to-reduce-emissions-17-08-2026)
- Advocacy pressure in the same window: CO2RE's three‑part series on UK DAC policy sequencing and an **open letter to DESNZ** urging a Carbon Takeback Obligation as a licence condition on Jackdaw and Rosebank. [CO2RE](https://co2re.org/open-letter-desnz-carbon-storage)
---
## 3. United States — restoration, retrenchment, and a state‑level scatter
**Negative / risk:**
- **2 Mar 2026** – **California SB 643 is dead**: the legislature did not override Newsom's 13 Oct 2025 veto, terminating what would have been a ~$50m state CDR purchase programme. Newsom cited budget pressure and duplication, pointing at SB 840's $85m climate‑innovation allocation. [leginfo](https://leginfo.legislature.ca.gov/faces/billHistoryClient.xhtml?bill_id=202520260SB643)
- **20 May 2026** – House **FY2027 Energy & Water** bill clears committee proposing to cut the **Carbon Dioxide Removal line to $15m (−67% vs FY26)**, Carbon Utilization −30%, Transport & Storage −15%, Point Source −33%, and to reprogramme ~**$2.8bn** of IIJA funds — jeopardising DAC Hubs (IIJA 40308), CarbonSAFE, CO₂ transport FEED studies and DAC prizes, with CIFIA effectively eliminated. Not law: still needs Senate and signature. [Carbon180](https://carbon180.org/blog/what-the-house-fy2027-energy-and-water-development-budget-means-for-carbon-removal)
- EPA's proposed **GHGRP rollback** (relieving 46 of 47 source categories of reporting) remains pending; the corpus flags it as underpinning DAC/BECCS measurement and 45Q eligibility.
**Positive / neutral:**
- **~1 Apr 2026** – After a federal court ruled the October 2025 mass grant terminations unconstitutional (suit led by the City of St. Paul), DOE's **preserved‑awards list to House appropriators retained both flagship DAC hubs** — Project Cypress (LA) and South Texas — roughly **$1.2bn combined**. The corpus rates this **medium confidence**: no DOE/OCED primary press release confirming restoration was located; it rests on trade press (Decarbonfuse, Latitude Media, Carbon Herald). Per‑hub figures are inconsistently reported. [Decarbonfuse](https://decarbonfuse.com/posts/doe-restores-1-2b-for-dac-hubs-and-5-hydrogen-projects)
- **10 Apr 2026** – EPA issues a final **Class VI permit** for the One Carbon Partnership project (Vault 44.01 / Cardinal Ethanol) in Randolph County, Indiana — 450,000 t/yr, ~13.5 Mt over 30 years; Indiana's first. [Carbon Herald](https://carbonherald.com/first-indiana-carbon-capture-and-storage-project-approved)
- **Marine CDR permitting precedent, contested.** EPA issued only its **second‑ever MPRSA research permit** to **Carboniferous** for biomass sinking in the Orca Basin off Louisiana (up to 16 t of crushed sugarcane; deployment Sept 2026–Feb 2028). Note a date conflict in the records: the instrument entry and the E&E News item date the permit to **March 2026**, while the Carbon Herald item is dated **16 April 2026** — likely the same permit reported twice. On **8 Apr 2026**, **Friends of the Earth** asked EPA to withdraw, modify, or hold a hearing on it, citing ecological damage from biomass dumping. [Carbon Herald](https://carbonherald.com/the-us-epa-issues-its-second-ever-marine-protection-research-and-sanctuaries-act-permit-for-an-mcdr-pilot-in-louisiana); [E&E News](https://www.eenews.net/articles/environmental-group-asks-epa-to-block-climate-test)
**New federal bills (all proposed, none enacted):**
- **~1 Aug 2026** – **Carbon Dioxide Leadership Act** reintroduced by Tonko, Peters, Whitehouse and Coons: DOE procurement of increasing volumes of DAC/durable removal, MMRV standards, a **declining per‑ton price ceiling**, and a set‑aside for newer pathways. [Troy Record](https://www.troyrecord.com/2026/08/01/tonko-others-reintroduce-carbon-dioxide-removal-bill)
- **25 Jun 2026** – **Methane Removal Research and Innovation Act** (Mullin, Salinas, Miller): DOE programme, **$25m/yr through 2030**.
- **RESCUE Oceans Act** re‑promoted (Murkowski, Schatz, Bonamici, Carter) — NOAA‑led mCDR research grants, field‑study site designation, code of conduct, NOAA–DOE interagency working group. [Senate one‑pager](https://www.murkowski.senate.gov/imo/media/doc/one-page_summary_of_the_rescue_oceans_act.pdf)
**States:**
- **Maryland SB0625 (Ch. 411)** – signed 12 May 2026, effective 1 Jul 2026: MDE must adopt permitting **regulations for biochar and wood‑vault projects by 1 Jan 2028**. First‑of‑kind state rulemaking for those pathways. [mgaleg](https://mgaleg.maryland.gov/mgawebsite/legislation/details/sb0625)
- **New Hampshire HB 1205** – signed by Gov. Ayotte, May 2026: **prohibits** state and county lands from enrolling in timber carbon sequestration projects. [NHPR](https://www.nhpr.org/nh-news/2026-05-12/new-law-prohibits-state-county-lands-from-carbon-sequestration-projects)
- **California AB 706 (FOREST Act)** – passed Senate unanimously 26 Aug 2026, back to Assembly for concurrence; biomass power incentives from 1 Jan 2027 with ≥60% Tier 1/2 high‑hazard‑fire feedstock, framed as preserving infrastructure adaptable to biochar/BiCRS. Also **GFO‑25‑307** (CEC DAC pre‑commercial demonstration) opened 1 May, closed 31 Jul 2026.
- **Michigan HB 6247/6248/6249** – introduced (recorded 1 Sept 2026) to transfer Class VI primacy from EPA to EGLE.
---
## 4. Canada
- **28 Apr 2026** – Spring Economic Update proposes making **enhanced oil recovery an eligible storage use under the CCUS ITC at half rates**: 30% DAC capture, 25% other capture, 18.75% transport/storage/use, versus 60/50/37.5% for dedicated geological storage. Requires ≥95% permanent storage of CO₂ intended for EOR in a province with an adequate storage regime (AB, BC, SK expected to qualify). Applies from 28 Apr 2026 subject to legislation; rates halve again 2036–2040. [Budget.gc.ca PDF](https://budget.canada.ca/update-miseajour/2026/report-rapport/pdf/update-miseajour2026-eng.pdf)
- **15 May 2026** – **Canada–Alberta MOU Implementation Agreement** signed, rewriting TIER: headline carbon price to **CAD 130 by 2035**, a price floor from 2030, a Direct Investment Program and **Carbon Contracts for Difference**, with the federal benchmark to be aligned. This sits in the corpus's publication lane (ClearBlue Markets webinar record), not as a typed policy instrument — so treat the specifics as secondary‑sourced. [ClearBlue](https://www.clearbluemarkets.com/public-webinar-canada-alberta-mou-may-2026)
---
## 5. Asia‑Pacific and Nordics
- **Japan: 1 Apr 2026** – GX‑ETS **mandatory Phase 2 began**, covering entities emitting ≥100,000 tCO₂/yr (FY2023‑25 average). The government‑run allowance trading market under the GX Acceleration Agency is scheduled for **autumn 2027**. This is the most consequential new compliance market in the window, though the corpus holds no record yet of how removals will be treated within it. [ICAP](https://icapcarbonaction.com/en/ets/japan-gx-ets)
- **Australia: 30 Apr → 22 May 2026** – DCCEEW released and closed consultation on the exposure draft **Carbon Credits and Other Legislation Amendment (Integrity and Transparency) Bill 2026**; government now considering feedback before introduction. [DCCEEW](https://consult.dcceew.gov.au/cc-legislation-amendment-integrity-transparency-bill)
- **Denmark: June 2026** – The **DKK 28.7bn CCS Fund's first tender concluded**, with **Aalborg Portland** (ACCSION) signing for up to **DKK 16.5bn** to capture and store ~**1.25 Mt CO₂/yr from 2030**. Point‑source cement rather than removal, but it is the largest single subsidy award in the window. [Carbon Herald](https://carbonherald.com/denmark-awards-billions-to-aalborg-portland-after-bumpy-carbon-capture-tender/)
---
## 6. Registry and standards layer (quasi‑policy)
- **7 May 2026** – Puro Standard **General Rules v4.4** approved.
- **13 Aug 2026** – ICVCM approves **Isometric's Agroforestry Protocol** for the CCP label — the first dedicated agroforestry protocol to earn it, and Isometric's **ninth** ICVCM‑approved removal protocol; suppliers 3T and FCF India can now carry the label. [Isometric](https://isometric.com/writing-articles/icvcm-approves-isometrics-agroforestry-protocol)
- **2026 (undated)** – CDR.fyi published **OSCAR**, a free open‑standard offtake contract template, and a **CDR Buyer's Guide** — market plumbing rather than policy, but relevant to procurement friction.
---
## What I'd flag as genuinely uncertain
1. **The DAC Hubs "restoration" is the weakest‑sourced major claim here.** No DOE primary confirmation; month‑precision dating; inconsistent per‑hub amounts. If you need this for a decision, verify against DOE/OCED directly and against the House Appropriations record.
2. **The EU 250 Mt is a proposal at the start of ordinary legislative procedure**, not law. It now goes to Council and Parliament; the three integration designs (public purchaser vs. operator integration vs. one‑in‑one‑out) are still live, and the biochar exclusion will be fought.
3. **The corpus is silent on multilateral/Article 6 developments** in this window and on most of Asia, Latin America and Africa. If a COP30 follow‑up, an Article 6.4 removals methodology decision, or a Chinese/Indian/Korean removals policy moved after March 2026, it would not appear here.
4. **Nothing after 17 August 2026 is in scope.** Given the pace of the EU file and the US FY2027 appropriations cycle, that is a material blind spot — please check the ETS review's Council/Parliament progress and the Senate FY2027 mark independently.
Asked, verbatim: “How strong is the field evidence for ocean alkalinity enhancement? How much of the literature is actual in-water measurement vs modeling?”
Judge’s note — The answer is a well-structured, expert-grade synthesis with a sharp and correct central thesis (in-water data verify dispersal and safety; models produce the tonnes), and it is rich in beyond-corpus specifics: LOC-NESS August 2025 (65,000 L NaOH over 1 km², 760 L rhodamine, 3,600 km² tracked domain, Ocean Sciences Meeting presentation), the April 2025 EPA permit process, Vesta's Duck NC 8,200 t berm and its unpublished carbon results, Ebb Carbon/PNNL Sequim and the Microsoft 350,000 t / 1,333 t initial deal under Isometric's protocol, Planetary's Halifax outfall work, and a named NOAA OAP grant. Citations are mostly named and checkable (Tyka 2025 Biogeosciences, Zhou et al. NCC 2024, NASEM 2021, Cyronak et al. 2023, Isometric 2024), though several block quotes lack explicit attribution. Coverage honesty is excellent: it explicitly states no rigorous bibliometric breakdown was located, labels its percentages as estimates rather than citable statistics, and lists three specific unresolved questions. Where it falls down is on the gold corpus facts: it never surfaces the ~384-paper in-scope set or that the pathway query is a complete result set, never gives the 16 field-trial count (only an inferred '~5% or less, on the order of 10–20 papers', which is close enough in substance to credit that item), badly under-weights the lab share relative to modeling (real split is near-parity ~102 vs ~96, answer says ~50% vs ~22%), treats reviews as a residual rather than the ~77-paper block they are, and omits entirely the caveat that ~76 papers are unclassified so the split describes only the classified subset. The estimated breakdown is thus directionally right on the headline (field is a tiny fraction) but structurally wrong on the interior composition, and it is presented as guesswork where hard numbers were available.
I'll research the current state of OAE field trials and the composition of the literature. Let me dig into the specific trials and quantify the literature composition. ## Bottom line Field evidence for ocean alkalinity enhancement (OAE) is **thin but improving rapidly, and it does not yet include a single direct, in-water measurement of net CO₂ removal from any deployment.** What field trials have demonstrated so far is (a) that alkalinity can be dispersed as engineered, (b) that plumes can be tracked with dye and sensors, and (c) an absence of detected acute ecological harm at small scale and short duration. The actual carbon accounting — the tonnes — is in every case produced by a model, calibrated against near-field measurements. That is not a scandal; it is a structural property of the pathway. But it means the "evidence base" and the "carbon claims" rest on different foundations, and buyers and policymakers frequently conflate them. --- ## 1. The open-water field record (as of mid-2026) | Trial | Operator / date | Scale | What was actually measured | |---|---|---|---| | Halifax Harbour | Planetary Technologies + Dalhousie, 2023–24 | Wastewater outfall, magnesium hydroxide/oxide | Near-field chemistry, operational MRV learnings | | Duck, NC | Vesta, July 2024 | 8,200 t olivine sand nearshore berm | Water quality, ecology, sand tracking; multi-year | | PNNL–Sequim, WA | Ebb Carbon, ~2023– | ~100 t CO₂/yr electrochemical system | Local carbonate chemistry, oyster/eelgrass response | | LOC-NESS | WHOI, Aug 2025 | 65,000 L NaOH over 1 km² | Dye-tracked plume, ship + AUV chemistry, biology | The single most informative event is LOC-NESS. Woods Hole Oceanographic Institution announced preliminary findings from the first US EPA-permitted OAE field trial, conducted in federal waters in the Gulf of Maine in August 2025, with results presented at the Ocean Sciences Meeting in Glasgow. In August 2025 the team conducted an engineered six-hour dispersal in the Wilkinson Basin area, introducing highly purified sodium hydroxide — the same grade used to adjust drinking-water pH — into surface waters. The design is worth noting because it defines the detection problem: they dispersed 65,000 litres of sodium hydroxide over 1 km², accompanied by 760 litres of fluorescent rhodamine dye used to monitor dispersal across a 3,600 km² zone — i.e. the tracked domain was ~3,600× the injection footprint. Using remote sensors and autonomous vehicles, researchers found the dispersal lowered acidity in the area and had no significant biological impacts on the organisms there. The permitting record is also part of the evidence base, in a governance sense: EPA approved the LOC-NESS proposal in April 2025 after a year-long review including two public comment periods totalling 75 days and more than 250 comments, with EPA consultations with NOAA Fisheries and the US Fish and Wildlife Service finding minimal potential environmental impact. **What is still missing from the published record:** Vesta's Duck deployment, the largest mass-based intervention, has not yet produced peer-reviewed carbon results. The project requires three years of ecological monitoring, with monitoring led by Hourglass Climate and supported by UNC Greensboro, UNC Wilmington, ECU's Coastal Studies Institute and the USACE Field Research Facility, with results to be published in peer-reviewed journals , and preliminary data from ongoing monitoring are still being analysed . That was the stated position as of the most recent company communications I could find — a two-year-old deployment without published carbon verification is a meaningful gap. --- ## 2. Why in-water measurement of CDR is not merely rare — it is close to impossible This is the crux, and it is well established in the literature rather than a matter of opinion. Long equilibration timescales, the potential for premature subduction of surface water parcels, and extensive horizontal transport and dilution of added alkalinity make direct experimental measurement of induced CO₂ uptake challenging ; the problem of MRV therefore hinges on general circulation models (Tyka, *Biogeosciences* 22, 341–353, 2025). Concretely: an alkalinity anomaly dilutes below the ~1–2 µmol kg⁻¹ detection threshold within days to weeks, while air–sea CO₂ equilibration takes months to years, and is incomplete. Zhou et al. (*Nature Climate Change*, Nov 2024) present global maps of OAE efficiency and assess seasonal variation, noting that incomplete air–sea CO₂ equilibration reduces the efficiency of carbon removal. The measurement window and the carbon-uptake window barely overlap. Consequently, operational MRV is explicitly hybrid. MRV for OAE relies on the combination of direct measurement of the rate and concentration of alkalinity-enhanced seawater returned to the surface ocean, direct measurement of water-quality parameters near the outfall for permit compliance, and ocean modelling to quantify carbon removal and storage. A 2025 tracer study is blunter still: research on open-water OAE deployment and its associated MRV is currently lacking; the only monitoring framework for OAE exists as part of a commercial wastewater outfall MRV protocol (Isometric, 2024), which focuses on in-water measurements not for the measurement of CDR but instead to verify dispersal limits in the near field of the outfall pipe and to calibrate a model which is then used for CDR calculations. That sentence should be read carefully by anyone buying OAE credits: **the in-water measurements verify dispersion and safety; the model produces the tonnes.** Credits are nonetheless being transacted at scale — Ebb Carbon signed an agreement with Microsoft to remove up to 350,000 tonnes of CO₂ over ten years using electrochemical OAE, with credits measured, reported and verified under Isometric's OAE protocol, the first standard established for the pathway , structured as an initial 1,333 tonnes with an option for the further 350,000 tonnes , built on a 100-tonne-per-year system at DOE's Pacific Northwest National Laboratory in Sequim, Washington . --- ## 3. How much of the literature is field measurement vs. modelling? **No rigorous published bibliometric breakdown of the OAE corpus exists that I could locate**, so treat the following as my structured estimate, not a citable statistic. Authoritative qualitative statements do exist and point the same direction. The 2021 US National Academies research strategy concluded that the vast majority of results on OAE impacts on chemical and biological processes come from modelling studies, and there is an urgent need to generate empirical data from laboratory and mesocosm studies and to explore these questions in small field trials . Two years later, the OAE Guide's field-experiment chapter (Cyronak et al., 2023) stated that to date only a limited number of OAE field studies have been conducted, and guidelines for such experiments are still evolving . My estimate of the composition of the peer-reviewed OAE-specific literature (roughly 2019–2026): - **Modelling (global/regional efficiency, carbon accounting, plume/MRV design): ~45–55%.** This is the growth engine of the field and includes the papers that generate all headline efficiency numbers and potential estimates. - **Laboratory and bench work (mineral dissolution kinetics, secondary precipitation thresholds, single-species toxicology): ~20–25%.** - **Mesocosm studies: ~15–20%.** Disproportionately from a small number of GEOMAR campaigns, which each spawn many papers. E.g. a Gran Canaria experiment using nine 8.3 m³ mesocosms with sediment traps, running a CO₂-equilibrated total alkalinity gradient from ~2,400 to ~4,800 µmol L⁻¹ over 33 days ; and a month-long experiment in which most biogeochemical pools — inorganic nutrients, particulate organic carbon and phosphorus, biogenic silica — remained unaltered across all OAE levels, with exceptions of a minor decrease in particulate organic nitrogen and an increase in the C:N ratio of particulate organic matter . Note the caveat that a missing component in all closed-system mesocosm experiments is dilution through mixing with unperturbed waters — mesocosms therefore overstate exposure and cannot test dispersal or air–sea flux realistically. - **Open-water field studies with primary in-water data: ~5% or less.** In absolute terms this is on the order of 10–20 papers, and the number reporting *quantified, measurement-anchored CDR* is effectively **zero**. - **Reviews, guides, governance, social licence, LCA: the remainder.** A useful sanity check on the ratio: even the field-experiment guidance chapter concedes that incorporating physical mixing models with biogeochemical processes will likely be the end goal of many field experiments focused on MRV . Field work in OAE is largely designed to *parameterise and validate models*, not to replace them. One more caution about the lab-heavy portion: transferability is poor for mineral-based OAE. The National Academies flagged discrepancies between field and lab observations, and large uncertainties regarding which factors govern mineral dissolution . Olivine dissolution rates measured in beakers have repeatedly failed to reproduce in sediments, where armouring, secondary mineral coatings and biological reworking intervene. This is the single largest reason to discount lab-derived efficiency claims for coastal enhanced weathering specifically. --- ## 4. What the field evidence does and does not currently support **Reasonably supported by in-water data:** - Alkalinity can be delivered and dispersed in open water to a pre-specified plan, and tracked with dye plus autonomous platforms. - Short-duration, moderate-ΔTA exposure produces no detected acute ecosystem harm at trial scale — from both mesocosms and the LOC-NESS observations. A 2025 study found OAE did not cause cellular stress in a subtropical Atlantic phytoplankton community across a 2,400–4,800 µmol L⁻¹ alkalinity gradient. - Local pH/acidity is measurably raised, which matters for the ocean-acidification co-benefit claim independent of the carbon claim. **Not yet supported by field data:** - Net tonnes of CO₂ removed, at any site, by any operator. - Realised air–sea equilibration efficiency (η) in situ, versus the ~0.6–0.9 typically assumed from models. - Multi-year durability and the fraction of alkalinity lost to secondary carbonate precipitation under real deployment conditions. - Mineral dissolution rates in real coastal sediments over multi-year timescales. - Cumulative or repeated-dosing ecological effects — every trial to date is a single, short pulse. **Open questions I could not resolve in this research:** whether the LOC-NESS team has published a quantified CO₂-uptake estimate with uncertainty bounds (press coverage from spring 2026 indicated carbon uptake was detected, but I did not locate the peer-reviewed quantification); whether Vesta's Duck carbon results have entered peer review; and how many Isometric-verified OAE credits have actually been issued and retired versus contracted. Those three data points would materially change the assessment and are worth tracking directly. --- ## 5. How I'd frame the maturity judgment OAE is roughly where enhanced rock weathering on land was in 2021: physically plausible, chemically well understood at first principles, thermodynamically favourable, and supported by a governance and instrumentation build-out — but with a verification architecture in which models carry nearly all the load and field data serves mainly to bound the models' input parameters and demonstrate safety. Public funding is now flowing into precisely that gap; for example, NOAA's Ocean Acidification Program has funded a $1,877,644 project (FY2023–2025, PI David Nicholson) to assess carbon dioxide removal and ecosystem response for an OAE field trial, with results intended to help establish MRV practices for future carbon removal . The honest position for a buyer or policymaker: the *safety* evidence is genuinely improving and is increasingly empirical; the *efficacy* evidence remains almost entirely model-derived, and will likely stay that way, because the physics prevents direct measurement of the quantity being sold. The right response is not to dismiss OAE but to demand model validation against tracer-release experiments, published uncertainty distributions rather than point estimates, and conservative discounting — and to treat any operator quoting a precise tonnage without disclosing the model, its resolution and its error bars as making an unverifiable claim.
Judge’s note — All four gold checklist items are surfaced explicitly and quantitatively: 384 in-scope OAE papers with corpus provenance, 16 field trials at 4.2% of all / 5.2% of classified, the modeling (102) / lab (96) / review (77) dominance, and the 76 unclassified papers with an explicit statement that the split describes the classified subset (the only soft spot is that it frames the corpus as 'a curated floor, not a census' rather than asserting the pathway query is a complete OpenAlex result set). The answer goes well beyond the checklist with a genuinely expert-level critique: it reads the 16 'field trials' and disaggregates them into natural analogues, mesocosms, ship incubations, and the small handful of true engineered open-water releases with plume tracking (Mg(OH)2 outfall, Ebb Sequim Bay ~2.5 m detectable signal, Tasmania NaOH, Martha's Vineyard rhodamine proxy, Muping olivine) — a distinction most answers miss. Beyond-corpus findings include: LOC-NESS Feb 2026 Ocean Sciences preliminary results (2–10 t measured, ~50 t modeled, EPA permit, ecological monitoring); Planetary's 625.6 t Isometric-verified delivery and buyers; the Isometric ESM protocol effective date; the EC CRCF ERW/OAE technical workshop (Oct 2025); Horizon Europe OAE calls; London Protocol amendment ratification status; Ocean Visions–Fugro EIA framework timing. Citations are consistently named (journal + month/year, outlet, registry) and thus checkable; coverage honesty is exemplary (classifier limitation, 20% unclassified, non-English and preprint exclusion, unconfirmed Ebb Macoma status, search-limit admission). Structure is tight and decision-useful, with a strong closing evidence-vs-commitment contrast (~642 kt contracted vs ~626 t verified). Minor risk that some specific figures/dates are unverifiable, and the completeness claim for the OpenAlex query is arguably undercut, which keeps it just short of a 10.
I'll start by pulling the evidence profile for OAE and the paper-level breakdown. Now let me get the precise study-type distribution and the full field-trial list, then verify against the web. ## Bottom line The field evidence for ocean alkalinity enhancement (OAE) is **thin but no longer absent, and it improved sharply in 2025–26**. The literature is dominated by models and lab work; genuine open-water, engineered alkalinity releases with in-water tracking number in the *single digits* worldwide, and the total CO₂ removal ever directly measured in the field is on the order of **tens of tonnes** — against roughly **640,000 tonnes already under contract**. That gap, not the science of the chemistry, is the central credibility problem. --- ## 1. The literature split, counted rather than eyeballed From the Carbon Curve paper layer (OpenAlex CC0 metadata, 2010→, English, in-scope classification; **384 in-scope OAE papers**, 7,535 citations, fetched 2026-09-02): | Study type | Papers | Share of all 384 | Share of the 308 *classified* | |---|---|---|---| | **Modeling** | 102 | 27% | **33%** | | **Lab** | 96 | 25% | 31% | | **Review** | 77 | 20% | 25% | | Unclassified (no rights-cleared abstract) | 76 | 20% | — | | **Field trial** | **16** | **4.2%** | **5.2%** | | Survey (social science) | 9 | 2.3% | 2.9% | | LCA | 8 | 2.1% | 2.6% | So: **roughly one in twenty classified OAE papers involves going in the water at all**, and modeling alone outnumbers field work **6.4 to 1**. Modeling papers in this corpus date back to 2013; the earliest paper classified as an OAE field trial is February 2022. The theory has a decade's head start on the observations. Two honest caveats on these counts: study types were assigned by a model from title and abstract (classifier v1, 2026-08-12), and 20% of papers carry no classification because no licensed abstract exists — so the field-trial share could be modestly higher or lower. ## 2. "Field trial" is a generous label — the real number is smaller I read all 16. The bucket is doing a lot of work it shouldn't: - **Natural analogues and observational studies, not interventions:** a 26-year South China Sea time series showing surface total alkalinity rising 0.56 µmol kg⁻¹ yr⁻¹ (*GRL*, Sept 2025); benthic alkalinity fluxes in seagrass meadows; oyster-shell dissolution in Narragansett Bay at 4–56 µmol L⁻¹ d⁻¹; basalt dust weathering in Icelandic soils; pyrite oxidation in mined landscapes. Informative about alkalinity cycling, silent on engineered OAE. - **Mesocosms** — enclosed bags in real seawater, which is field-*adjacent*, not open-water: Bergen fjord, Norway (47 days, ΔTA 0–600 µmol kg⁻¹); Helgoland, North Sea (39 days, ΔTA up to 1,250 µmol kg⁻¹); La Spezia, Italy (Aug 2026). - **Ship-based incubations:** 19 experiments in the Equatorial Pacific comparing NaOH, olivine and steel slag (*Comms Earth & Env*, Apr 2025). That leaves a very short list of **actual engineered releases into open water with plume tracking**, all of them proof-of-concept: | Trial | What was measured | |---|---| | Mg(OH)₂ into wastewater → coastal ocean (*Comms Earth & Env*, June 2024) | 74% aqueous CO₂ reduction in the wastewater; ocean signal only a few metres from outfall (−27.2 µatm pCO₂). The authors state plainly it achieved **no CDR** because of its scale. | | Ebb Carbon electrochemical release, Sequim Bay, WA (*Frontiers in Env. Eng.*, Sept 2025) | pH held at the permitted maximum of 9.0 in the mixing tank; measurable signal confined to **~2.5 m** of the discharge pipe; no change in temperature, salinity, turbidity or oxygen. | | NaOH addition, coastal Tasmania — first Australian OAE trial (*Scientific Reports*, Dec 2025) | ΔTA ≈ +545 µmol kg⁻¹ and pCO₂ down up to 370 µatm at the release point; downstream decreases of just 22–77 µatm, i.e. **signal strengths of under 1% to 5%**. | | Rhodamine tracer release south of Martha's Vineyard, Sept 2023 (*Biogeosciences*, Oct 2025) | A *proxy* study — no alkalinity added. Over 36 h the calculated CDR signal (+4 µatm fCO₂, +1.8 µmol kg⁻¹ DIC) reached only **10% of total potential CDR**. | | Olivine addition, Muping marine ranch, Yantai (2022) | Particle-attached microbial community composition shifted; free-living community unchanged. | Add one enabling-technology paper: high-frequency **in situ** total-alkalinity sensing for OAE field trials (*Communications Engineering*, Apr 2026). ## 3. The strongest field result to date is not yet in the corpus Beyond the corpus's 2026-09-02 paper cutoff, the most consequential result is **LOC-NESS**, which I found via web search rather than the corpus: Woods Hole Oceanographic Institution announced preliminary findings on **25 February 2026** at the Ocean Sciences Meeting in Glasgow, from the first **EPA-permitted** OAE field trial, conducted in **August 2025** in federal waters in the Wilkinson Basin area of the Gulf of Maine (whoi.edu press release; also reported by *Chemistry World*, ecomagazine, Ocean News & Technology). The design: a six-hour engineered dispersal of highly purified sodium hydroxide with rhodamine tracer, followed by four to five days of tracking using ships plus four autonomous underwater vehicles and gliders. The headline numbers, per *Chemistry World*'s report of the briefing: **an additional 2–10 tonnes of CO₂ drawn into the ocean during the four-day monitoring window, with modelling suggesting this could reach about 50 tonnes over a year**. WHOI's Adam Subhas framed the significance as tracking precision rather than tonnage. On ecology, the team reported no measurable reductions in phytoplankton or zooplankton and no impacts higher in the food chain, with fish and lobster explicitly monitored. EPA's own permit determination held that the project "demonstrates scientific rigor" and was not expected to significantly affect human health or the marine environment. Read that carefully: **the best-instrumented OAE field trial ever run measured single-digit-to-low-double-digit tonnes of removal, and extrapolated the rest by model.** ## 4. Why in-water measurement can never carry the whole load This is a structural point, not a funding gap, and it is why the modeling share of the literature will stay high: - **Air–sea equilibration is slow.** The Bergen fjord mesocosm estimated **~1,050 days** to reach 95% equilibration for ΔTA of 600 µmol kg⁻¹ (*Biogeosciences*, Jan 2026). The near-coast modeling study found most locations plateau at only **0.6–0.8 mol CO₂ per mol alkalinity after 3–4 years**, with Hawaii taking 8–10 years and downwelling regions such as around Iceland losing up to half the CDR potential to bottom waters (*Biogeosciences*, Jan 2023). - **The signal dilutes below detection almost immediately.** Every trial above confirms it: metres, not kilometres. The tracer study concluded OAE signals are detectable only above TA >10 µmol kg⁻¹, pH >0.01 and fCO₂ >10 µatm. - **So MRV is model-based by necessity.** A 2025 *Biogeosciences* paper on OAE efficiency metrics states that MRV will rely to a great extent on general circulation models, parameterised and constrained by experimental measurements. An EGU 2026 abstract puts the core difficulty as the weak OAE signal against natural carbonate-system variability. Isometric's Electrolytic Seawater Mineralization protocol (effective 2025-03-17) leans on direct measurement of inputs and outputs plus reconciliation calculations rather than measuring the atmospheric flux itself. Field trials, in other words, are best understood as **model-validation exercises**, and that is exactly how the Martha's Vineyard authors framed theirs. ## 5. Two findings that should temper confidence - **Efficiency can collapse.** Lab work on quick and hydrated lime found runaway CaCO₃ precipitation could cut uptake efficiency from ~0.8 down to **0.1 mol CO₂ per mol alkalinity added** (*Biogeosciences*, Aug 2022). A companion study found net alkalinity *loss* above ΔTA ~600 µmol kg⁻¹ for Mg(OH)₂, and that commercially available ultrafine Ca(OH)₂ generally produced net TA loss at tested doses (*Biogeosciences*, Feb 2023). Feedstock choice is not a detail. - **The ecological literature may be testing the wrong doses.** "Lethal by Design?" (*JGR Oceans*, Aug 2026) argues that ΔTA >1,000 µmol kg⁻¹ — common in OAE experiments — occurs in reality only for minutes in a minuscule fraction of perturbed water, while the ubiquitous real-world range is **1–100 µmol kg⁻¹**, which is rarely the focus of environmental assessments. Reassuring mesocosm results at high dose and worrying ones both risk being miscalibrated. Note the uncomfortable overlap: the realistic perturbation range sits right at or below the stated detection thresholds. ## 6. Evidence versus commercial commitment The corpus holds **14 OAE deals totalling 642,102 tonnes contracted**, value disclosed in only 2 of them: - Microsoft–Ebb Carbon, Oct 2024: up to **350,000 t** (later coverage), described at the time as the largest marine CDR deal - Frontier–Planetary Technologies, Aug 2025: **116,964 t for $31.3M ≈ $272/t**, Tufts Cove, Halifax, delivering 2026–2030 - Frontier–CREW Carbon, Dec 2024: **72,492 t for $32.1M ≈ $447/t**, New Haven, CT - Boeing–Equatic, May 2023: **62,000 t**, Singapore - Microsoft–CREW, Aug 2026: up to 23,602 t; Google–Ebb, Dec 2025: 3,500 t (Saudi Arabia, with the Saudi Water Authority) - Two deals with Running Tide, which shut down in 2024 — a reminder that contracted ≠ delivered Against that, **actual independently verified delivery stands at 625.6 tonnes**: Planetary's Halifax credits, verified by Isometric under its coastal-outfall OAE protocol and bought by Stripe, Shopify and British Airways (June 2025, per Planetary and Isometric). That is roughly **0.1% of contracted volume**, and about one-thousandth of it. ## 7. What I'd watch, and what I couldn't confirm - **The LOC-NESS peer-reviewed papers.** February 2026 findings were preliminary, presented at a conference. Full publication is the single biggest pending update to this evidence base. - **Whether Planetary hits its 2026 ramp.** Frontier deliveries began in 2026; verified issuance volumes through late 2026 are outside the corpus and I hit my search limit before confirming them. - **CRCF certification.** The European Commission held its first technical workshop on whether ERW and OAE are ready for certification (Oct 2025, per *CDR Policy Scoop*); Horizon Europe 2025–27 carries explicit OAE calls on impacts, risks and MRV protocols within €389M of CDR-relevant calls. - **Regulatory scaffolding is still provisional.** The London Protocol's 2013 marine-geoengineering amendment had only six acceptances as of October 2023 and is not in force; the 2023 LC/LP statement still says marine geoengineering beyond legitimate scientific research should be deferred. The Ocean Visions–Fugro environmental impact assessment framework is due 2027. **Explicit gaps in what I've told you:** the corpus is a curated floor, not a census — deal coverage starts in 2023, prices are disclosed for 1.5% of all contracted tonnes market-wide, and papers stop at 2026-09-02, so anything published in the last months of 2026 is invisible here. Non-English literature and preprints are excluded. I could not verify current Ebb Carbon Project Macoma operating status or late-2026 issuance data. And the 4–5% field-trial share rests on a machine classification I have spot-checked by reading, not on an independent audit.
Judge’s note — All four gold checklist items are present with the exact counts (384 in-scope, 16 field trials, 102 modeling / 96 lab / 77 review, 76 unclassified explicitly excluded from the denominator and flagged as such). The answer goes well beyond the tally: it reads the 16 field-trial records individually and shows that only ~5 are deliberate open-water alkalinity additions, with several misfiled natural-alkalinity or dye-tracer studies — a genuinely more informative framing than the raw percentage. Beyond-corpus contributions include the LOC-NESS Gulf of Maine August 2025 EPA-permitted trial with dosing volumes and the 2–10 t measured / ~50 t modeled result, the June 2025 Planetary–Isometric first verified OAE credits (625.6 t) and the earlier 138 t announcement, plus explicit acknowledgement of non-English/preprint gaps. Citations are named and checkable (journals, outlets, dates, project names). Honesty is a strength: model-assigned labels flagged as unaudited, mesocosm/lab boundary fuzziness quantified in its effect, deal layer explicitly not a market census, derived $/t marked as division rather than quoted, unverified project statuses listed. Minor quibbles: the field-trial share is stated as ~5% of classified where the checklist frames it as ~4%, and the 16-vs-21 discrepancy, though explained, adds slight noise. Structure and decision-usefulness (contracted ~642k t vs ~600 verified t; confidence tiering by claim) are excellent for an expert reader.
I'll start by checking what the corpus covers, then pull the OAE evidence profile.
Now let me get the year distribution and the full field-trial list, then search the web.
Now the required web searches for what the corpus can't see.
## Bottom line
OAE has crossed from "modeling exercise" into "measured in water," but only barely, and only at scales three to four orders of magnitude below the tonnages already under contract. On the Carbon Curve corpus's own classification, **roughly 5% of the classified OAE literature is field work; about two-thirds is modeling or lab/mesocosm bench science.** And when you read the field-trial papers individually, a majority of them are not deliberate alkalinity releases at all — they are mesocosms, shipboard incubations, or observations of *natural* alkalinity sources. The number of published, deliberate, open-water alkalinity additions with in-water carbon accounting is in the single digits worldwide.
---
## 1. The literature split (Carbon Curve Intelligence, papers fetched 2026-09-02)
The corpus holds **384 in-scope OAE papers** (OpenAlex CC0 metadata, 2010+, English), of which 308 carry a classified study type:
| Study type | Papers | Share of classified (308) |
|---|---|---|
| Modeling | 102 | 33.1% |
| Lab (incl. many mesocosms) | 96 | 31.2% |
| Review / synthesis | 77 | 25.0% |
| **Field trial** | **16** | **5.2%** |
| Survey | 9 | 2.9% |
| LCA | 8 | 2.6% |
| *Unclassified* | *76* | *(excluded from the base)* |
So: **modeling + lab = 198 papers, 64% of the classified corpus. Field trials = 16, or 4.2% of all 384 records.** Review papers alone (77) outnumber field trials nearly five to one — a classic signature of a field that is theorising faster than it is measuring.
Publication volume is growing steeply: 26 papers in 2022 → 47 in 2023 → 70 in 2024 → **125 in 2025** → 80 in the first ~7.5 months of 2026. Citation weight still sits with the older synthesis and modeling work: the two most-cited items are Renforth & Henderson's 2017 *Reviews of Geophysics* review (638 citations) and a 2019 *Frontiers in Climate* modeling paper on EW/OAE ecosystem risks (450 citations).
**An important caveat on these counts:** the study-type labels were assigned by a model from title and abstract (v1, 2026-08-12), not by human coding. The boundary between "lab" and "field-trial" is genuinely fuzzy for mesocosms — a 2026 La Spezia port mesocosm on benthic calcifiers is filed as *lab*, while the Helgoland and Bergen fjord mesocosms are filed as *field-trial*. Move mesocosms to one side or the other and the field-trial share swings by several points. Treat 5% as an order-of-magnitude figure, not a precise one.
## 2. Reading the 16 field trials individually makes the picture thinner, not thicker
Pulling the field-trial list (a broader filter returns 21 papers tagged OAE, including secondary tags), the set breaks down roughly as:
**Deliberate in-water alkalinity addition, open water or outfall — the real thing (~5):**
- **Tasmania, Australia (2025, *Scientific Reports*)** — the first Australian OAE field trial, continuous NaOH addition at a coastal site. At the release point pCO₂ fell up to 370 µatm and alkalinity rose ~545 µmol/kg, but **downstream pCO₂ decreases were only 22–77 µatm, signal strengths of <1–5%**, with the plume dispersing within metres.
- **Sequim Bay, WA (2025, *Frontiers in Environmental Engineering*)** — Ebb Carbon's electrochemical alkalinity discharged through a wastewater plant at PNNL. pH held at the permitted maximum of 9.0 in the mixing tank; at the outfall **the measurable signal was confined to ~2.5 m of the discharge pipe**, with no detectable change to temperature, salinity, turbidity or oxygen.
- **Magnesium hydroxide into wastewater (2024, *Communications Earth & Environment*)** — a 10% alkalinity increase cut aqueous CO₂ 74% in the wastewater, but the ocean signal was 27.2 µatm pCO₂ over a few metres, and the authors state plainly that **the trial did not achieve CDR because of its small scale**.
- **Muping Marine Ranch, Yantai, China (2022)** — coastal olivine addition, measuring microbial community shifts.
- **High-frequency in-situ TA analyzer (2026, *Communications Engineering*)** — first field deployment of an autonomous Lab-on-a-Chip alkalinity analyser during a Mg(OH)₂ slurry trial; 314 alkalinity measurements over 40 days, 60 m from discharge, detecting a ~40 µmol/kg rise after ~210 tonnes of alkaline addition.
**Enclosed mesocosms (~4):** Bergen fjord (Norway, 47 days, ΔTA 0–600 µmol/kg), Helgoland North Sea (39 days, ΔTA to 1,250 µmol/kg), the Oikopleura zooplankton study (53 days, Norwegian fjord), La Spezia. These are in seawater but not in the ocean — no open-system dispersion, no real air-sea equilibration geometry.
**No alkalinity added at all (~7):** a rhodamine dye tracer release south of Martha's Vineyard used as an OAE *proxy*; 26 years of natural alkalinity trend in the South China Sea; oyster farms, oyster reefs, bivalve shell dissolution, seagrass sediments; and two records that appear misfiled to OAE entirely (mountaintop-mine pyrite weathering, Southern Ocean profiling floats).
That last group matters for your question: several of the papers the taxonomy counts as OAE "field evidence" are observations of natural or agricultural alkalinity fluxes, not tests of an intervention.
## 3. What the corpus does not have: LOC-NESS (found by web search)
The most consequential field result to date post-dates or sits outside the corpus's paper records. Per **Woods Hole Oceanographic Institution's February 2026 announcement** (Newswise, 25 Feb 2026) and coverage in *Chemistry World* (15 May 2026) and the *Boston Globe* (26 Feb 2026):
- The **LOC-NESS** project ran the **first US EPA-permitted OAE field trial** in federal waters in the Gulf of Maine in **August 2025**, with preliminary findings presented at the Ocean Sciences Meeting in Glasgow in February 2026.
- It dispersed **~65,000 litres of sodium hydroxide over 1 km²** alongside 760 litres of rhodamine dye, tracked across a 3,600 km² zone using three vessels, four AUVs, CTD casts, plankton tows and NASA's PACE satellite.
- Headline result: an estimated **2–10 tonnes of additional CO₂ drawn into the ocean over the four-day monitoring window**, with modelling suggesting up to ~50 tonnes over a year. WHOI reported **no measurable impact on ocean life**, with laboratory biological effects appearing concentration- and exposure-dependent.
This is the strongest open-ocean datapoint yet — and note what it says about MRV: the *measured* window is four days and a handful of tonnes; the *claimed annual* figure is a model extrapolation. That is the structural condition of OAE evidence today.
## 4. The MRV problem is the evidence problem
The field's own MRV synthesis (Ho et al., *State of the Planet*, 2023) concedes that direct observation should not be expected to track every molecule of additional CO₂, because doing so would be an insurmountable barrier — the field instead defines what can feasibly be observed. Sensor precision is a hard constraint: pH uncertainty of 0.01 translates to roughly 10 µatm of pCO₂ uncertainty, against field signals that the Tasmanian and Sequim trials measured at single-digit percentages of background.
The 2025 Martha's Vineyard tracer study (*Biogeosciences*) put numbers on the achievable detection floor: with suitable conditions, OAE signals should be detectable at TA >10 µmol/kg, pH >0.01 and fCO₂ >10 µatm — and over 36 hours the calculated CDR signal amounted to **10% of the total potential CDR**, with the rest necessarily inferred. Its explicit purpose was to **validate the model-based MRV approaches currently in use**, which is a candid statement that models, not measurements, are doing the accounting today.
A 2026 *JGR Oceans* modeling paper ("Lethal by Design?") adds a sharp methodological critique in the other direction: ΔTA >1,000 µmol/kg — a perturbation common in OAE experiments — would occur only **for minutes in a minuscule fraction** of real OAE-perturbed water, while the realistic range of 1–100 µmol/kg is not what environmental assessments usually study. In other words, much of the existing ecological safety literature may be testing the wrong doses.
## 5. Contracted tonnes vastly exceed delivered, measured tonnes
The corpus holds **14 OAE deals totalling ~642,102 tonnes contracted**, of which only **2 disclose a value** — so any price statement here describes a disclosing minority, not the market. (Corpus-wide, 44 of 385 deals disclose a value, covering 1.5% of contracted tonnes.) The two disclosed: **Frontier–Planetary Technologies, 2025-08-26, 116,964 t for $31.3M (≈$272/t, derived by division, not quoted)** and **Frontier–CREW Carbon, 2024-12-17, 72,492 t for $32.1M (≈$447/t, derived)**. The largest single record is **Microsoft–Ebb Carbon, 2024-10-01, 350,000 t**, value undisclosed. Most of these records are machine-extracted (`review_status: auto`); the Microsoft–CREW 2026 record is flagged `pending` with an explicit note that its pathway classification needs a human call.
Against that, actual verified delivery: Planetary and Isometric issued the **first-ever verified OAE credits in June 2025 — 625.6 tonnes**, from 1,103.2 tonnes of alkalinity dosed at Tufts Cove, Halifax, to Stripe, Shopify and British Airways (Planetary Technologies; Carbon Herald, 17 June 2025). Planetary had announced the first 138 tonnes of net removals a month earlier. **Roughly 600 verified tonnes against ~642,000 contracted.** The ratio is about 1:1,000.
## 6. Where I'd put the confidence
- **Well supported by measurement:** alkalinity can be added to coastal seawater safely at small scale; plumes disperse fast and locally; short-term ecological effects at realistic doses appear small to undetectable across Norwegian, German, Tasmanian, Italian, US and Chinese studies.
- **Partly supported:** carbonate-chemistry efficiency (mol CO₂ per mol TA) — measured in lab, constrained in situ only for hours to weeks, extrapolated for years. Runaway CaCO₃ precipitation risk is well characterised in the lab (Moras et al. 2022; Hartmann et al. 2023) but not in open deployment.
- **Not yet supported by field measurement:** durability over decades, basin-scale efficiency, air-sea equilibration completion (the Bergen mesocosm estimated ~1,050 days to 95% equilibration for ΔTA 600 µmol/kg), cumulative or repeated-dosing ecological effects, and net CDR at commercial scale. All of these currently rest on models.
## 7. Gaps in what I've told you
- The corpus's paper layer stops at **2026-09-02** and holds only English-language, OpenAlex-indexed work; preprints (including much CDRXIV material) and non-English literature — notably Chinese and Japanese mCDR work — are under-covered. LOC-NESS's peer-reviewed papers, if published after that date, are not in these counts.
- Study-type labels are model-assigned and unaudited; the 16 vs. 21 discrepancy between the evidence profile and a direct filtered search reflects secondary pathway tags and in-scope/out-of-scope flags, not two different realities.
- The deal layer is explicitly **not a market census** (CDR.fyi is the fuller scoreboard), and unannounced or sub-threshold OAE contracts are invisible here.
- I did not verify current operating status for Ebb Carbon's Project Macoma, Equatic's Singapore plant, Limenet, Pronoe or Planeteers — the corpus records their deals but not their delivery track records, and project stage in this database is "what the source last said," not a live status.
Judge’s note — All four gold checklist items are surfaced explicitly and quantitatively: 384 in-scope papers described as computed over the full corpus rather than a sample, 16 field trials at 4.2%, modeling 102 / lab 96 / review 77, and the 76 unclassified papers with a correct re-basing onto the 308 classified subset. The answer goes further usefully by noting a count discrepancy (16 in the summary profile vs 21 enumerated under the same filter) and by decomposing the 'field trial' bucket into genuine open-water alkalinity releases, MRV method development, enclosed mesocosms, and natural analogues — which is the substantively correct critique of the headline number. The modeling-reliability section (inter-model divergence, regional vs global efficiency mismatch, detection-floor errors an order of magnitude above signal) is exactly the argument an expert would want. Coverage honesty is exemplary: it flags the absent web-search capability up front, labels the LOC-NESS release as a media claim rather than a result, marks crawler metadata (Planetary site_status) as unverified, and closes with a specific list of unresolved external checks. Structure and decision-usefulness are high, with a clear epistemic verdict separating strong chemistry from weak in-water carbon accounting. The main shortfall is the beyond-corpus probe: no post-fetch literature, no named 2025-26 campaign results sourced outside the corpus, and no non-English coverage — the answer is candid that it could not attempt this, which preserves honesty but not credit. Citation quality is strong on named-journal-plus-date traceability, though a few venue attributions look shaky enough that a reader would want to spot-check them before quoting.
I'll start by mapping what the corpus holds on ocean alkalinity enhancement. ## Bottom line OAE has strong *chemistry* evidence, moderate *ecological* evidence, and weak *in-situ carbon-accounting* evidence. The literature is dominated by models, tanks and enclosures; genuine open-water alkalinity releases number in the single digits, all are tiny, and **not one of them has directly measured the CO₂ uptake it claims** — in every published trial the signal dissipated within metres to a few kilometres of the outfall and net removal had to be inferred from models. Meanwhile ~642,000 tonnes of OAE have been contracted against ~626 tonnes ever independently verified. ⚠️ **Tooling caveat:** I have the Carbon Curve corpus (papers fetched 2026-09-02; deals to 2026-08-28) but **no web-search tool in this session**, so I could not run the recommended checks for post-fetch literature and announcements. Everything below is corpus-sourced; I flag at the end what needs verifying externally. --- ## 1. The split: modeling vs in-water measurement Across **384 in-scope OAE papers (2013 → 2026-08-14, 7,535 citations)**: | Study type | Papers | % of all 384 | % of the 308 classified | |---|---|---|---| | Modeling | 102 | 26.6% | 33.1% | | Lab (incl. tank/mesocosm) | 96 | 25.0% | 31.2% | | Review | 77 | 20.1% | 25.0% | | **Field trial** | **16** | **4.2%** | **5.2%** | | Survey | 9 | 2.3% | 2.9% | | LCA | 8 | 2.1% | 2.6% | | Unclassified (no rights-cleared abstract) | 76 | 19.8% | — | *(Source: Carbon Curve evidence profile for `oae`, computed over the full corpus, not a sample.)* One inconsistency worth stating: the profile's summary counts **16** field trials, but paginating the same `pathway=oae, study_type=field-trial` filter enumerates **21** records — the extra ones are papers where OAE is a secondary tag (e.g. an ERW natural-analogue study). Either figure lands in the same place: **roughly 4–6% of the OAE literature is field work, versus ~57% modeling + lab.** Topic split (overlapping tags; 98 of 384 untagged): environmental risks/co-benefits 162, MRV-quantification 96, governance-legal 44, markets-economics 26, policy 24, public perception 14. Output is also very new: **125 papers in 2025 and 80 in the first eight months of 2026**, versus 26 in all of 2022. The field is young enough that most of it is still hypothesis-generating. --- ## 2. What the "field trials" actually are The 16–21 field-trial records are not 16 alkalinity releases. Sorting them: **(a) Genuine intentional alkalinity addition to open water — about five studies, all sub-kilotonne:** - **Ebb Carbon / PNNL, Sequim Bay, WA** (Frontiers in Environmental Engineering, 2025-09-18): electrochemically derived alkalinity discharged via a wastewater outfall; mixing-tank pH held at the permit ceiling of 9.0; **at the outfall the measurable signal was confined to within ~2.5 m of the discharge pipe**; no detectable change in temperature, salinity, turbidity or oxygen. Explicitly framed as "proof-of-concept." - **Tasmania NaOH release** (Scientific Reports, 2025-12-12) — first Australian OAE field trial: ΔTA ≈ +545 µmol/kg and ΔpCO₂ up to −370 µatm *at the release point*, but downstream maxima of only 22–77 µatm, i.e. **signal strengths of <1–5%** against background. - **Mg(OH)₂ into wastewater discharged to sea** (Communications Earth & Environment, 2024-06-28; Planetary's Cornwall work): efficiency 0.73 mol CO₂/mol alkalinity *inside the wastewater*, but in the receiving water the signal was −27.2 µatm pCO₂ within a few metres. The authors state plainly it **"did not achieve CDR due to its small scale."** - **Muping Marine Ranch olivine addition, Yantai, China** (Frontiers in Microbiology, 2022-02-10) — ecological, not carbon accounting. - **Equatorial Pacific ship-board additions** (Comms Earth & Environment, 2025-04-07): 19 experiments at realistic ΔTA 16–29 µmol/kg comparing NaOH, olivine and steel slag. **(b) MRV method development without adding alkalinity:** the WHOI rhodamine tracer release south of Martha's Vineyard (Biogeosciences, 2025-10-10) — 36 h tracking, dynamic baselining cut baseline variability 25% for fCO₂ and 60% for TA, but the modelled CDR signal reached only **10% of total potential CDR** over the tracking window. Also new high-frequency in-situ TA instrumentation (Communications Engineering, 2026-04-22). **(c) Enclosed mesocosms classed as field work:** Bergen fjord, Norway (Biogeosciences, 2026-01-08; ΔTA 0–600 µmol/kg, 47 days, estimated **~1,050 days to 95% air-sea equilibration**), Helgoland/North Sea (Biogeosciences, 2026-07-08), La Spezia (Marine Pollution Bulletin, 2026-08-14). **(d) Natural analogues, not interventions:** a 26-year South China Sea time series showing natural TA rise of 0.56 µmol/kg/yr (GRL, 2025-09-02); Icelandic basalt-dust soils; oyster-reef and bivalve-shell alkalinity regeneration in Narragansett Bay; seagrass benthic alkalinity fluxes. These are physically informative but are not tests of engineered OAE. **The largest release on record isn't in the peer-reviewed set yet.** A corpus signal from a Guardian Seascape feature (2026-08-20) reports WHOI's **LOC-NESS** programme released ~65,000 litres of sodium hydroxide with dye 40 miles offshore in the Gulf of Maine's Wilkinson Basin, monitored for four days by AUVs; preliminary lobster work showed no lethal or behavioural effects, RNA analysis pending. Treat as a media claim — the corpus holds only LOC-NESS's community-engagement "field notes" paper (ICES JMS, 2026-06-01), not results. --- ## 3. Why the modeling dependence is the core problem — and how shaky it currently is Because OAE plumes dilute below detection in metres, quantification is model-based by necessity. Three 2026 papers puncture the assumption that models are ready to carry that load: - **Inter-model divergence:** matched pulse-injection simulations in CESM2/MARBL (1°) vs ECCO-Darwin (⅓°) at twelve locations show differences "well beyond the expected intrinsic variation of each model," mostly attributable to vertical transport representation, with the biggest gaps in the first seven years post-injection (Biogeosciences, 2026-07-21). - **Regional vs global mismatch:** a 2 km model of Bass Strait finds efficiencies of **0.11–0.27 mol CO₂/mol TA versus 0.31 from a global model in the same region**, with 70.7% ± 10.5% of added alkalinity subducted below the mixed layer at the shelf break before equilibrating (Frontiers in Climate, 2026-07-03). - **Detection floor:** local-scale errors in state-of-the-art global products/models are 19–37 µatm fCO₂ (products) and 25–56 µatm (models), **more than an order of magnitude larger than the ~1 µatm signal expected from open-ocean mCDR** (Environmental Research Letters, 2026-06-22). Add the efficiency literature: near-coast equilibration plateaus at **0.6–0.8 mol CO₂/mol TA after 3–4 years**, with downwelling regions losing up to half their CDR potential (He & Tyka, Biogeosciences, 2023-01-03); and mineral-based open-ocean OAE loses >75% efficiency if grain size doubles, requiring forsterite near **1.7 µm** to work — "questioning the suitability of olivine" (Biogeosciences, 2026-05-12). --- ## 4. Where the evidence *is* solid - **Carbonate-chemistry limits are well characterised.** Runaway CaCO₃ precipitation thresholds are mapped: no precipitation at Ω_arag ≈ 5, aragonite forming above ~7, with efficiency collapsing from ~0.8 to ~0.1 mol CO₂/mol TA in runaway conditions (Moras et al., Biogeosciences, 2022-08-01); alkalinity-stability limits by feedstock (Hartmann et al., Biogeosciences, 2023-02-20); a 2026 study found stability for a month at +3,000 µmol/kg but rapid precipitation at +14,000, with **no runaway observed** (Frontiers in Marine Science, 2026-06-18). - **Moderate-dose ecological effects look modest so far.** Coastal Tasmanian phytoplankton at +500 µmol/kg: "moderate" effects (Biogeosciences, 2022-12-01); *Oikopleura dioica* unaffected to ΔTA 600 (PLoS ONE, 2026-03-30); macro-benthic calcifiers unaffected below the site-specific saturation threshold (Mar. Poll. Bull., 2026-08-14); Antarctic krill biomagnification factors <1 for steel-slag metals (ICES JMS, 2026-05-01). - **Feedstock choice matters more than "OAE" as a category.** In the Equatorial Pacific trials, NaOH had negligible plankton effects with predictable alkalinity; olivine disrupted cyanobacteria, heterotrophic bacteria and picoeukaryotes while delivering only **0.06 mmol alkalinity/g** versus **8 mmol/g** for steel slag. - **Experimental design is being corrected.** "Lethal by Design?" (JGR Oceans, 2026-08-01) shows ΔTA >1,000 µmol/kg — common in OAE experiments — occurs for only minutes in a minuscule fraction of real perturbed volume, while the ubiquitous real-world range is **1–100 µmol/kg**, which most environmental assessments don't test. Much of the "reassuring" ecotoxicology may be testing the wrong doses in both directions. --- ## 5. The commercial gap | Metric | Value | |---|---| | OAE deals in corpus | 14, totalling **642,102 t** contracted (2 disclose value) | | Largest | Microsoft–Ebb Carbon, 350,000 t (2024-10-01), "world's largest marine CDR deal to date" | | Priced deals | Frontier–Planetary 116,964 t / $31.3M / **$272/t** (2025-08-26); Frontier–CREW 72,492 t / $32.1M / **$447/t** (2024-12-17) | | OAE **projects** (supply side) | **4**; 1 operating (Vycarb, East River, 60 t/yr); stated annual capacity sums to **3,710 t/yr** | | Independently verified credits | **625.6 t** — Planetary at Tufts Cove, Halifax, from dosing 1,103.2 t of alkalinity; first-ever verified OAE credits, Isometric protocol, 2025-06-16, to Stripe/Shopify/British Airways | That is roughly **1,000:1 between contracted tonnes and verified tonnes**. Two of the deals (Microsoft 12,000 t and Shopify, both with Running Tide) are with a company that ceased operations in June 2024. Also note: the corpus's entity record for **Planetary Technologies** carries `site_status: "dead"` at a 2026-09-01 feed check — that is unverified crawler metadata, not a corporate event, but it needs checking before anyone treats Planetary's 117,590 contracted tonnes as live. Governance is likewise immature: the **London Protocol 2013 marine-geoengineering amendment is still not in force** (six instruments of acceptance as of October 2023; two-thirds needed), and its standing statement is that marine geoengineering beyond legitimate scientific research should be deferred. Isometric's Electrolytic Seawater Mineralization protocol went operational 2025-03-17; Ebb's Project Macoma secured an NPDES permit. Tellingly, the EU ETS review proposal of 2026-07-17 earmarks central purchase of 250 Mt of permanent removals over 2031–2040 for **DACCS and BioCCS — not OAE**. --- ## 6. How I'd summarise the strength of the evidence - **Thermodynamics and dissolution chemistry: strong.** We know how much CO₂ a mole of alkalinity can take up and what breaks it. - **Near-field environmental safety at realistic doses: moderate and improving,** with a real caveat that dose ranges in much of the ecotoxicology don't match deployment, and that olivine looks materially worse than NaOH/slag/limestone. - **Quantified, in-water, additional CO₂ removal: weak.** No field trial has closed the carbon budget by measurement. The plume vanishes; the models that must fill the gap disagree with each other by factors of ~2–3 regionally and have local errors an order of magnitude above the signal. - **Practical implication:** OAE credits sold today are model-attested, not measurement-attested. That is defensible for a research-stage pathway with tight protocols, but it is a different epistemic product from what buyers get from, say, DAC with geological storage — and buyers contracting hundreds of thousands of tonnes should be explicit about it. --- ## 7. What I could not check (please verify externally) Because no web search was available, these are open: 1. **OAE literature published after 2026-09-02**, and non-English or preprint work (CDRXIV items like "A framework for understanding efficiency losses of OAE," 2026-05-04, are held but unclassified). 2. **Peer-reviewed LOC-NESS results** from the Wilkinson Basin NaOH release — currently only a media signal. 3. **Planetary Technologies' current operating status** (the `site_status: dead` flag) and whether Frontier's 2026–2030 delivery schedule is on track. 4. **Carbon to Sea, NOAA and CSIRO Carbonlock trial results** — CSIRO's programme appears only as an August 2026 media signal quoting Elizabeth Shadwick and Richard Matear, who frame it as assessing "whether this is a good idea or not." 5. **CDR.fyi / AlliedOffsets OAE delivery totals**, which are the fuller transaction scoreboard; this corpus is explicitly a curated floor, not a census.
Asked, verbatim: “What does the research literature say about enhanced rock weathering's MRV challenges, and which companies are selling ERW removals anyway?”
Judge’s note — Strong on the substantive science: it surfaces a genuinely expert-level decomposition of ERW MRV failure modes with named sources (Knapp et al. ES&T 2023 multiproxy; the Frontiers in Climate methods review's acid-type caveat; the inorganic-carbon-vs-cation-based argument; the immobile-tracer-enrichment bias that systematically overestimates dissolution; centrifugation vs. lysimeter divergence at the Welsh forest trial; the published dispute over Kantola et al. 2023; CarbonPlan's protocol comparison across Puro v1.0 and CSI Global Rock C-Sink v0.9). The directional-bias point and the downstream-reversal discount-factor gap are exactly the diligence hooks an expert buyer needs. Commercial coverage is also good and clearly post-cutoff-aware (Terradot/Eion consolidation and the 400k+ t contracted portfolio, InPlanet–Microsoft 28,500 credits, first Isometric verification, Eion's first Puro issuance, Drawdown's ~10,000 t delivered-to-date), and the contracted-vs-delivered framing is the right analytical punchline. Weaknesses: it captures only 2 of 4 gold items — no corpus-scale accounting (~1,582 works fetched / ~440 in scope) and, more importantly, no mention that ERW carries ~78 field trials, the largest field-trial count of any pathway, which is a substantive empirical fact about the pathway's evidence base rather than mere metadata. Frontier–Lithos Carbon appears only as an admitted gap. A few citations are bare PMC IDs without titles/authors, and some claims (Mati XPRIZE, deal closure) are explicitly unverified. Honesty is exemplary — background knowledge is clearly separated from retrieved sources, and the gap list is specific and actionable. Well organized and decision-useful.
I'll research this thoroughly — both the scientific literature on ERW measurement challenges and the current commercial landscape.
I ran a set of searches on this; my search budget ran out partway through, so I've flagged below which claims come from retrieved sources versus my own background knowledge that you should verify.
---
# 1. What the literature says: why ERW MRV is genuinely hard
The framing in the peer-reviewed work is unusually blunt. Multiple papers open by stating that MRV *is* the central problem, not a side issue. Knapp et al., writing in *Environmental Science & Technology* (2023, "Quantifying CO₂ Removal at Enhanced Weathering Sites: a Multiproxy Approach"), describe monitoring, reporting and verifying the amount of carbon removed as the main challenge associated with enhanced weathering — and to get around it they had to use radiocarbon, δ¹³C, ⁸⁷Sr/⁸⁶Sr and major-element data at a 40-year-old steel-slag site in Consett, County Durham, UK. A separate *ES&T* paper (2024) on environmental impacts of ERW on agricultural land repeats the same framing and notes "a scarcity of experimental and field trial data" on risks and monitoring strategies.
The problem decomposes into several distinct failure modes:
**a) Dissolution is not the same as carbon removal.** The rock has to dissolve, but whether that dissolution yields net CO₂ drawdown depends on which acid did the dissolving. Weathering driven by carbonic acid removes CO₂; weathering driven by strong acids (nitric and sulfuric acid from fertiliser, atmospheric deposition, or sulfide oxidation) does not. The Frontiers in Climate review of measurement methods (2024) tabulates this explicitly as a shared weakness — nearly every method it compares carries the caveat "no consideration for acid type or CO₂ losses."
**b) Cation-based methods may not be sufficient on their own.** A 2024 Frontiers in Climate paper argues directly in its title that inorganic carbon-based approaches may be *required* to complement cation-based approaches, on the grounds that uncertainty in measuring the sequestration itself has been the binding constraint on the field.
**c) Soil heterogeneity destroys signal-to-noise.** A recent methods paper (PMC12713775) presents "an updated framework and a signal-to-noise analysis for using soil-based mass balance approaches," and makes a specific technical correction: mass balance must account for enrichment of immobile tracer elements in topsoil caused by feedstock mass loss, because omitting that process *systematically overestimates* feedstock dissolution. That is a directional bias, not just noise — an important distinction for anyone buying credits quantified this way. The Frontiers review makes the same broader point: soil sampling gives temporally integrated signals but is "limited by soil heterogeneity which can lead to large uncertainties," and assumes no cation leaching between sampling intervals.
**d) Pore-water sampling is unreliable.** A study on a forest ERW trial in Wales (PMC12750519) found that centrifugation-based pore water extraction detected elevated total alkalinity and Ca in amended plots, while the same effect was *not detectable* using lysimeters. In other words, the choice of sampling hardware can be the difference between measuring a removal signal and measuring nothing.
**e) Time lag and downstream reversal.** An Oxford Department of Physics seminar on uncertainty quantification in open-system CDR (April 2024) lists the outstanding questions as the time lag between weathering and CO₂ storage, downstream reversal risks, and the impact of soil heterogeneity on weathering rates. Carbon captured as bicarbonate has to survive transit through soils, groundwater and rivers to the ocean; secondary carbonate precipitation and CO₂ degassing en route can reverse part of it.
**f) Protocol divergence.** CarbonPlan's explainer on quantifying enhanced weathering makes the governance point: there is a persistent "lack of common understanding about what different quantitative methods — measurements, models, and forms of record keeping — can actually tell us about net carbon removal." Their comparison table covers the Puro Enhanced Rock Weathering Methodology v1.0 and the Carbon Standard International Global Rock C-Sink Guideline v0.9 (alongside the Beerling et al. 2023 field-trial approach), and notes that some protocols simply don't account for certain weathering phases at all ("N/A" entries).
**g) Contested field results.** There is an active published dispute over specific field estimates — one paper is titled "On error, uncertainty, and assumptions in calculating carbon dioxide removal rates by enhanced rock weathering in Kantola et al., 2023," which tells you the community is litigating individual trial numbers, not just methods in the abstract.
**h) Co-impacts that need their own monitoring.** The 2024 *ES&T* impacts paper argues for monitoring beyond carbon accounting, drawing on literature from agriculture, water chemistry and human health to identify both negative consequences and co-benefits of scale-up. (Trace-metal accumulation — nickel and chromium in olivine and some basalts — is the standard concern here; that specific attribution is my background knowledge rather than from a retrieved source.)
**Consensus takeaway:** no single method closes the accounting. The literature is converging on multi-proxy approaches — solid-phase mass balance plus pore-water/alkalinity flux plus isotopic tracers (δ¹³C, δ¹⁸O, ⁸⁷Sr/⁸⁶Sr, ¹⁴C) — while noting these are expensive. The Frontiers review flags radiocarbon in particular as expensive to analyse.
---
# 2. Who is selling ERW removals anyway
The gap between contracted and delivered volume is the headline.
**Scale reality check:** Project Drawdown's ERW page states that more than a dozen companies are selling ERW-based carbon removal credits, with **nearly 10,000 tCO₂ reported to have been removed as of early 2025**. Drawdown's own verdict on the pathway is "Keep Watching," citing unpredictable effectiveness, complex monitoring requirements, and high costs.
| Company | Geography | Status / notable deals |
|---|---|---|
| **Terradot** | US, Brazil | Launched with a $60M investment and record ERW sales (Carbon Herald). Acquired the assets of **Eion** on 6 Feb 2026; the combined entity holds **400,000+ tons** of carbon removal contracts with buyers including **Microsoft, Google and Frontier** — described as one of the largest contracted ERW portfolios in the market (Businesswire / AgFunderNews) |
| **Eion** (now part of Terradot) | US farmland, olivine | Secured a **$33M Frontier offtake, ~79,000 tons by 2030, just under $420/tCO₂**; general pricing $300–400/t. Google participated. Recently secured its first credits through **Puro.earth** from part of its 2023 deployment (AgFunderNews, Carboncredits.com) |
| **InPlanet** | Brazil, tropical soils | First company to have enhanced weathering credits verified by **Isometric** (Jan 2025); later inked a deal to sell **28,500 ERW credits to Microsoft** (Carbon Herald, Quantum Commodity Intelligence). Carbon Herald also reports a company being first to deliver under Isometric's **ERW protocol v1.2** (June 2026) |
| **Mati Carbon** | India (smallholder focus) | Listed in the Enhanced Weathering Alliance as specialising in ERW to durably remove and store CO₂. *(My background knowledge: Mati won the $50M XPRIZE Carbon Removal grand prize in April 2025 — please verify independently, I could not confirm this in search.)* |
| **UNDO** | UK/Canada, basalt | Named among leading ERW credit projects (Regreener, 2026) |
| **Flux** | Africa | EWA member; positioning ERW across the continent |
| **ZeroEx** | Germany | Described as the largest ERW project in Germany, listed on the Senken marketplace |
Pricing in the retail/marketplace segment runs roughly **€200–500 per tonne** depending on project, volume and contract structure, per Regreener's 2026 survey. Note that Regreener's list includes **Carbfix**, which is geological CO₂ injection into basalt rather than soil-applied ERW — a common conflation worth watching for in vendor comparisons.
The registry/standard layer is where credibility is being contested: **Puro.earth** (ERW Methodology v1.0), **Isometric** (EW protocol, now at v1.2), and **Carbon Standard International** (Global Rock C-Sink Guideline v0.9). AgFunderNews acknowledges the reputational overhang, noting that Eion's model relies on selling carbon credits, "some of which have been deemed 'worthless'" — referring to the broader voluntary market rather than ERW specifically.
---
# 3. My read, and the gaps
**The core tension is quantifiable:** roughly 10,000 tonnes reported removed industry-wide as of early 2025, against a single merged company holding 400,000+ tonnes of contracts. Almost all ERW value in the market today is a forward promise resting on measurement methods that the peer-reviewed literature says are not yet settled — and where at least one identified error (immobile-element enrichment in mass balance) biases estimates *upward*.
**Buyer behaviour suggests they know this.** Google's carbon removal program manager framed the Eion deal as being partly about "building a robust, transparent understanding of enhanced weathering's potential" — i.e. purchases are functioning as R&D subsidy with data rights, not purely as offset procurement.
**What I could not verify in this session:**
- Current CDR.fyi totals for ERW purchased vs. delivered (my search allowance ran out before I could pull them). The ~10,000 t Drawdown figure is now likely a year or more stale.
- Whether the Terradot–Eion deal has closed, and the financial terms.
- The Mati Carbon XPRIZE claim.
- Several additional sellers I'd expect to be active (Lithos Carbon, Silicate, Alt Carbon, Greenearth) that did not surface in the results I retrieved.
- Whether any independent third party has re-audited issued ERW credits post-issuance.
**The unresolved scientific question that matters most commercially** is downstream reversal — the fraction of alkalinity generated in soil that actually reaches long-term ocean storage. Most current protocols apply discount factors to this rather than measure it, and I did not find field evidence in the retrieved literature that validates any particular discount rate. If you're evaluating ERW offtakes, that assumption is where I'd focus diligence.
Judge’s note — Nearly all gold items are surfaced substantively: the 440 in-scope ERW literature count with a full study-type breakdown (108 lab, 78 field trials, 78 modelling, etc.), though the ~1,582 fetched figure and the 'largest field-trial count of any pathway' comparison are not stated explicitly; the ERW deal layer is reconstructed supplier-by-supplier with Frontier–Eion ($33M / 80,892 t) and Lithos Carbon present; and the Isometric Enhanced Weathering in Agriculture Protocol (v1.2, conservative discounting §10.1.4) plus Puro.earth methodology form a clear certification-layer discussion. Beyond-corpus/recent-literature depth is strong and specific rather than gestural: the CDRXIV Part 3 EW review's 0.84 tCO2/ha/yr median and downward-revised 0.2–2 Gt/yr technical potential; the Frontiers in Climate uncertainty framework concluding geochemical variance is likely too high for element–element mixing models; the ES&T 'Overlooked Soil Acidity' paper (with honest flagging that only metadata was available); the Biogeosciences carbon-vs-cation MRV paper; the aggregation paper's warning that biased plot selection can manufacture removal from noise; the SATuration–Centrifugation porewater method; plus delivery-side facts (InPlanet's Jan 2025 first verified issuance, Alt Carbon's 9,566 t as largest certified EW deployment, Google–Terradot deliveries starting 2029), Puro General Rules v4.4/ICVCM CCP, CRCF Delegated Regulation 2026/285 excluding ERW, and DOE FOA 3082 mineralisation awards. Citation quality is high — journals, dates, protocol versions, direct quotes, and derived vs stated prices are distinguished. Coverage honesty is exemplary: it distrusts and explains the anomalous 3.67 Mt Lithos figure, identifies the Canadian Wollastonite record as a feedstock contract miscategorised as a credit deal, flags Cella as geological mineralisation rather than agricultural ERW, notes 30-of-35 truncation, notes durability values are pathway defaults not measurements, and states explicitly that search budget was exhausted and nothing after ~Sept 2026 was checked. Clarity is excellent: a sharp thesis (measurement confidence falling while contracted volume rises), a scannable supplier table, and a genuinely useful closing 'thing to watch' on reconciling registry discount factors with published variance. Main residual risk is that a few precise figures (citation counts, award amounts) are unverifiable from the answer itself, and two gold framings (total fetched count, cross-pathway field-trial ranking) are only implicit.
I'll start by pulling the corpus's evidence profile for enhanced rock weathering, then check the literature and deals layers directly. Now let me pull the MRV-specific literature and the full supplier breakdown of ERW deals. Let me verify against the web, including the recent critical MRV literature and anything after the corpus freshness dates. ## Bottom line The ERW literature and the ERW market are moving in opposite directions on confidence. The measurement science has spent the last two years getting *less* certain — narrowing global potential estimates, documenting that the standard quantification models may not resolve the signal at all, and finding systematic upward biases — while contracted volumes have grown to roughly 4.6 Mt in the Carbon Curve corpus alone, sold by a dozen named suppliers under registry protocols that are themselves still on version 1.x. --- ## Part 1: What the literature actually says about MRV **Scale of the evidence base.** The corpus holds 440 in-scope ERW papers: 108 lab studies, 78 field trials, 78 modelling papers, 64 reviews, 15 LCAs, 9 surveys, and 88 unclassified. That is a large literature, but the most-cited works are old and theoretical — the top three by citations are a 2013 *Reviews of Geophysics* review, Beerling et al.'s 2020 *Nature* cropland modelling paper (675 citations), and Strefler et al. 2018, which estimated cropland potential at up to 95 GtCO₂/yr for dunite and 4.9 GtCO₂/yr for basalt at costs of $60/t (dunite) to $200/t (basalt). Those headline numbers are what built the sector's narrative; the field-measurement work testing them is far younger. ### The five recurring MRV problems **1. Global potential has been revised down by roughly an order of magnitude.** The most recent systematic synthesis in the corpus — *An Ecosystem of Carbon Dioxide Removal Reviews – Part 3: Enhanced Weathering* (CDRXIV preprint, 14 July 2026) — reports a median area-normalised CDR flux across empirical and modelling studies of **0.84 tCO₂ per hectare per year, spanning several orders of magnitude**, and converges on a maximum technical potential of **~0.2–2 GtCO₂/yr**. That is against Strefler's 4.9–95 Gt/yr and the 2018 *ERL* review's 2–4 Gt/yr consensus range. Critically, the review attributes the flux variance not only to real site differences but to "the diversity of quantification approaches, which differ by use case and vary in terms of their system boundaries and treatment of loss processes." **2. The workhorse solid-phase method may not resolve the signal.** An uncertainty-aware framework published in *Frontiers in Climate* (12 January 2026) tested how spatial heterogeneity propagates into feedstock-dissolution estimates from element–element mixing models — the dominant approach in commercial MRV. Its conclusion is blunt: analysis of existing high-density soil sampling campaigns "indicates that geochemical variance is likely too high for element-element mixing models to serve as effective near-term constraints on feedstock dissolution." This is a methodological finding about the industry's primary measurement, not a marginal caveat. **3. Loss processes after the initial reaction are unresolved.** The July 2026 review flags "significant uncertainty… regarding the magnitude, persistence, and timing of loss processes after initial CDR has occurred, including secondary phase formation and cation exchange," and notes these depend on local soil and climate conditions. Since ERW credits are sold on the premise of ocean bicarbonate storage over 10,000+ years, everything between the field and the ocean is an accounting assumption, not a measurement. **4. Acid attribution — the newest and most pointed challenge.** A paper in *Environmental Science & Technology* dated 5 August 2026 is titled "Overlooked Soil Acidity Leads to Significant Overestimation of Carbon Sequestration in Enhanced Rock Weathering." The corpus holds only metadata (the paper is closed-access, no abstract), so I cannot report its magnitude, but the mechanism it names is the well-known one: if a strong acid (nitric from fertiliser, sulphuric) rather than carbonic acid dissolves the rock, cations are released without atmospheric CO₂ being consumed — cation-based MRV counts it as removal anyway. A related 2026 *Biogeosciences* paper works the same seam, on "Carbon vs. cation based MRV of Enhanced Rock Weathering and the issue of soil organic carbon" (bg.copernicus.org). The same literature also raises N-cycle side effects: pH change can shift soil emissions of non-CO₂ greenhouse gases. **5. Sampling and aggregation design can manufacture removal from noise.** *Aggregated monitoring of enhanced weathering on agricultural lands* (CDRXIV, 15 October 2025) argues quantification error falls when signals are pooled across many fields and larger catchments — but adds the warning that "aggregation sets must be defined in advance and all plots included, as biased selection of fields can generate apparent removal signals out of statistical noise." Separately, a January 2026 paper in *Cambridge Prisms: Carbon Technologies* introduces a new porewater extraction method (SATuration–Centrifugation) precisely because existing in-field techniques are moisture-dependent, limiting "comparability across sites and seasons." ### What the field trials do show The trials are more encouraging on *agronomy* than on carbon. A CDRXIV preprint (9 August 2026) reports **>20% yield benefits** in a two-year corn-soy strip trial in the southeastern US, with pH increases and cation losses as weathering evidence. An 11.5-hectare, 6,400-tree, four-year reforestation trial (*Communications Sustainability*, 30 June 2026) found silicate rock amendment raised aboveground carbon stocks 27% in broadleaf stands. A Vermont dairy trial (March 2026) found agronomic-rate basalt posed "minimal trace-metal risk." Yield and safety evidence is firming up faster than tonne-accounting does — which matters, because yield is not a credit. --- ## Part 2: Who is selling ERW removals anyway The corpus holds **35 ERW deals, 4,592,794 tonnes contracted, spanning 2022 to May 2026** — but only **6 deals disclose a value** ($117.8M across 328,161 t). Suppliers, by corpus records: | Supplier | Deals | Tonnes contracted | Notable buyers | |---|---|---|---| | Lithos Carbon | 4 | 3,669,280 ⚠️ | Microsoft (11,400 t), Carbonx (3,000 t) | | Canadian Wollastonite | 1 | 450,000 ⚠️ | UNDO (rock feedstock, 3-yr) | | Terradot | 3 | 289,998 | Google (200,000 t), Frontier (89,998 t @ $300/t), Microsoft | | Eion | 3 | 84,392 | Frontier (80,892 t, $33M ≈ $419/t), Perdue AgriBusiness (3,500 t inset) | | UNDO | 8 | 33,043 | Microsoft (15,000 t + 5,000 t 2023), Barclays (6,538 t), British Airways, Auto Trader (340 t) | | InPlanet | 4 | 31,491 | Microsoft (28,500 t, 2026–28), AXA Schweiz (1,950 t), Adyen | | Alt Carbon | 4 | 21,417 | Frontier (9,566 t + 1,851 t), Mitsui O.S.K. Lines (10,000 t), NextGen CDR | | Mati Carbon | 3 | 7,013 | Shopify (5,000 t), Frontier (1,593 t + 420 t) | | Cella | 1 | 3,212 ⚠️ | Frontier (option) | | Flux | 2 | 1,682 | Frontier, Milkywire (540 t @ $370/t) | | Silica | 1 | 1,266 | Frontier | | ZERO / ZeroEx | 1 | n/d | Bain & Company (intent) | ⚠️ **Three caveats on that table.** The Lithos figure of 3.67 Mt is dominated by a record I could not inspect (the corpus truncated its deal list at 30 of 35) and is wildly out of line with every visible Lithos deal; treat it as unverified. The 450,000 t Canadian Wollastonite entry is UNDO *buying rock*, not anyone buying credits — a feedstock supply contract miscategorised in the deal layer. And the Cella record describes CO₂ injection into deep basalt in Kenya, i.e. geological mineralisation, not agricultural ERW. **Prices, where disclosed:** $419/t (Frontier–Eion, derived), $370/t (Milkywire–Flux, traded on Kenya's CYNK platform), $300/t (Frontier–Terradot, derived), ~$200/t average (NextGen–Alt Carbon, as stated). Trade reporting cited in the corpus put current ERW credits at $300–400/t with claimed pathways to $25/t at scale — a 12-fold cost-down that no field data currently supports. ### The delivery gap is the real story Against 4.6 Mt contracted, the largest *verified issuance* on record is **9,566 tonnes** — Alt Carbon's Monsoon Harvest project in Darjeeling and the Terai, delivered to Frontier on 27 May 2026 via Watershed. Isometric's registry operations head confirmed this as the largest single enhanced-weathering deployment the registry has certified (esgpost.com, June 2026), which makes Alt Carbon the top ERW firm by issued credit volume. The first verified ERW credits in the world were only issued in **January 2025**, by InPlanet in São Paulo State under Isometric's Enhanced Weathering Protocol, facilitated by ClimeFi. Google's 200,000-tonne Terradot contract — its largest-ever CDR purchase, signed December 2024 — does not begin delivering until **2029**. ### Buyers are partly funding the MRV they depend on Several deals are explicitly research subsidies in credit clothing. Microsoft's September 2024 follow-on with UNDO includes funding for ERW MRV science with field sites at the University of Guelph, Queen's University Kingston and Newcastle University. Microsoft's Terradot arrangement provides direct support for field trials, lab work and data infrastructure rather than a straight credit purchase. Auto Trader's 340-tonne UNDO purchase funds MRV and agronomy trials. InPlanet and Alt Carbon both contribute to Cascade Climate's ERW "Data Quarry," an industry data-sharing system. ### The certification layer Two registries are doing the work: **Isometric's Enhanced Weathering in Agriculture Protocol** (UK, now at v1.2, which explicitly notes that new quantification technologies "will be regularly assessed… for inclusion") and **Puro.earth's** ERW methodology (General Rules v4.4, approved 7 May 2026; Puro received ICVCM programme-level CCP eligibility on 11 December 2025). Cascade Climate has been coordinating a community ERW quantification standard since November 2023, and its *Foundations for Carbon Removal Quantification in ERW Deployments* is the reference document practitioners point to on measurement problems. Microsoft and Carbon Direct's *Criteria for High-Quality CDR* (2023 edition) also covers ERW. **Public policy has not caught up.** The EU's CRCF Delegated Regulation (EU) 2026/285, adopted 3 February 2026, sets certification methodologies for permanent removals — DACCS, BECCS and biochar. **ERW is not in it.** The Commission held its technical workshop on ERW and OAE on 25 September 2025, and a DG CLIMA presentation from May 2026 lists enhanced rock weathering among methodologies where work has only "started." Elsewhere, ERW is eligible under the US DOE's CDR Purchase Pilot Prize and received four mineralisation pilot awards under FOA 3082 (Eion $3.99M, Lithos $3.98M, Stanford $3.99M, Northwestern), and the UK funds it through the GGR Demonstrator programme. --- ## Gaps and cautions - **Durability is asserted, not measured.** Almost every ERW deal in the corpus carries `durability: durable-1000y` with `durability_basis: pathway-default` — a taxonomy default, not a supplier claim or a verified figure. Deal notes range from "over 10,000 years" to "100,000+ years" to "hundreds of thousands of years" for what is physically the same process, which is itself a signal about how loosely permanence is being described in the market. - **The corpus is a floor.** It is not a census; CDR.fyi is the fuller transaction scoreboard, and 27 of the 35 ERW deals are machine-extracted without spot-check. Contracted tonnes ≠ delivered tonnes ≠ issued credits, and must not be summed against project capacity. - **My web verification was partial.** I confirmed the Isometric/InPlanet/Alt Carbon issuance record, the Cascade/Isometric MRV infrastructure and the CRCF status through mid-2026, but exhausted my search budget before I could verify the current status of Lithos Carbon (whose 3.67 Mt corpus figure I distrust) or read the August 2026 *ES&T* soil-acidity paper beyond its title. I could not check for deals or literature after roughly September 2026. - **The unresolved question** is whether registry protocols can price uncertainty honestly. The literature says field-scale dissolution estimates may be under-determined by the dominant measurement model; the market has nonetheless issued credits against it since January 2025. Isometric applies conservative discounting (protocol §10.1.4), but no published reconciliation exists between the discount factors in use and the variance the 2026 uncertainty papers report. That comparison is the thing to watch.
Judge’s note — Nearly all gold items are surfaced, though with partial credit stretches: the answer gives '440 in-scope ERW papers' but never the ~1,582 works fetched or the completeness framing, and it reports 78 field trials without noting that this is the largest field-trial count of any pathway (in fact it frames the evidence base as skewed away from field work). Frontier–Eion is explicitly itemized with tonnage and value; the large Lithos contract is quantified but not attributed to Frontier. The Isometric/Puro/CRCF protocol layer is handled thoroughly, including protocol version and the fact that CRCF Delegated Regulation 2026/285 omits ERW. Beyond-corpus contributions are numerous and specific: Su et al. ES&T soil-acidity overestimation, Isson & West Science 2026 durability Perspective, Schiedung et al. Nature Rev. Earth & Env. uncertainties taxonomy (land–ocean continuum), Carbon Direct's ~12,000 t first-ever ERW/OAE issuances vs millions contracted, Lithos 5,160 t Dec 2025 issuance, a Microsoft–Terradot 12,000 t March 2026 deal flagged as a corpus gap, Climeworks–Lithos distribution partnership, and Isometric v1.1 / Spring 2026 module update. MRV criticisms are named with mechanisms (no consensus quantification method; pre-registration needed for aggregation sets to prevent signal manufactured from noise; porewater moisture dependence; SOC entanglement; nine-week lysimeter non-detection). Citations carry journals, dates and DOIs. Honesty is exemplary: it distinguishes corpus vs web, flags 27/35 auto-reviewed deal records, unmasks the 450,000 t Canadian Wollastonite row as a feedstock supply agreement inflating pathway totals, notes prices are derived by division, and admits it could not verify the magnitude of the overestimation result. Structure is tight and decision-useful. Deductions only for the two incompletely surfaced corpus facts and the fact that some numeric specifics (e.g., exact page-level details) are asserted without independent cross-check.
I'll start by mapping the corpus, then pull the ERW evidence profile and deal set. Now the MRV-specific literature and the full supplier breakdown. Now the required web checks — recent MRV literature, the soil-acidity overestimation finding, protocol status, and anything after the corpus freshness dates. ## What I checked **Carbon Curve corpus** (freshness at time of query: deals to 2026‑08‑28, ingest 2026‑09‑01; papers fetched 2026‑09‑02; policy events to 2026‑08‑17): `describe_coverage`, the full `erw-ag` evidence profile, a paper search filtered to topic `mrv-quantification`, and an `aggregate_records` run over every ERW deal grouped by supplier. **Web** (to fill the post‑freshness and off‑corpus gap): searches on ERW MRV uncertainty, the ES&T soil‑acidity paper, Isometric protocol versions, CRCF methodology status, and 2026 ERW deals/issuances. I flag below which is which. --- ## 1. The literature: how big and how mature The corpus holds **440 in‑scope ERW papers**, split by study type as: 108 lab, 78 field‑trial, 78 modeling, 64 review, 15 LCA, 9 survey, 88 unclassified (no rights‑cleared abstract to type from). That shape is the first finding: ERW's evidence base is still weighted toward benchtop and model work, and the field trials that exist mostly measure *agronomy and safety*, not *carbon*. Of the ten most recent field trials the profile returns, the topics tagged are dominated by "environmental‑risks‑cobenefits" — yield, trace metals, soil biology — with only two tagged `mrv-quantification`. A corpus search for ERW papers tagged `mrv-quantification` returned only **three** records. That thinness is itself the story. --- ## 2. What the MRV literature actually says the problems are **(a) There is no agreed quantification method.** "Mass‑Balance MRV for Carbon Dioxide Removal by Enhanced Rock Weathering" (CDRXIV preprint, 2025‑05‑28, doi 10.70212/cdrxiv.2025379.v1) states plainly that reliable, scalable quantification in a medium as complex as soil is hard and that no consensus exists on the best method. It reviews cation‑stock mass balance, offers a software package for uncertainty analysis and trial design, and argues for Bayesian inference so that the answer comes with an explicit uncertainty representation rather than a point estimate. **(b) The signal is small relative to the noise, and the noise can be mined.** "Aggregated monitoring of enhanced weathering on agricultural lands" (CDRXIV, v2 2025‑10‑15, field‑trial, doi 10.70212/cdrxiv.2025394.v2) is the most operationally important MRV paper in the set. Three findings: larger agricultural catchments are at least as good as small ones for detecting a weathering signal in river chemistry, sometimes better because background variability is lower; quantification error from in‑field solid‑phase soil sampling falls when signals are aggregated across many fields; and — the warning that matters for credit integrity — **aggregation sets must be defined in advance with all plots included, because biased selection of fields can manufacture an apparent removal signal out of statistical noise.** That is a direct description of how ERW MRV could be gamed without anyone falsifying a measurement. **(c) The measurement instrument changes the answer.** "A novel soil porewater extraction technique… SATuration–Centrifugation" (Cambridge Prisms: Carbon Technologies, 2026‑01‑01, doi 10.1017/cat.2026.10011) makes the point that alkalinity and dissolved‑ion measurements underpin ERW carbon accounting, but in‑field extraction is constrained by soil moisture, which breaks comparability across sites and seasons. Their SAT‑C method is proposed precisely to decouple porewater sampling from moisture conditions. **(d) Proving the rock actually weathered is a separate problem from proving CO₂ was removed.** "Direct evidence for enhanced mineral weathering in cropland verified by quantitative X‑ray powder diffraction" (Applied Geochemistry, 2026‑05‑09, doi 10.1016/j.apgeochem.2026.106869) is tagged `mrv-quantification` — an indication that as of 2026 the field was still publishing papers establishing *that* dissolution occurred in a field setting. **(e) Counter‑evidence on non‑detection.** A 2026 lysimeter field study (Soil and Tillage Research, 2026‑04‑22) reports **no visible basanite weathering in soil over a nine‑week planted experiment** — a reminder that timescales for a detectable solid‑phase signal can exceed a growing season, which is the interval over which credits are often sampled. **(f) The overestimation result.** The corpus holds, as a metadata‑only closed‑access record with no assigned study type: **"Overlooked Soil Acidity Leads to Significant Overestimation of Carbon Sequestration in Enhanced Rock Weathering"** (Environmental Science & Technology, 2026‑08‑05, doi 10.1021/acs.est.6c05139). *From web search* I confirmed the authorship (Su, Kang, Driscoll, Zhu, Goll, Huang, Wang, Fang) and that it is being cited across the 2026 weathering literature. The mechanism implied by the title — protons from soil acidity, not atmospheric CO₂, consuming part of the released alkalinity — attacks the core assumption of cation‑based mass balance. I could not retrieve the quantified magnitude of the overestimation from the abstract within my search budget; treat the size of the effect as unverified here. **(g) Durability along the land–ocean continuum is unresolved.** *Web only, published after nothing equivalent appears in the corpus's MRV set:* Isson & West, "Can enhanced alkalinity store carbon durably?", **Science 392, 808–810 (22 May 2026)**, doi 10.1126/science.adz2620 — a Perspective arguing that key questions remain about removal strategies operating in open environmental systems. Also *web only*: Schiedung, Harrington, Dupla et al., **"Uncertainties of enhanced rock weathering for climate‑change mitigation", Nature Reviews Earth & Environment 7, 253–266 (2026)**, doi 10.1038/s43017-026-00761-7, which organises the uncertainties into feedstock availability (constrained by ultramafic rock composition and toxic‑element content), plant–soil system impacts, **CDR efficiency along the land–ocean continuum**, and socio‑economics. The land–ocean continuum point is the crux: carbon counted as removed at the field edge can be re‑released by carbonate precipitation downstream, and no field MRV system currently tracks a tonne that far. **(h) Carbon accounting is entangled with soil organic carbon.** "Enhanced Rock Weathering Affects Formation of Mineral‑Associated Organic Carbon in Soil" (KITopen, 2026‑08‑11, tagged `mrv-quantification`) points at an interaction that complicates attributing measured changes to inorganic weathering alone. **(i) What the field trials are good at.** By contrast, the co‑benefit and safety evidence is firming up: a 2‑year corn–soy strip trial in the southeastern US reports **>20% yield benefit** and grower economics above $250/ha (soy) and $750/ha (corn) with subsidised rock (CDRXIV, 2026‑08‑09); a Vermont dairy trial applying 20 t/ha Fe–Al‑rich basalt found **no increase in extractable Ni, Cr, Pb, Al, Mn, Zn, Cu or Cd** versus controls (CDRXIV, 2026‑03‑11); and an 11.5‑hectare reforestation trial monitoring 6,400 trees over four years found silicate amendment raised aboveground carbon stocks 27% in broadleaf stands (Communications Sustainability, 2026‑06‑30). **The asymmetry is the point: agronomic efficacy is being demonstrated faster than carbon quantification.** --- ## 3. Who is selling ERW removals anyway `aggregate_records` over the whole ERW deal layer — exact over what is held, but a **lower bound**, not a census: - **35 deals**, **4,592,794 tonnes contracted**, spanning 2022 → 2026‑05‑27 - **6 of 35 deals disclose a value**: $117.8M total, covering **328,161 of 4.59M tonnes (7%)** - **27 of 35 records are `review_status: auto`** — machine‑extracted from the cited announcement with no human spot‑check | Supplier | Deals | Tonnes contracted | Value disclosed | |---|---|---|---| | Lithos Carbon | 4 | 3,669,280 | $57.1M on 154,880 t | | Canadian Wollastonite | 1 | 450,000 | – | | Terradot | 3 | 289,998 | $27.0M on 89,998 t | | Eion | 3 | 84,392 | $33.0M on 80,892 t | | UNDO | 8 | 33,043 | – | | InPlanet | 4 | 31,491 | – | | Alt Carbon | 4 | 21,417 | $0.5M on 1,851 t | | Mati Carbon | 3 | 7,013 | – | | Cella | 1 | 3,212 | – | | Flux | 2 | 1,682 | $0.2M on 540 t | | Silica | 1 | 1,266 | – | | ZERO (ZeroEx) | 1 | no tonnage | – | **Two data caveats I'd flag before anyone quotes that table.** First, the Lithos bucket is dominated by a single very large contract; the tonnage is not evenly distributed and one record drives 80% of the pathway total. Second, the 450,000 t "UNDO ↔ Canadian Wollastonite" row is a **feedstock supply agreement** — UNDO buying rock from a Kingston, Ontario quarry over three years — not a credit purchase. It sits in the deal layer with an ERW pathway tag and materially inflates the pathway total. Roughly 4.1M of the 4.59M tonnes is attributable to two rows of unusual character. **Buyers on record:** Microsoft (UNDO 15,000 t 2024, Lithos 11,400 t, InPlanet 28,500 t Dec 2025, Terradot), Frontier (Terradot 89,998 t / $27M, Eion 80,892 t / $33M, Alt Carbon ×2, Mati ×2, Flux, Silica, Cella), Google (Terradot 200,000 t, Dec 2024, its largest CDR purchase at the time), Shopify (Mati 5,000 t), Barclays (UNDO 6,538 t), British Airways (UNDO), Mitsui O.S.K. Lines (Alt Carbon 10,000 t), NextGen CDR (Alt Carbon), AXA Schweiz, Adyen, Auto Trader (UNDO 340 t), Bain & Company (ZeroEx), Carbonx (Lithos 3,000 t), Milkywire (Flux 540 t), and Perdue AgriBusiness (Eion 3,500 t — an **insetting** deal that issues no credits at all). **Prices — the disclosing minority only.** Three per‑tonne figures are derivable by division, not quoted: Frontier–Eion **$419/t**, Frontier–Terradot **$300/t**, Milkywire–Flux **$370/t** (traded on Kenya's CYNK platform). NextGen's Alt Carbon announcement stated an average of **~$200/t**. A trade‑press source in the Microsoft–Terradot record cites current ERW at **$300–400/t with projections toward $25/t at scale** — a supplier‑side projection, not a market observation. With 93% of contracted tonnes carrying no disclosed value, none of this supports a price trend. **How MRV is being handled commercially:** Isometric's Enhanced Weathering in Agriculture protocol dominates (InPlanet, Alt Carbon, Mati), with Puro.earth used by UNDO and Mati; Alt Carbon's early Frontier deal relied on **in‑house MRV** with IISc Bangalore scientists; Microsoft's UNDO follow‑on explicitly funded MRV research at Guelph, Queen's and Newcastle; InPlanet and Terradot share data through **Cascade Climate's ERW Data Quarry**. Buyers are, in effect, funding the measurement science with the same cheque that buys the tonnes. --- ## 4. What web search added beyond the corpus - **Deliveries are three orders of magnitude below contracts.** Carbon Direct's 2026 VCM trends piece (Feb 2026) reports ERW and OAE issued their **first credits in 2025, roughly 12,000 tonnes combined**, against the ~4.6M tonnes contracted in the table above. Lithos issued 5,160 registry‑certified tonnes in Dec 2025 (then a record, ~7× the prior largest); Alt Carbon's 9,566 t issuance to Frontier in May 2026 — which *is* in the corpus — superseded it. - **A Microsoft–Terradot deal the corpus appears to be missing:** Data Center Dynamics reported (~31 March 2026) a **12,000‑tonne agreement for delivery 2026–2029**, described as Microsoft's fourth ERW contract. The corpus holds an earlier, untonned Microsoft–Terradot record dated 2025‑08‑14. Treat this as a genuine coverage gap. - **Climeworks–Lithos Carbon partnership** announced ~25 March 2026 (Carbon Herald) to bring verified ERW credits to market — a DAC company distributing an ERW supplier's credits. Not in the corpus. - **Protocol state:** Isometric certified an updated Enhanced Weathering Protocol in **January 2025** after a 30‑day public consultation; the live registry version is **Enhanced Weathering in Agriculture v1.1**, and Isometric's Spring 2026 update extended its GHG Accounting Module across EW, mineralisation, OAE and RAE. The corpus record for this protocol carries a "confirm current version at review" note — v1.1 is the answer. - **EU policy — the binding constraint.** **Delegated Regulation (EU) 2026/285** sets CRCF certification methodologies for DACCS, BioCCS and biochar. **ERW is not among them.** The Commission's own pages list ERW as a pathway being *explored*, and an ISCC state‑of‑play deck (May 2026) shows work merely *started* on ERW and mineralisation methodologies. So ERW credits sold into Europe today rest entirely on private registries (Isometric, Puro), not on the EU framework. --- ## 5. Where the two disagree, and what neither covers The corpus and the web don't contradict each other so much as sit at different points on the same curve: the corpus's deal layer shows millions of contracted tonnes, while web sources show issuance in the low thousands. Both are true; the gap between them is the MRV problem expressed in tonnes. **Not covered by either:** delivered‑vs‑contracted reconciliation per supplier (CDR.fyi is the fuller scoreboard); any deal or paper after 2026‑09‑02; non‑English literature; the quantified magnitude of the ES&T soil‑acidity overestimation, which I could not verify; whether the Su et al. finding has been rebutted; supplier financial health and delivery track records; and any ERW activity in jurisdictions outside the 28 the corpus tracks. Two of the corpus's own ERW price points are derived by division from press releases, and 27 of 35 deal records have never been human‑checked. I would not build a purchasing decision on the price column.
Judge’s note — Coverage of the gold items is strong on the numbers but partial on their framing: the answer gives the ~440 in-scope ERW paper count with a full type breakdown and the 78 field trials, but never states the ~1,582 works fetched nor that ERW is held as a complete result set, and never makes the cross-pathway point that 78 is the largest field-trial count of any pathway. Deal coverage is excellent and exceeds the checklist — Frontier–Eion is given with tonnage, value and implied $419/t; Lithos deals are itemised (though the Frontier–Lithos pairing specifically is not called out, Frontier's 11-deal book is quantified). The registry/protocol layer is well handled (Isometric EW in Agriculture, Puro.earth 2022→2025 edition transition, Rainbow Standard ICVCM eligibility, Cascade's Data Quarry). On the beyond-corpus probe, the answer explicitly states no web tool was available, so nothing is verified live; however it does surface a substantial set of specific, named, recent MRV criticisms — the ES&T soil-acidity overestimation paper, the CDRXIV spatial-uncertainty framework undermining element–element mixing models, the ES&T Letters cation-mobility and MAOM-C papers, the prospective LCA's median 11-year lag, the PNAS 'farming practice not CDR' framing, and the counterfactual/stoichiometry reframing preprint — plus issuance-versus-contracted delivery data (Alt Carbon 9,566 t, Lithos 5,160 t, Mombak 46 credits) and 2026 policy items (CRCF carbon-farming methodologies excluding ERW, EU ETS 250 Mt proposal). Nothing is clearly dated after the stated ingest cutoffs, and the answer asserts slightly later cutoffs than the probe specifies. Citation quality is exemplary: journal, date and DOI for nearly every claim, with explicit flags where only title/venue is held. Honesty is also exemplary — it declares the missing web tool, corpus floors, paywalled titles-only papers, misclassified deal rows (UNDO–Canadian Wollastonite feedstock supply, Frontier–Cella injection), and labels the $25/t projection a company claim. Structure is crisp and the closing contracted-versus-delivered 100:1 framing is genuinely decision-useful. Docked mainly for the missing corpus-completeness/superlative framing and the absence of any live verification. only weakness beyond that is that several DOIs and figures cannot be independently confirmed from within the answer.
I'll start by mapping the corpus coverage and pulling the full ERW evidence profile. Now let me pull the MRV-specific literature and the full supplier/buyer breakdown. ## Bottom line The ERW literature has converged on a fairly brutal consensus: **the pathway works, but nobody can yet measure how much of it works at a given field, in a given year, with the precision a tonne-denominated credit implies.** Meanwhile roughly a dozen companies have contracted at least 4.6 MtCO₂ of ERW removals — and delivered, verifiably, something on the order of tens of thousands of tonnes. An important caveat up front: **I was not able to run web searches in this session** — no web tool was available to me. Everything below comes from the Carbon Curve corpus, which is explicitly a floor rather than a census, with papers fetched to **2026-09-02**, deals to **2026-08-28**, and policy events to **2026-08-17**. Anything after those dates, non-English work, and any unannounced deal is invisible here. Treat the deal tonnages as lower bounds; CDR.fyi and AlliedOffsets are the fuller transaction scoreboards. --- ## 1. What the literature actually says about ERW MRV The corpus holds **440 in-scope ERW papers**: 108 lab, 78 field trials, 78 modeling, 64 reviews, 15 LCAs, 9 surveys, 88 unclassified. That's a real evidence base — but the MRV-tagged subset is dominated by papers documenting why quantification is hard rather than solving it. ### (a) The estimates span orders of magnitude, and the spread is partly methodological, not physical The most useful single synthesis is **"An Ecosystem of Carbon Dioxide Removal Reviews – Part 3: Enhanced Weathering"** (CDRXIV preprint, 2026-07-14, [doi:10.70212/cdrxiv.2026417.v1](https://doi.org/10.70212/cdrxiv.2026417.v1)). It reports a **median area-normalised CDR flux of 0.84 tCO₂/ha/yr, "spanning several orders of magnitude,"** and is explicit that the variance "reflects not only context-dependent differences in EW performance, but also **the diversity of quantification approaches, which differ by use case and vary in terms of their system boundaries and treatment of loss processes.**" Its global technical potential — **~0.2–2 GtCO₂/yr** — is a marked downgrade on the earlier modelling in the corpus: Strefler et al. (ERL 2018, [doi:10.1088/1748-9326/aaa9c4](https://doi.org/10.1088/1748-9326/aaa9c4)) put cropland potential at 4.9 Gt/yr for basalt and 95 Gt/yr for dunite, and Fuss et al. (ERL 2018) gave 2–4 Gt/yr. ### (b) The counterfactual problem: strong acids vs carbonic acid This is the live controversy and the one with the most direct crediting consequence. If crushed basalt is dissolved by nitric or sulfuric acid from fertiliser rather than by carbonic acid from atmospheric CO₂, the cations mobilise, the geochemical signals look identical — and **no atmospheric CO₂ has been removed.** - **"Overlooked Soil Acidity Leads to Significant Overestimation of Carbon Sequestration in Enhanced Rock Weathering,"** *Environmental Science & Technology*, 2026-08-05 ([doi:10.1021/acs.est.6c05139](https://doi.org/10.1021/acs.est.6c05139)). The corpus holds title and venue only (closed access, no abstract) — the title is the finding. - **"Reframing Carbon Accounting in Enhanced Rock Weathering: Counterfactuals, stoichiometry, and MRV implications"** (2026-07-15, [doi:10.13140/rg.2.2.15013.26080](https://doi.org/10.13140/rg.2.2.15013.26080)). - A signal in the corpus reports a *Science of the Total Environment* study finding that in acidic tropical soils, **part of basalt weathering is driven by naturally occurring soil acids rather than carbonic acid** — with researchers warning that reliable verification methods are needed "particularly if farmers are to receive carbon credits." This matters because tropical acidic soils (Brazil, India) are exactly where the largest commercial deployments sit. ### (c) Spatial heterogeneity may defeat the industry's dominant measurement method **"A Framework for Integrating Spatial Uncertainty into Solid-Phase Measurement and Verification of Enhanced Weathering"** (CDRXIV, 2025-03-12, [doi:10.70212/cdrxiv.2025334.v1](https://doi.org/10.70212/cdrxiv.2025334.v1)) concludes that "field-scale variance in baseline soil concentrations must be accurately estimated before designing a sampling plan and, even then, **such variance is likely too high for element–element mixing models to be effective near-term constraints on enhanced weathering.**" Immobile-element mass balance is the workhorse of commercial ERW MRV; this paper says the soil is too noisy for it right now. ### (d) Loss processes after the initial capture are poorly constrained The Part 3 review names the open items: **secondary phase (carbonate/clay) formation and cation exchange**, with "significant uncertainty… regarding the magnitude, persistence, and timing of loss processes after initial CDR has occurred." Corroborating: **"Soil Carbon Dioxide Removal Estimates from Enhanced Weathering Deployments in the Southeastern United States Are Likely Tied to Cation Mobility"** (*ES&T Letters*, 2026-04-22, [doi:10.1021/acs.estlett.5c01172](https://doi.org/10.1021/acs.estlett.5c01172)). ### (e) ERW perturbs the organic carbon pool it is being measured against **"Enhanced Rock Weathering Affects Formation of Mineral-Associated Organic Carbon in Soil"** (*ES&T Letters*, 2026-07-25, [doi:10.1021/acs.estlett.6c00633](https://doi.org/10.1021/acs.estlett.6c00633)) found fresh olivine-rich basalt caused **~13% loss of native MAOM-C** (0.86 ± 0.29 mg C/g soil) in a 6-month Cambisol incubation, while weathered basalt **increased** new straw-derived MAOM-C by 20%. The authors state plainly this "implies the need to include MAOM-C responses in ERW carbon accounting." A companion KIT record appeared 2026-08-11. A *Sustainability* systematic review (2026-07-09, [doi:10.3390/su18147011](https://doi.org/10.3390/su18147011)) adds that ERW studies sample too shallow — it argues for **>60 cm depths** and measurement of *both* SOC and SIC, noting studies "only provided selective coverage." ### (f) Timing: credits are sold years before net removal occurs **"Delayed climate benefits and toxicity risks could hinder the sustainable deployment of enhanced weathering"** (prospective LCA, CDRXIV, 2026-08-03, [doi:10.70212/cdrxiv.2026531.v2](https://doi.org/10.70212/cdrxiv.2026531.v2)) finds a **median 11-year lag between rock application and the onset of net CDR** for inland scenarios starting in 2030, driven by the up-front supply-chain emissions of mining, crushing, hauling and spreading. It concludes "commercialization of CDR credits should be preceded by a thorough evaluation of near-term net climate benefits." Related: **"Bridging carbon removal time lags in agricultural enhanced weathering via methane emission reduction"** (*ERL*, 2026-05-29) states the radiative-forcing lag "complicates using the practice to offset carbon emissions." ### (g) MRV protocols and LCA are not saying the same thing **"Life Cycle Assessment in the MRV of Land-Based Carbon Dioxide Removal: Gaps and Opportunities"** (*ES&T*, 2025-03-27, [doi:10.1021/acs.est.4c09510](https://doi.org/10.1021/acs.est.4c09510), 26 citations) compares registry MRV protocols against the LCA literature for BECCS, biochar, ERW and AR, and finds divergence on **baselines, additionality, uncertainty treatment, multifunctionality, environmental safeguards, emission factors and "overlooked carbon pools."** ### (h) MRV has to police co-risks too, and the feedstock hierarchy is unsettled - **ERW Metal Accumulation Calculator** (CDRXIV, 2026-07-29, [doi:10.70212/cdrxiv.2026546.v1](https://doi.org/10.70212/cdrxiv.2026546.v1)): under basalt deployment, modelled soil metal concentrations stay below most regulatory thresholds though 8 metals can rise above baseline; under **peridotite, Ni and Cr often exceed the median regulatory threshold**, with 7–11× the Ni/Cr intensity of basalt. - The 2026-08-03 LCA **contradicts the conventional wisdom** that basalt is the safe feedstock, arguing basalt-derived **zinc** may pose a greater long-run toxicological risk than dunite-derived nickel, and recommending deployment "prioritize non-agricultural areas until the risks of food chain contamination are better understood." - Countervailing field evidence: a Vermont dairy trial at 20 t/ha (CDRXIV, 2026-03-11) found **no increase** in extractable Ni, Cr, Pb, Al, Mn, Zn, Cu or Cd versus controls, and no change in forage yield or tissue metals. ### (i) The sharpest framing, and the encouraging counterweight The most pointed paper in the window is **"Enhanced rock weathering has greater promise as a sustainable farming practice than a CO₂ removal technology,"** *PNAS*, 2026-06-24 ([doi:10.1073/pnas.2602421123](https://doi.org/10.1073/pnas.2602421123)). The corpus holds title and venue only — I have not read the argument, and you should before citing it. That framing is well supported on the agronomy side: a 2-year corn-soy strip trial in the southeastern US reports **>20% yield benefit** and grower gains of >$250/ha (soy) and >$750/ha (corn) if rock is subsidised (CDRXIV, 2026-08-09); an 11.5-hectare, 6,400-tree, four-year reforestation trial found silicate amendment raised aboveground carbon **27% in broadleaf stands** (*Communications Sustainability*, 2026-06-30). And genuine MRV tooling is arriving — direct XRD verification of cropland weathering (*Applied Geochemistry*, 2026-05-09) and the SAT-C porewater extraction method (*Cambridge Prisms: Carbon Technologies*, 2026). --- ## 2. Who is selling ERW removals anyway The corpus holds **35 ERW deals, 4,592,794 tCO₂ contracted, spanning 2022 to 2026-05-27 — with only 6 disclosing a value ($117.8M across 328,161 t).** Two records need flagging before you read the table: the 450,000 t "UNDO ↔ Canadian Wollastonite" row is a **three-year rock-feedstock supply agreement**, not a credit sale; and the 3,212 t Frontier ↔ Cella row is **basalt injection mineralisation in Kenya**, tagged erw-ag but not agricultural ERW. Strip those and the genuinely agricultural contracted volume is ~4.14 Mt, of which 3.5 Mt is one 2023 Climeworks–Lithos deal. | Supplier | Deals | Tonnes contracted | Geography | Notable buyers | |---|---|---|---|---| | **Lithos Carbon** (US) | 4 | 3,669,280 | US croplands | Climeworks (3.5 Mt, 2023, decade-long), Microsoft (11,400 t), Carbonx (3,000 t) | | **Terradot** (Brazil) | 3 | 289,998 | Brazil | Google (200,000 t, Dec 2024 — Google's largest-ever CDR deal), Frontier (89,998 t / $27M / **$300/t**), Microsoft | | **Eion** (US, Princeton) | 3 | 84,392 | Mid-Atlantic, Southern/Midwestern US; olivine | Frontier (80,892 t / $33M / **$419/t**, delivery 2027-30), Perdue AgriBusiness (3,500 t **inset**, no credits issued), New Jersey Rocks! ($5k community campaign) | | **UNDO** (UK) | 8 | 33,043 | UK, Ontario, Québec | Microsoft (15,000 t follow-on 2024 + earlier 5,000 t), Barclays (6,538 t), McLaren Racing (6,165 t), Auto Trader (340 t), British Airways (×2) | | **InPlanet** (Germany/Brazil) | 4 | 31,491 | Brazil, >12,000 ha | Microsoft (28,500 t, Dec 2025, delivery 2026-28), AXA Schweiz (1,950 t), Adyen | | **Alt Carbon** (India) | 4 | 21,417 | Darjeeling/Terai tea estates, rice paddies | Frontier (9,566 t May 2026 + 1,851 t), Mitsui O.S.K. Lines (10,000 t), NextGen CDR | | **Mati Carbon** (India/Zambia/Tanzania) | 3 | 7,013 | smallholder farms, Global South | Frontier (420 t + 1,593 t), Shopify (5,000 t) | | **Flux** (Kenya) | 2 | 1,682 | Kenyan Rift Valley sugarcane | Milkywire (540 t @ **$370/t** via CYNK), Frontier (1,142 t) | | **Silica** | 1 | 1,266 | — | Frontier | | **ZeroEx / ZERO** (Munich) | 1 | undisclosed | — | Bain & Company (intent, Oct 2025) | | **Mombak** (Brazil) | — | first issuance **46 credits** (Project Cajueiro, Apr 2026) | Brazil | three EW projects certified by Isometric | **Buyers, ranked:** Frontier (11 deals, 345,861 t, $117.6M disclosed — by far the deepest ERW book); Microsoft (6 deals, 59,900 t); Climeworks (3.5 Mt as offtaker/reseller — described in one source as "the single largest offtaker in the ERW credits category"); Google (200,000 t); then a long tail of Barclays, Mitsui O.S.K. Lines, British Airways, McLaren, Bain, Adyen, AXA, Auto Trader, Perdue, Shopify, Milkywire, NextGen CDR (South Pole/Mitsubishi JV), Carbonx. **Prices, from the disclosing minority only:** $419/t (Frontier–Eion), $370/t (Milkywire–Flux), $300/t (Frontier–Terradot), ~$200/t stated average (NextGen–Alt Carbon). One trade press source cites a current range of **$300–400/t with projections to $25/t at scale** — treat that projection as a company claim. ### The verification stack these sales run on - **Isometric Enhanced Weathering in Agriculture Protocol** (UK registry) — the dominant certifier: InPlanet's January 2025 issuance was billed as the world's first independently verified ERW credits; Alt Carbon, Terradot and Mombak all sit on it. - **Puro.earth ERW methodology** — moving from Edition 2022 to Edition 2025; the transition deadline for legacy facility audit packages was **18 February 2026**. - **Rainbow Standard** — ICVCM CCP-eligible as of 2026-03-05, with an ERW methodology among its suite. - **Cascade Climate's ERW "Data Quarry"** — the industry's cross-project data-sharing system, used by InPlanet and Terradot. - **Carbon Direct / Microsoft, Criteria for High-Quality CDR (2023 edition)** — the buyer-side screen. --- ## 3. The gap that matters Three numbers, side by side: 1. **~4.6 Mt contracted** in this corpus (a floor). 2. **Verified issuances measured in tens of thousands of tonnes.** The largest single ERW issuance on record here is Alt Carbon's **9,566 t** (May 2026); Lithos's largest delivery was **5,160 t**; Mombak's first issuance was **46 credits**. That is roughly a **100:1 contracted-to-delivered ratio.** 3. **A literature that, as of August 2026, is still arguing about whether the acid doing the weathering came from the atmosphere.** Policy has noticed. The EU's **CRCF** adopted permanent-removal methodologies (BioCCS, DACCS, biochar) and, on 2026-07-10, three carbon-farming ones (mineral soils, peatland rewetting, afforestation) — **ERW is not among them**; the Commission was still running its *first technical workshop* on ERW and OAE certification as of late 2025. The **EU ETS review proposal** published 2026-07-17 proposes central purchasing of 250 Mt of CRCF-certified permanent removals over 2031-40, and names **DACCS and BioCCS**, not ERW. ERW's public support is still concentrated in R&D and prize instruments — UK's GGR-D Demonstrator and DAC/GGR Innovation Programme, DOE's **FOA 3082 Carbon Negative Shot Pilots** (Eion $3.99M, Lithos $3.98M, Stanford, Northwestern) and the DOE CDR Purchase Pilot Prize. Read charitably, the sellers' position is defensible: the buyers are almost entirely sophisticated advance-market-commitment purchasers (Frontier, Microsoft, Google) who explicitly fund MRV science as part of the deal — Microsoft's 2024 UNDO contract funded field trials at Guelph, Queen's and Newcastle; its Terradot arrangement is described as direct support for field trials, lab work and data infrastructure "not just purchasing carbon credits." That is R&D procurement wearing a credit's clothes. Read less charitably, ~4.6 Mt of tonne-denominated obligations have been written against a quantification method that a 2025 uncertainty analysis says is too noisy to work yet, and against a stoichiometry that a 2026 *ES&T* paper says is systematically overestimated. **What I could not check:** anything published or announced after 2026-09-02 (papers) / 2026-08-28 (deals). I would specifically want to read the full text of the *PNAS* piece (2026-06-24) and the *ES&T* soil-acidity paper (2026-08-05) — both are behind paywalls in this corpus and I have only their titles — and to verify current issued-credit totals directly on the Isometric and Puro.earth registries, which is the only way to get the delivered-versus-contracted ratio right.
Asked, verbatim: “I'm diligencing Vaulted Deep. What do we know about their deals, and how does their pathway's evidence base look?”
It found every checklist fact on this question — not every question favors the database, and this page reports those too.
Judge’s note — All three gold items are surfaced substantively and accurately: the Frontier May 2024 offtake with exact volume/value (plus a derived ~$382/t), the Microsoft 4.9 Mt/2038 deal correctly framed as the pathway's largest-of-record at announcement, and a clear description of slurry biomass injection beneath caprock with 10,000+ year durability claims and Isometric's Biomass Geological Storage protocol. Well beyond the corpus, it adds: the Advantek spinout origin, named founders/executives, the $32.3M Series A (Prelude/Lowercarbon, Nov 2024) with candid reconciliation of conflicting totals, XPRIZE $8M second runner-up with 9,986 t demonstration tonnes, the 2023 Frontier prepurchase (1,666 t delivered), the Google 50 kt/2030 deal and the associated Isometric avoided-methane research program, delivery track record (>23 kt verified in 2025), verifier identity (350Solutions) and monthly issuance cadence, Hutchinson/LA site detail with tipping-fee dual revenue, and 2026 awards — roughly a dozen distinct findings. Analysis is genuinely expert-grade: the counterfactual/baseline-decay dependency is correctly identified as the load-bearing assumption, avoidance-vs-removal is flagged, the LA disposal-fee reduction is used as a concrete induced-demand example, and the ~17x scale-up gap between 2025 actuals and contracted volumes is quantified. It also raises an original, well-labelled analytical flag (in-reservoir methanogenesis of biodegradable organics) and separates it from sourced findings. Citations are named with dates and outlets but lack links, and some 2026-dated trade-press items are not independently checkable; a few numbers (PitchBook/Tracxn totals) are self-flagged as inconsistent, which is handled honestly rather than papered over. Coverage honesty is exemplary — an explicit search-quota caveat, a dedicated gaps section covering unverified cumulative registry tonnage, undisclosed Microsoft/Google prices, missing UIC permit and buffer-pool/reversal detail, and absence of an independent peer-reviewed LCA. Structure is clean, with a deal table, an evidence-strength/evidence-weakness split, and a bottom line that names the single most likely repricing trigger. Deductions only for unlinked sourcing and mild reliance on unverifiable secondary trade reporting.
I'll research Vaulted Deep's deals and the evidence base for their pathway.
**Date of research: 2 September 2026.** Note upfront: my searches returned material through roughly mid‑2026, and I hit my search quota before I could sweep H1‑2026 news, registry totals, or permit filings. Treat the "current state" numbers below as at least a quarter stale, and see the "gaps" section for what I could *not* verify.
---
## 1. Company snapshot
- Spun out of **Advantek Waste Management Services** in 2023; Advantek was a ~10‑year‑old industrial waste business that pumped oilfield waste into deep disposal wells (Frontier, May 2024). Founders: **Julia Reichelstein** (CEO, ex‑climate VC) and **Omar Abou‑Sayed** (executive chairman, Advantek founder).
- Two operating sites: **Hutchinson, Kansas** ("Great Plains Organic Waste Sequestration" on the Isometric registry) and **Los Angeles**, where Advantek has taken City sewage sludge since 2008 (Waste360, June 2025; Waste Dive, March 2026).
- Feedstocks: biosolids, manure, paper‑mill sludge, food/ag residues — wet, "sludgy" streams that are otherwise landfilled, incinerated or land‑applied.
- Funding: seed led by **Lowercarbon**; **$32.3M Series A** closed 2024, announced 14 Nov 2024, led by **Prelude Ventures** with Lowercarbon, Earthshot Ventures, WovenEarth, Fall Line Capital, Rethink Impact (PRNewswire). Waste Dive (Mar 2026) puts total at **~$56M equity + awards**; PitchBook shows $48M; Tracxn shows $32.3M and ~60 employees as of May 2026. *These three numbers disagree — reconcile with the cap table.*
- **$8M XPRIZE Carbon Removal** second‑runner‑up prize, April 2025, having delivered **9,986 net tonnes** during the demonstration period.
## 2. The deal book
| Counterparty | Date | Volume | Value / price | Notes |
|---|---|---|---|---|
| Frontier (prepurchase) | Sept 2023 | 1,666 t | undisclosed | Fully delivered and verified |
| **Frontier (offtake)** | 1 May 2024 | **152,480 t**, 2024–2027 | **$58.3M** → **~$382/t implied** | Frontier's largest offtake at the time; 18,321 t slated for 2024; funded three new wells; includes options on future tons **at lower prices** |
| **Microsoft** | 17 July 2025 | **up to 4.9 Mt** over 12 yrs, through **2038** | undisclosed | Billed as the second‑largest CDR deal on record at announcement |
| **Google** | 17 Sept 2025 | **50,000 t by 2030** | undisclosed | Isometric‑certified; bundled with a **research program with Isometric to quantify avoided methane**; follow‑on to Google's 2024 exposure via Frontier |
| Waste customers | ongoing | n/a | tipping fees | Dual revenue: "finances its infrastructure through tipping fees and carbon removal credits" (company statement, 2026) |
Delivery track record: >2,000 t in the first four months at Hutchinson (April 2024); ~5,000 t in year one; **>23,000 t of verified removal in 2025** (company/Inc., Dec 2025). Monthly issuance cadence on Isometric, verified by **350Solutions**, under the **Biomass Geological Storage protocol v1.1**.
**Commercial read‑through.** The Frontier contract is the only one with a public price, and $382/t is a real, high‑quality‑market price — but Frontier explicitly bought options on future tons *cheaper*, so that number is a ceiling, not a run‑rate. The Microsoft and Google deals are "up to" volumes over long horizons, i.e. optionality contingent on Vaulted permitting and building new sites; they are not bankable revenue in the way a fixed take‑or‑pay would be. Buyer concentration is extreme: essentially Frontier + Microsoft + Google.
**Scale math to stress‑test.** 23,000 t in 2025 versus 152,480 t owed to Frontier by 2027 and up to 4.9 Mt by 2038 (≈400 kt/yr average). That implies a **~17x step‑up on the Microsoft contract alone**, gated on new wells, UIC permits, feedstock contracts and trucking logistics. Ask for the site pipeline with permit status and dates, and for the Frontier delivery schedule versus actuals to date.
## 3. How the pathway's evidence base looks
**Strong relative to most CDR:**
1. **Storage physics are borrowed, not invented.** Slurry injection beneath impermeable caprock is mature oilfield practice; Advantek/LA has run since 2008. Claimed durability is **10,000+ years**. This is the least speculative part of the story.
2. **MRV is mass‑balance, not modelled.** Carbon is *weighed* in the biomass, then lifecycle emissions (transport, energy) are subtracted (Frontier). That's a far shorter inference chain than enhanced weathering or ocean alkalinity, and it's why Vaulted could issue monthly credits from December 2023 onward.
3. **Low energy intensity** — company claims **<0.14 GJ per tonne CO₂ removed**, because the waste is injected with minimal processing rather than converted to pure CO₂.
4. **Third‑party scrutiny exists**: peer‑reviewed Isometric protocol (published April 2024), independent verifier, public per‑delivery records on the registry, plus XPRIZE's four‑year judging process.
5. **Regulatory tailwind on the waste side.** PFAS is pushing states to restrict land application of biosolids; a municipal customer quoted by Waste360 cited odour and weather constraints and called cavern disposal a more controlled alternative. Tipping fees mean the business is not purely dependent on voluntary carbon demand.
**Where the evidence is thin — and this is where I'd concentrate diligence:**
1. **The counterfactual is the whole credit.** Nothing is captured from the air by Vaulted; the credit is the *difference* between injection and the assumed alternative fate. Canary Media noted the Isometric protocol frames waste biomass counterfactually — i.e. that unstored material would release CO₂. The decay fraction, decay timescale, and how much land‑applied carbon actually persists in soil are **modelled, not measured**. Small changes in baseline assumptions move net tonnes materially.
2. **Methane quantification is openly unresolved.** The Google/Isometric work exists precisely because the avoided‑methane benefit isn't yet rigorously quantified — Vaulted's own framing was that it knows the process prevents methane and now needs to "prove how much." Two implications: (a) a large slice of the claimed climate benefit is **avoidance, not removal**, and should be priced and disclosed separately; (b) the methodology could move in either direction when the research lands.
3. **System‑level additionality risk.** Heatmap (Feb 2025) captured the concern well: as BiCRS scales and a market for waste biomass emerges, do the "it would have been incinerated/land‑applied" counterfactuals still hold? A related, concrete example in the same reporting: because of the Frontier deal, Vaulted **dropped its disposal fee to LA** so the city would divert more waste — good for volume, but it means the carbon revenue is subsidising diversion, which is exactly the kind of induced‑demand dynamic that baseline rules struggle with. Frontier says its technical team vets every delivery against a "higher and better use" test.
4. **In‑reservoir behaviour of organic solids.** This is my analytical flag, not a sourced finding: injecting *biodegradable* organics into a warm anaerobic formation is not the same as injecting inert drill cuttings. I found no public evidence addressing subsurface methanogenesis, gas generation and pressure build‑up, or long‑term fate monitoring of injected organic carbon. Ask for the protocol's treatment of in‑situ decomposition, any downhole gas/pressure data, and whether "10,000 years" is asserted by geological analogy or supported by site‑specific modelling.
5. **Well integrity, seismicity and permitting.** Public materials describe pipe‑in‑pipe wells with cement layers and an "exemplary safety record" since 2008, but I could not verify UIC permit classes, permitted injection volumes, seismic monitoring arrangements, or state regulator correspondence in Kansas or California. Deep‑well injection is a politically sensitive permitting category, and the buildout thesis depends entirely on repeatable permitting.
6. **PFAS cuts both ways.** Keeping PFAS out of soil and water is the marketing case; permanently emplacing PFAS‑bearing sludge underground is a long‑tail liability and a plausible future regulatory target. Ask about indemnities and post‑closure liability allocation.
7. **Related‑party structure.** Vaulted "partners with Advantek" at the LA facility and books carbon revenue off those contracts (Waste Dive, Mar 2026). Diligence the commercial terms, IP licence for the patented slurry technology, and whether the LA carbon economics survive an arm's‑length restatement.
## 4. Explicit gaps in what I could establish
- **No verified total delivered tonnage as of Sept 2026** — the last hard figure I have is >23,000 t verified in 2025 plus 9,986 t in the XPRIZE window. Pull the Isometric registry supplier page for cumulative issuances and retirements.
- **No 2026 news coverage** beyond award announcements (CNBC Disruptor 50 2026; Fast Company #3 most innovative in North America, March 2026; Wichita Business Journal). Awards are marketing signal, not evidence.
- **No disclosed prices** for Microsoft or Google, so unit economics and gross margin are unknown; likewise the tipping‑fee/credit revenue split.
- **Conflicting basics**: HQ (Houston in press releases vs San Francisco in Tracxn), total raised ($32.3M / $48M / $56M), and whether Frontier's 2024 tranche of 18,321 t landed on schedule.
- **No peer‑reviewed, independent LCA** of this specific pathway surfaced — the protocol is peer‑reviewed, but I found no academic literature independently testing the baseline decay and methane assumptions.
- **No public information** on insurance, buffer‑pool/reversal provisions, or delivery‑shortfall remedies in the offtakes.
## 5. Bottom line
The storage side of Vaulted's story is unusually solid for CDR — real tonnes, monthly third‑party verification, mature injection technology, and a second revenue line from waste tipping fees that most CDR companies lack. The contracted book (Frontier $58.3M at ~$382/t, Microsoft up to 4.9 Mt to 2038, Google 50 kt to 2030) is among the strongest in the industry on paper. The risk is not "does the carbon stay down" but **"how many tonnes does the counterfactual actually entitle them to claim"** — baseline decay and avoided methane are the load‑bearing assumptions, they are model‑based, Google and Isometric are actively funding work to firm them up, and a methodology revision is the single most plausible way this asset gets repriced. Second‑order risks are execution: a ~17x volume scale‑up gated on repeatable deep‑well permitting, three‑buyer concentration, and an unpriced Microsoft contract.
Judge’s note — All three gold items are present and substantive: the Frontier 2024 offtake with $58.3M and a derived $382/t (tonnage stated as 154,146 vs gold's 152,480 — a minor discrepancy the answer itself flags as auto-extracted and needing primary-source check), the Microsoft 4.9 Mt deal with the important correction that it is an 'up to' ceiling through 2038 (~380 kt/yr) rather than a firm floor, and clear pathway framing (slurry injection of wet organic waste into deep wells, geological storage, Isometric Biomass Geological Storage protocol, durable). Beyond-corpus content is rich: $32.3M Series A investors, $8M XPRIZE, Google 50k t deal plus the Isometric methane-quantification workstream, Frontier's ~52.1 kt delivered as of Aug 2026, Isometric monthly registry issuances and first-ever issued credits (Lowercarbon, Apr 2024), Great Plains/Hutchinson throughput figures (>90 kt waste, >26 kt removal; >150 kt since Aug 2023), ICVCM CCP approval of the protocol, conflicting cumulative-raise figures (PitchBook ~$48M vs ~$56M), absence of ratings-agency coverage, and the July vs September 2025 Microsoft announcement-date conflict. Analytical value is high: 96% single-buyer concentration, Class V vs Class VI permitting asymmetry, counterfactual/'ineligible biomass' deduction as the core integrity question, methane co-benefit not yet credited, PFAS/feedstock social-licence exposure. The evidence-base verdict — essentially zero peer-reviewed field trials or LCAs on this specific method, with only two tangential papers (coal-seam BiCRS lab study; hydrochar feasibility modelling), so integrity rests on registry protocol plus oilfield injection practice — is exactly the right framing and is honestly labelled as a gap in the literature, not just the corpus. Citations are named and mostly checkable (Isometric, Frontier, Trellis, Canary Media, Waste Dive, Waste Today, Lowercarbon, PitchBook) though without links, and a few figures are hard to verify. Coverage honesty is exemplary: unverified signals labelled as such, PFAS explicitly flagged as unconfirmed, ICVCM date and CDR.fyi ranking marked unverified due to search budget, and missing project-layer records called out. Structure is crisp and decision-useful for an investment memo.
I'll start by pulling what the corpus has on Vaulted Deep and its pathway, then verify against recent public sources.
Now let me pull the pathway evidence base and the project/supply side.
Let me check the supply side and verify against the web.
## Bottom line
Vaulted Deep is a genuinely revenue-generating, registry-verified BiCRS supplier with an unusually strong delivery record for its stage — but its contracted book is ~96% one buyer (Microsoft) on an "up to" basis, and the *peer-reviewed* literature specific to its pathway is close to non-existent. The integrity case rests almost entirely on one registry protocol (Isometric's Biomass Geological Storage) plus decades of oilfield/waste-injection engineering practice, not on published field science.
Below, corpus-held records are marked **[corpus]** (Carbon Curve, deals ingested through 2026-09-01, papers through 2026-09-02) and web-verified items **[web]**.
---
## 1. The deal book **[corpus]**
Three deals on record, 5,104,146 tonnes contracted, one with a disclosed value:
| Announced | Buyer | Tonnes | Value / price | Window | Notes |
|---|---|---|---|---|---|
| 2024-05-01 | **Frontier** (Stripe, Shopify et al.) | 154,146 | $58.3M / **$382/t** (derived, not quoted) | 2024–2027 | Slurry injection into Class V wells, Kansas + Los Angeles. Frontier's portfolio page shows **~52,100 t delivered as of Aug 2026** |
| 2025-09-08 (see date conflict below) | **Microsoft** | 4,900,000 | undisclosed | 12 years | "One of the largest CDR transactions ever signed"; funds new facilities across the US |
| 2025-09-17 | **Google** | 50,000 | undisclosed | by 2030 | Follow-on to Google's March 2024 Frontier-channel purchase; paired with a methane-quantification initiative with Isometric |
Earlier: Vaulted was in **Frontier's Sept 2023 $7M prepurchase cohort** (14 suppliers) **[corpus]**.
Non-dilutive and equity capital on record **[corpus]**: **$32.3M Series A** (Nov 2024, led by Prelude Ventures, with Lowercarbon, Earthshot, WovenEarth super-pro-rata, plus Fall Line and Rethink Impact) and an **$8M XPRIZE Carbon Removal** award from the $100M purse.
### Three things to correct or probe in that table
- **Date conflict on Microsoft.** The corpus dates it 2025-09-08 off an Arnold & Porter advisory note; Isometric's own announcement page and Trellis's coverage are dated **17–21 July 2025**. Most likely the deal was announced mid-July 2025 and the law-firm release followed. Treat July 2025 as the announcement date.
- **"Up to," not firm.** The corpus flags the Microsoft deal `commitment_type: firm` at a flat 4.9 Mt. Isometric's page describes it as delivering **"up to 4.9 million tonnes"** through to **2038**. That is a ceiling over 12–13 years (~380 kt/yr average), contingent on Vaulted building sites that do not yet exist. Do not model it as a floor.
- **Price is one datapoint.** $382/t is derived from the 2024 Frontier deal only. Trellis (July 2025) noted the Microsoft price was undisclosed and referenced the ~$380/t prior benchmark. Frontier's 2024 write-up said Vaulted has "line of sight" to sub-$100/t — an aspiration, not a contracted price. Corpus-wide, only 44 of 385 deals disclose value (1.5% of tonnes), so there is no reliable market comp here.
## 2. Delivery track record — the strongest part of the file
- **[corpus]** Frontier lists ~52,100 of 154,146 t delivered as of Aug 2026 (~34% through roughly two-thirds of the 2024–2027 window).
- **[corpus]** At Series A (Nov 2024): >7,000 t delivered in year one, running at >1,500 t/month.
- **[web]** Isometric's registry shows monthly issuances to Vaulted (e.g. a 2,786-credit issuance dated Aug 2025; a 202-credit delivery to Google in Dec 2024), and Isometric's own case study describes validation, site visit and first-batch verification completed by end-2023 with issuance now on a monthly cadence weeks after injection.
- **[web]** Waste Today reports the Great Plains facility in Hutchinson, Kansas processed **>90,000 t of organic waste** and delivered **>26,000 t of CO₂ removal** in the most recent year after an expansion that tripled capacity; Waste Dive (March 2026) reports **>150,000 t of waste sequestered** at Great Plains since acquisition in Aug 2023.
- **[web]** Lowercarbon documented Isometric's very first-ever issued credits going to Vaulted (2,474 t removed, 2,166 delivered to Stripe and Shopify) in April 2024.
Monthly ex-post issuance with public registry data is a materially better verification posture than most durable-CDR suppliers. **Coverage gap: the corpus holds no *project* records for Vaulted at all** — Hutchinson and LA County are absent from the project layer, so facility-level stage, capacity and permit status must be diligenced directly.
## 3. The pathway evidence base — thin to the point of being an outlier
**[corpus]** `biomass-injection` returns **2 in-scope papers** out of 22,411 held (6,636 CDR-classified):
1. *Stable isotope analysis of biomethane generation in Powder River basin coals using BiCRS + RNG* (Energy Conversion and Management X, 2026-04-10) — a **lab** study, and about **Cowboy Clean Fuels' coal-seam injection**, not slurry injection into disposal wells.
2. *Feasibility of subsurface storage of hydrochar in the Netherlands* (Int. J. Greenhouse Gas Control, 2025-12-05) — modelling/feasibility, unclassified study type.
**Zero field trials. Zero LCAs. Zero papers on Vaulted's actual method.** Compare with ERW, DAC or biochar, where the corpus holds hundreds of studies including field trials. This is the single biggest analytical gap in the file, and it is a gap in the *literature*, not just in the corpus.
What carries the weight instead:
- **[web]** **Isometric Biomass Geological Storage protocol** (v1.0, now v1.1) governs the credits. Its counterfactual is defined narrowly: the deduction is the CO₂ that *would have remained durably stored* in the biomass absent the project ("ineligible biomass"). Canary Media (May 2024) characterised the protocol as treating waste biomass counterfactually — that not storing it results in CO₂ release.
- **[web]** **ICVCM** has approved Isometric's Biomass Geological Storage (v1.0–1.1) and Bio-oil Geological Storage protocols for the **CCP label**, following independent assessment against all ten Core Carbon Principles including permanence and quantification. That is third-party governance validation of the *method*, distinct from published science.
- **[corpus]** Puro.earth's standard also covers `biomass-injection` (Geologically Stored Carbon), and the pathway sits inside **US DOE grant programmes** (FOA 3082 Carbon Negative Shot BiCRS pilots; the up-to-$100M pilot-scale CDR FOA) — public money, but no Vaulted award on record here.
- **[corpus]** EPA **Class VI UIC** is tagged to this pathway, but Vaulted's Frontier-era wells are described as **Class V** — a different, lighter regime for non-hazardous fluid injection. Confirm the well class, permit terms and state primacy for every current and planned site; this is a real regulatory-asymmetry question that CO₂-injection precedent does not answer.
## 4. Analytical risks I'd put in the memo
1. **Buyer concentration.** Microsoft is ~96% of contracted tonnes. A corpus **signal** (uncorroborated, from a Consumer Watchdog release citing the NYT, Aug 2026) asserts Microsoft is stepping back from CDR; a separate signal cites Microsoft's emissions rising 25% and reaffirmed CDR commitment. These conflict and neither is verified — but Microsoft's posture is the single largest exogenous variable in this asset.
2. **Removal vs. avoided emissions.** The methane story is *not* currently credited: Vaulted's own Sept 2025 post says injection prevents methanogenic conditions and asks "how much?", with the Google/Isometric work proceeding in phases toward a scientific report. So today's credits are CO₂-only, and the methane co-benefit is an option, not revenue. Conversely, the CO₂ counterfactual depends on what the biosolids/manure/paper-sludge stream would otherwise have done — land application retains some carbon, and the protocol's "ineligible biomass" deduction is exactly where feedstock-specific diligence should concentrate.
3. **No credit-rating coverage.** Trellis (July 2025) noted the method was too new to have been assessed by ratings agencies. Check whether BeZero/Calyx/Renoster have since initiated coverage.
4. **Capital vs. obligation.** Reported cumulative raise is inconsistacross sources **[web]**: PitchBook ~$48M, Tracxn $32.3M, Waste Dive (March 2026) "roughly $56 million." Even at $56M equity, a 4.9 Mt/13-year build-out across multiple new US sites implies substantial future project finance. Frontier's 2024 note said the offtake enabled **three new wells**; Canary Media reported a target of three new sites online by 2027 — verify how many are actually permitted and operating.
5. **Feedstock quality and social licence.** Vaulted's stated advantage is accepting wet/contaminated streams (biosolids, manure, paper sludge) that BECCS and other BiCRS cannot, managed via an internal waste-qualification protocol. That is also its exposure: contaminant regimes (notably evolving US federal and state rules on PFAS in biosolids) could reshape both feedstock economics and permitting. I have **not** verified any specific PFAS regulatory action affecting Vaulted — flagging it as an open question, not a finding.
## 5. Gaps and next steps
- No Vaulted **project** records in the corpus; no policy instrument tied to Vaulted specifically; no media/podcast items; **no Series B on record** (Waste Dive's ~$56M figure suggests capital raised beyond the $32.3M Series A that is not captured anywhere I can verify).
- The corpus's Vaulted deals are **machine-extracted, `review_status: auto`** — the Microsoft date error is a live example of why each should be checked against the primary announcement.
- Pull directly: the Isometric registry project page (issuance-by-month, feedstock mix, LCA deductions per batch), the Biomass Geological Storage v1.1 diff versus v1.0, well permits by class and state for Hutchinson/LA/new sites, the Microsoft contract's delivery schedule and off-ramps, and CDR.fyi's latest delivered-tonnage league table for where Vaulted now ranks. I hit my web-search budget before I could confirm the ICVCM approval date and the current CDR.fyi ranking — treat both as unverified timing.
Judge’s note — All three gold items are substantively present: the Frontier 2024 offtake with $58.3M and a derived ~$382/t (tonnage stated as 154,146 vs the gold 152,480 — a minor discrepancy that doesn't obscure the deal), the Microsoft 4.9 Mt 12-year offtake explicitly framed as ~95% of pathway tonnage, and the biomass-injection/slurry → geological/durable pathway framing, with a sharp caveat that the durability label is a taxonomy default rather than a source claim. Beyond-corpus material is rich and mostly checkable: the $32.3M Prelude-led Series A, XPRIZE $8M award, Frontier 2023 $7M prepurchase, first-year delivery of >7,000 t at >1,500 t/month, Frontier portfolio ~52,100 t delivered as of 2026-08, Google 50,000 t/Isometric verification, the Isometric July 2025 registry announcement and 2038 horizon, CNBC 2026 Disruptor 50 and Q1 2026 waste-volume growth — roughly eight distinct items, several post-dating the corpus. Analytical quality is high for an expert reader: the 'up to 4.9 Mt' ceiling vs stored firm min=max, Class V vs Class VI permitting mismatch, additionality/counterfactual on avoided methane, one-buyer concentration risk, and the observation that the apparent pathway price curve is entirely Frontier's disclosure practice. Evidence-base assessment is candid — two tangential papers, zero field trials, assurance carried by Isometric protocol rather than literature. Citations name sources (Frontier portfolio page, Isometric, Arnold & Porter advisory, Renewable Energy Magazine) and flag record provenance (approved vs auto), though some quoted trade-press claims lack precise attribution and are only loosely hedged as 'company-sourced'. Coverage honesty is exemplary: search-budget exhaustion, missing project/facility records, undisclosed prices, and specific re-run queries are all stated. Deductions are small — the tonnage slip and a couple of unverifiable growth/list claims.
I'll start with the coverage baseline and pull everything on Vaulted Deep. Now the pathway evidence base and the supply side. Let me check the supply side by pathway filter alone, and run the suggested web searches. ## Vaulted Deep — diligence brief **Sources:** Carbon Curve Intelligence (deals ingested through 2026-09-01, latest announcement held 2026-08-28; papers fetched 2026-09-02) plus web search. I flag which is which throughout. Web-search budget was exhausted after four queries, two of which returned results — so the web layer here is thinner than I'd want and I note the specific gaps at the end. --- ### 1. The deal book (database) The corpus holds **three offtakes for Vaulted Deep totalling 5,104,146 tonnes**, of which **one discloses a value** ($58.3M covering 154,146 t). Date range 2024-05-01 to 2025-09-17. | Announced | Buyer | Tonnes | Value | Notes | |---|---|---|---|---| | 2024-05-01 | Frontier | 154,146 | $58.3M → **$382/t derived** | Slurry injection into Class V wells (Kansas, Los Angeles); delivery window 2024–2027. Record is `approved` (human-reviewed). Frontier's portfolio page listed **~52,100 t delivered as of 2026-08** against 154,146 contracted. | | 2025-09-08 | Microsoft | 4,900,000 | undisclosed | 12-year offtake, US, "one of the largest carbon removal transactions ever signed." Record is `auto` (machine-extracted from an Arnold & Porter advisory note, no human spot-check). | | 2025-09-17 | Google | 50,000 | undisclosed | Firm, delivery by 2030, verified by Isometric; follow-on to Google's March 2024 purchase via Frontier. Paired with a joint Google/Isometric initiative on quantifying avoided methane — but the methane work is not the commitment; the 50,000 t removal is. Record is `auto`. | All three are tagged `biomass-injection`, `geological` storage, `durable-1000y` — but note the durability label is **pathway-default on the Microsoft and Google rows, not stated by the source**, and the pathway code itself was only recoded from `other-emerging` on 2026-09-01. Treat the 1,000-year label as a taxonomy convention, not a verified claim in those two deals. **Funding attached to the name:** Series A of **$32.3M led by Prelude Ventures (2024-11-14)**, with Lowercarbon, Earthshot and WovenEarth returning super pro rata plus Fall Line Capital and Rethink Impact. That release is also the best operational datapoint in the corpus: **>7,000 t delivered in the first year of operations, running at >1,500 t/month** at that time. Vaulted also took **$8M in the XPRIZE Carbon Removal awards** (waste-storage category), and was a recipient in Frontier's September 2023 $7M prepurchase cohort. **What the database does *not* have:** zero project records for Vaulted Deep. Its wells (Kansas, Los Angeles) exist in the corpus only as a phrase inside the Frontier deal record, not as facility rows with a build stage. For a company whose thesis is permitted-and-operating infrastructure, that is a real hole in the supply-side picture. ### 2. What web search adds and where it conflicts - **A date discrepancy worth chasing.** The corpus dates the Microsoft deal 2025-09-08 off a law-firm advisory note. Isometric's own announcement is earlier: Vaulted Deep selected Isometric as registry for its 12-year CDR offtake with Microsoft, delivering up to 4.9 million tonnes through to 2038 , published July 2025. Renewable Energy Magazine's write-up frames the same deal as up to 4.9 million tons of durable CDR over 12 years through its waste management infrastructure, running through 2038, allowing Vaulted to expand an approach already permitted, operating and delivering . So the corpus row is the *legal-advisory* date, not the announcement date, and the delivery horizon is 2038 — longer than the "12 years" the corpus notes imply from a 2025 start. Also note "**up to** 4.9 Mt": that is a ceiling, and the corpus stores it as a firm min=max=4,900,000, which overstates the floor. - **Registry choice is confirmed** — Isometric on both the Microsoft and Google deals. - **Traction signals through 2026 (not in the corpus).** Vaulted Deep was named to CNBC's 2026 Disruptor 50 list , and reporting on its Kansas site notes nearly 200% year-over-year growth in managed waste volumes in Q1 2026 , with the company expanding a pipeline of municipal, industrial and corporate disposal contracts . Treat trade-press growth figures as company-sourced. - **The business model is the diligence crux.** The company is pursuing a US organic waste market estimated at over $100 billion, where landfilling, incineration and surface spreading face rising environmental and community pushback, offering deep well injection as compliant disposal that also delivers durable removal. Revenue therefore comes from two sides — tipping fees plus carbon credits — which is the source of both its cost advantage and its additionality question. Frontier's own 2024 diligence claimed 10,000+ year permanence, near-term scalability, local co-benefits and line of sight to under $100 per ton , versus the $382/t actually contracted. ### 3. The pathway's evidence base — thin, and honestly so `biomass-injection` (pumpable biomass to geological storage: bio-oil, slurry) is, in this corpus, **commercially far ahead of its literature**. **Papers: two.** One lab study, one unclassified. Neither is about Vaulted's process: - *Stable isotope analysis of biomethane generation in Powder River basin coals using BiCRS + RNG* (Energy Conversion and Management X, 2026-04-10, CC BY) — a lab isotope study of biomass injected into unmineable coal seams, commercialised by Cowboy Clean Fuels, developing a GC-EA-IRMS method to distinguish biogenic CH₄ from CO₂ retained in the formation. Relevant to MRV technique, not to slurry injection. - *Feasibility of subsurface storage of hydrochar in the Netherlands* (Int. J. Greenhouse Gas Control, 2025-12-05). **There are zero field trials tagged to this pathway in the corpus, and no peer-reviewed paper on organic-waste slurry injection at all.** Two caveats that cut both ways: the pathway code was only created/recoded on 2026-09-01, so older work may sit under other codes or be flagged out-of-scope; and the corpus holds no non-English work, no preprints, and nothing published after 2026-09-02. But even generously, this is a pathway where the assurance burden is carried almost entirely by **registry protocols rather than by published science**. **Where the rigour actually lives:** - **Isometric** — its Bio-oil Geological Storage Protocol is CCP-approved and was used for Charm's first verified credits, with underlying data and calculations published on the registry (corpus record, `auto`). Vaulted's credits sit on the same registry. - **US EPA UIC Class VI** (effective 2010-12-10, operational) covers CO₂ geologic sequestration wells; six states held primacy as of the Texas approval in November 2025. Note the mismatch: Vaulted's Frontier deal describes **Class V** wells, a different and lighter-touch permitting track than Class VI. That regulatory classification is a specific thing to verify in diligence. - **Puro.earth** carries a Geologically Stored Carbon methodology covering the pathway; **DOE FOA 3082 (Carbon Negative Shot Pilots)** and the 2024 pilot-scale FOA fund BiCRS pilots — $48.6M across nine projects in FOA 3082, none of them Vaulted. **Pathway market context (database, all 11 biomass-injection deals):** 5,355,636 t contracted, 4 deals disclosing value. Vaulted is **~95% of the pathway's contracted tonnes** — essentially one Microsoft deal. Disclosed prices in the pathway: Frontier↔Charm 2023 at **$473/t** (contract price declining at least 37%, up to 75%, 2024–2030), Frontier↔Vaulted 2024 at **$382/t**, Frontier↔NULIFE GreenTech 2025 at **$362/t**. All three are Frontier deals, so that apparent price curve describes one buyer's disclosure practice, not the market. Corpus-wide, only 44 of 385 deals disclose a value, covering **1.5% of contracted tonnes**. ### 4. Flags I would carry into a call 1. **Delivery rate vs contract.** ~52,100 t delivered against 154,146 t on a 2024–2027 Frontier window (as of 2026-08) is roughly a third, with about a year left. Ask for the delivery schedule and any renegotiation. 2. **"Up to 4.9 Mt" is a ceiling, not a floor.** The concentration risk is extreme: one buyer is 96% of contracted volume, over a term running to 2038. 3. **Additionality and counterfactual.** The removal claim rests on what the waste would otherwise have done (decompose, be land-applied, off-gas methane). The Google/Isometric methane-quantification initiative exists precisely because that accounting is unsettled. 4. **Class V vs Class VI permitting** and per-site permit status. 5. **No published science on the specific process.** Assurance is registry-mediated. That is not disqualifying — Isometric's data transparency is genuine — but it means protocol revision is a live commercial risk. ### 5. What neither source covers No records here for: Vaulted's facilities as projects; any funding after the November 2024 Series A (a later round would not appear — the corpus's funding lane is thin on private rounds and my web searches for 2026 financing were cut off by the search limit); the actual price on the Microsoft and Google deals; tipping-fee economics; community or permitting opposition at specific sites; and any literature or protocol change after 2026-09-02. I'd specifically re-run: *Vaulted Deep 2026 funding round*, *Vaulted Deep new well permit 2026*, *Isometric organic-waste slurry injection protocol version*, and check the Isometric registry directly for issued-vs-contracted tonnes.
Judge’s note — All three gold items are surfaced substantively: the Frontier 2024 offtake with the $58.3M value (tonnage cited as 154,146 vs the gold 152,480 — same deal, minor discrepancy, and the derived $382/t is correspondingly slightly off), the Microsoft 4.9 Mt 2025 offtake correctly identified as the dominant/largest contract in the pathway, and clear pathway framing as slurry/organic-waste geological injection with durability claims. Beyond the corpus, it adds the $32.3M Series A led by Prelude Ventures (Nov 2024), the $8M XPRIZE Carbon Removal award with peer context, the Google 50,000 t Isometric-verified purchase, and a delivery track record (>7,000 t in year one, ~1,500 t/month run-rate; ~52.1 kt of 154.1 kt delivered per Frontier) that is then used well to compute the ~23x scale-up gap versus the Microsoft obligation and to benchmark against Charm. Citation quality is good — named sources (Frontier portfolio page, Arnold & Porter release, Isometric registry, DOIs, EO/bill numbers) and explicit provenance flags (auto-extracted vs approved records, durability_defaulted=true) — though a couple of the paper/DOI references and derived figures could not be independently corroborated and the answer itself admits it could not verify them. Coverage honesty is exemplary: it states plainly that no web search was available, flags the ~12-month silence since Sept 2025 as possibly a coverage gap, notes the absence of any Vaulted project/permit record, notes only 1 of 3 deals is priced, and lists the specific searches a reader should run. Clarity and decision-usefulness are high: BLUF, deal table, concentration/delivery/ramp risk flags, the genuinely useful Class V vs Class VI regulatory distinction, and a prioritized diligence question list. Main deductions: the self-admitted lack of live verification means the most decision-relevant items (post-2025 deals, Microsoft contract firmness, Isometric issuance) remain open, and a few numbers drift slightly from the gold record.
I'll start by understanding the corpus coverage and pulling Vaulted Deep's entity profile.
Now let me pull the pathway evidence profile and Vaulted Deep's project/facility footprint.
## Bottom line up front
Vaulted Deep is, on the records I can see, the **largest single supplier of contracted tonnes in the "biomass-injection" pathway and roughly the sixth-largest CDR supplier by contracted volume in the corpus (5.10 Mt)** — but ~96% of that sits in one undisclosed-price Microsoft contract, its only priced deal implies $382/t, its delivery run-rate as of the last public datapoint was ~18 kt/yr against an implied ~408 kt/yr obligation, and the peer-reviewed evidence base for its specific pathway is **essentially non-existent (2 papers, 0 field trials)**. The integrity case rests almost entirely on private registry protocols (Isometric), not published science.
**Important caveat on method:** the Carbon Curve corpus explicitly instructs users to supplement with live web searches, and its freshness cut-offs are deals to 2026-08-28, policy events to 2026-08-17, papers to 2026-09-02. **I have no web-search tool available in this session, so I could not run those checks.** Everything below is from the corpus plus my own reasoning; treat the last ~12 months, and anything not publicly announced, as unverified. I flag the specific searches you should run at the end.
---
## 1. The deal book (3 records, 5,104,146 t)
| Date | Buyer | Tonnes | Value | Structure / notes |
|---|---|---|---|---|
| 2024-05-01 | **Frontier** (Stripe, Shopify et al.) | 154,146 t | **$58.3M → ~$382/t** (derived, not quoted) | Offtake, delivery 2024–2027. Slurry injection into **Class V wells, Kansas and Los Angeles**. Frontier's portfolio page lists **~52,100 t delivered as of 2026-08**. |
| 2025-09-08 | **Microsoft** | **4,900,000 t** | Undisclosed | 12-year offtake, US; "one of the largest carbon removal transactions ever signed." Source is **Arnold & Porter, Vaulted's own counsel** — a press release, not a filing. Contemplates "development of new carbon sequestration facilities across the United States." |
| 2025-09-17 | **Google** | 50,000 t | Undisclosed | Firm, delivery **by 2030**, **Isometric**-verified. A follow-on to Google's March 2024 purchase via Frontier. Bundled with a **joint Google/Isometric initiative to quantify avoided methane** — note that the methane piece is an *avoidance* claim, not the removal commitment. |
Earlier: Vaulted was one of 12 recipients in **Frontier's Fall 2023 $7M prepurchase cohort** (2023-09-07), and Isometric's registry launch (2024-04-22) states its **first-ever issued credits went to Frontier from Vaulted Deep**.
**Diligence flags on the deal book:**
- **Counterparty concentration is extreme.** Microsoft is 96.0% of contracted tonnes. Two corpus signals are relevant and cut in opposite directions: Microsoft's own emissions rose 25% YoY on AI datacentre build-out with restated CDR commitment (gasworld, uncorroborated), while a Consumer Watchdog release (2026-08-10) asserts, citing the NYT, that "Microsoft [is] stepping back from the industry it helped create." Both are filed as **uncorroborated signals, not verified records** — but the second is exactly the thing to verify before underwriting a 4.9 Mt contract.
- **Delivery is lagging its own comparable.** Frontier's page shows Vaulted at ~52.1 kt of 154.1 kt (~34%) with roughly two-thirds of the 2024–2027 window elapsed. Charm Industrial, the nearest analogue, shows ~100.6 kt of 112.0 kt (~90%) on a 2024–2030 window.
- **The scale-up gap is the central question.** At the Series A (2024-11-14) Vaulted stated >7,000 t delivered in year one and a run-rate >1,500 t/month (~18 kt/yr). Microsoft's 4.9 Mt over 12 years averages **~408 kt/yr** — a ~23× step-up. For context, Frontier's June 2026 Growth AMC page says its *seven* portfolio companies together delivered ~23,000 t in 2025.
- **Price visibility is one datapoint.** Only 1 of 3 deals discloses value. Corpus-wide only 44 of 385 deals disclose value, covering 1.5% of tonnes — so $382/t is a 2024 Frontier-catalytic price, not a read on what Microsoft paid.
- **Provenance:** 2 of the 3 Vaulted deals are `review_status: auto` (machine-extracted by claude-haiku-4-5 from the cited URL, auto-published, no independent verification). The Frontier one is `approved`. Verify tonnage and terms against primary documents.
- **Nothing recorded since 2025-09-17.** That is ~12 months of silence in a corpus that runs to 2026-08-28 and dropped its minimum deal size on 2026-08-24. Could be a genuine quiet period or a coverage gap.
**Funding:** $32.3M Series A (2024-11-14) led by **Prelude Ventures**, with Lowercarbon, Earthshot and WovenEarth taking super pro rata, plus Fall Line Capital and Rethink Impact. Also an **$8M XPRIZE Carbon Removal award** ("waste storage" category, part of the $100M purse — behind Mati's $50M and NetZero's $15M). The corpus tracks buyer-side commitments and equity rounds but not debt, so assume its financing picture is incomplete.
**Where Vaulted sits:** among the 40 largest supplier buckets, contracted tonnes rank roughly BTG Pactual Timberland 9.30 Mt · Stockholm Exergi 9.26 Mt · Catona 8.35 Mt · Climate Impact Partners 6.90 Mt · re.green 6.50 Mt · **Vaulted Deep 5.10 Mt** · CO280 4.36 Mt · Lithos 3.67 Mt. Every one of those buckets is a floor, not a census.
---
## 2. The pathway evidence base: thin to the point of absence
This is where I'd concentrate diligence.
**Peer-reviewed literature: 2 papers total.** Out of 22,411 papers held and 6,636 classified in-scope for CDR, `biomass-injection` has **two**:
1. *Stable isotope analysis of biomethane generation in Powder River basin coals using BiCRS + RNG* (Energy Conversion and Management X, 2026-04-10, DOI 10.1016/j.ecmx.2026.101836) — a **lab** study, and it is about **Cowboy Clean Fuels** injecting biomass into unmineable coal seams to make RNG. Not slurry, not Vaulted's process.
2. *Feasibility of subsurface storage of hydrochar in the Netherlands* (Int. J. Greenhouse Gas Control, 2025-12-05, DOI 10.1016/j.ijggc.2025.104539) — study type unclassified; hydrochar, not organic-waste slurry.
**There are zero field trials, zero LCAs and zero pathway-specific MRV papers.** Neither paper describes what Vaulted actually does. Compare this to ERW or DAC, where the field-trial and LCA literature runs to hundreds of entries. Note also that the taxonomy itself is young: Vaulted's deals were **recoded from `other-emerging` to `biomass-injection` on 2026-09-01**, with the corpus noting "the taxonomy names Vaulted Deep as the first tenant of biomass-injection." The category was in part defined around this company.
**Durability is asserted, not evidenced.** Both the Microsoft and Google records carry `durability: durable-1000y` with `durability_basis: pathway-default` and `durability_defaulted: true` — i.e. **the corpus assigned 1,000-year durability by pathway convention; the sources did not state it.** The Frontier record is `seed-curated`. No source in this dataset independently substantiates millennial permanence for organic-waste slurry.
**What the evidence actually rests on is private certification.** Isometric verifies the Frontier and Google volumes and issued Vaulted the first credits on its registry. Puro.earth's "Geologically Stored Carbon" methodology also covers the class (General Rules v4.4, approved 2026-05-07). For the adjacent bio-oil case, Isometric's Bio-oil Geological Storage Protocol is **ICVCM CCP-approved**, and Charm's first credits (107.21 t, 2024-08-22) were issued ex-post with full calculations published. That is a real integrity pathway — but it is a standards body, not replicated independent science, and I could not confirm from this corpus whether a slurry-injection protocol has itself cleared ICVCM.
**Supply-side records are empty.** The corpus holds **one** biomass-injection project, and it is Charm's. There is **no project record for any Vaulted facility** — no Kansas or Los Angeles site, no permit, no stated capacity. Given the Microsoft contract explicitly contemplates "new carbon sequestration facilities across the United States," the absence of a documented build pipeline here is a gap you must close directly with the company and with state permit databases.
---
## 3. Regulatory posture — the Class V / Class VI distinction matters
Frontier's description is specific: Vaulted injects into **Class V wells**. That is a different legal regime from the **Class VI** program (EPA UIC under the SDWA, effective 2010-12-10) that governs DAC/BECCS CO₂ storage — Class V permitting is lighter, largely state-administered, and designed for non-hazardous fluid disposal rather than supercritical CO₂. Commercially this is an advantage (faster, cheaper, no Class VI queue). For diligence it raises two questions: **(a)** does a Class V permit deliver the monitoring and post-injection site care that a 1,000-year durability claim implies, and **(b)** is that regime durable against the rising political friction around subsurface injection?
That friction is well documented in the corpus, even though it currently targets Class VI:
- **Louisiana** Gov. Landry's EO JML 25-119 (2025-10-15) suspended review of new Class VI applications; ~20 CCS bills filed in 2025 including a proposed statewide moratorium; HB 4 would let parishes ban wells outright.
- **Illinois** Senate voted 55-0 (2025-04-14) to ban sequestration within the Mahomet aquifer zone after the Decatur leaks.
- Counterweight: six states now hold Class VI primacy (ND, WY, LA, WV, AZ, TX — Texas effective 2025-12-15), and EPA issued a final Class VI permit in Indiana on 2026-04-10.
- **EPA's proposed GHGRP rollback** would relieve 46 of 47 source categories of reporting — the corpus notes GHGRP underpins carbon-management MRV and 45Q eligibility.
**Policy tailwinds** are real but generic: FY26 appropriations (H.R. 6938, effective 2026-01-26) preserved **$45M for the DOE CDR Purchase Pilot Prize** with ≥$4M ring-fenced for MRV and up to $71.5M for CDR R&D; DOE's FOA 3082 funded BiCRS pilots (Vaulted was **not** among the named selectees — Mote, Arbor, Carba, Carbon Lockdown, Clemson were); the **Carbon Dioxide Removal Investment Act** ($250/t PTC, Bennet–Murkowski) remains **proposed only**. California's SB 643 purchase program was **vetoed 2025-10-13, veto sustained 2026-03-02**.
**EU market access looks closed for now.** The CRCF has adopted permanent-removal methodologies for **BioCCS, DACCS and biochar**; the July 2026 EU ETS review proposal contemplates central purchase of **250 Mt of CRCF-certified permanent removals over 2031–2040, specified as DACCS and BioCCS**. Slurry/organic-waste injection is not named in either. LEGO's Feb 2026 commitment did include "biomass geological storage using slurry injection" (via Climate Impact Partners/ClimeFi, supplier unnamed) — worth chasing as a possible undisclosed Vaulted channel.
---
## 4. What I'd ask, in order
1. **Microsoft contract mechanics.** The Arnold & Porter release calls it a "carbon reduction unit offtake." Is the full 4.9 Mt firm and unconditional, or milestone/volume-flexed with termination rights? What's the delivery schedule shape, and what happens on shortfall?
2. **Delivery reconciliation.** Pull the Isometric registry directly: issued vs. retired tonnes by vintage. Reconcile against the Frontier portfolio figure (~52.1 kt of 154.1 kt as of 2026-08) and against the 2027 contract end.
3. **The 23× ramp.** Site list, permit status per site, capex per tonne, and the feedstock supply agreements (biosolids/organic waste is a locally-contracted, regionally-constrained input).
4. **Class V durability defensibility.** Formation characterisation, plume modelling, monitoring obligations, post-injection care, and whether any buyer or standard has stress-tested the 1,000-year claim.
5. **Counterfactual and additionality.** The Google framing is avoided-methane-from-landfilling. Where organic waste would otherwise be land-applied rather than landfilled, the baseline is materially different — and California's proposed SB 285 signals regulators tightening on biomethane-adjacent accounting.
6. **Protocol status.** Has Isometric's slurry/organic-waste protocol been through ICVCM assessment, as the bio-oil protocol has?
## 5. Searches I could not run — please run these
- `"Vaulted Deep" offtake OR agreement carbon removal 2026`
- `"Vaulted Deep" carbon removal announcement 2026-09` *(corpus's own suggested search)*
- `Vaulted Deep Class V well permit Kansas` and the equivalent for California/Colorado
- `Isometric registry Vaulted Deep credits issued` (delivery truth-check)
- `Microsoft carbon removal portfolio 2026 scaling back` (test the Consumer Watchdog/NYT claim)
- `biomass injection carbon removal field trial 2026` (test whether literature has appeared since 2026-09-02)
- CDR.fyi's public reports — the corpus itself names these as "the fuller transaction scoreboard" and concedes it is not a census.
**Net read:** commercially the strongest position in its pathway, with blue-chip buyers, an Isometric verification track record and a genuine cost advantage from Class V permitting. Scientifically the thinnest evidence base of any pathway I can examine here — the durability claim underpinning $5.1 Mt of contracts is a taxonomy default, not a published finding. The risk is not that the technology fails; it is that a single buyer, a single permitting regime and a single registry protocol all have to hold simultaneously, with no independent literature to fall back on if any of them is challenged.
Asked, verbatim: “Where is government procurement of carbon removal actually happening today, and how does the money compare to private buying?”
Judge’s note — Strong on the government side: named, dated, source-attributed instruments across Sweden (SEK 20bn / 800 kt/yr Stockholm Exergi reverse auction), Denmark (CCUS Fund/Ørsted 430 kt/yr, NECCS first tender winners and 160,350 t/yr, DKK 28.7bn CCS Fund with the correct caveat that it is not removal-only), UK DESNZ GGR business model as a 15-year CfD tied to HyNet, EU certification-not-demand framing, DOE $35M/~$30M purchase prize with 24 semifinalists and per-team volumes, and Canadian NRCan/CGF instruments. Several of these are primary-source budget figures beyond a static corpus, though nothing post-mid-2026 appears, so the recency half of the beyond-corpus probe is unmet. The analytical framing that most 'procurement' is really supply-side subsidy rather than credit purchase-and-retire is genuinely useful and correctly identifies DOE's prize as the rare true buyer. Weaknesses are on the comparison side, which is half the question. There is no aggregate count of instruments across jurisdictions, and the private-market side rests almost entirely on unverified recollection (≈8 Mt in 2024, Microsoft ~80%) rather than the record of hundreds of offtakes and ~400 funding commitments; the directional claim that private dominates contracted tonnes is made but not substantiated. Critically, the two disclosure caveats in the checklist are absent: no acknowledgement that deal values are disclosed for only about one deal in nine (~11% of contracted tonnes), and no statement that market-wide price claims therefore describe only the disclosing minority — a notable omission given the answer does assert price comparisons (~$150–165/t Swedish vs voluntary DAC prices). Internal sloppiness too: 'three orders of magnitude' immediately followed by 'roughly 60×'. Honesty and structure are excellent — unverified figures are explicitly flagged, the DOE program-status risk is highlighted as the key open question, and unsearched jurisdictions (Japan, Germany, Switzerland) are named as gaps — which lifts the score despite the missed quantitative caveats.
I'll research this systematically — government CDR procurement programs and private market purchase volumes. I ran through several rounds of live searching and hit my search-tool budget partway through, so I'll flag clearly below which numbers I verified in this session and which are from my prior knowledge and need re-checking before you use them. --- ## The short answer Real government money for carbon removal today is overwhelmingly **European, and it is not "procurement" in the sense the CDR industry usually means it.** It's supply-side subsidy — reverse auctions and contracts-for-difference that pay a developer per tonne stored — not a government buying and retiring credits. The one place a government has behaved like a *buyer of credits* at any scale is the US DOE's purchase prize, and that program is roughly 1/50th the size of a single Swedish award. --- ## Tier 1: Governments paying big money per tonne (Nordics) **Sweden — the largest single commitment in the world.** The Swedish Energy Agency ran a reverse auction for bio-CCS and announced Stockholm Exergi as winner in late January 2025. The award is just over **SEK 20 billion (~US$1.8bn)**, disbursed over up to 15 years from the start of geological storage, supporting **800,000 tonnes/year** of permanent removal. Six companies bid; the agency selected the lowest requested public support per tonne. The agency's own framing was capture and storage of **over 11 million tonnes** of biogenic CO₂ (Energimyndigheten; Stockholm Exergi press release, 28 Jan 2025). That implies an effective public price of roughly **SEK 1,650–1,800/tonne (~$150–165/t)** — a genuine market signal, and notably below voluntary-market DAC prices. The interesting wrinkle: the *same* facility already has offtakes from Microsoft and from Frontier (buying for Alphabet, Meta, JPMorgan Chase, H&M, per ESG Today, 28 Jan 2025). So public and private money are stacked on one project. Who gets to claim those tonnes — the Swedish national inventory or the corporate buyers — is a live accounting question I would not assert an answer to. **Denmark — the most mature repeat buyer.** Three separate pots, run by the Danish Energy Agency: - **CCUS Fund** (~DKK 8bn): won by Ørsted, **430,000 t/yr from 2026 for 20 years** (biogenic feedstock, so effectively BECCS). - **NECCS Fund** (negative emissions specifically): first tender completed May 2024, three winners — BioCirc CO₂, Bioman, and Carbon Capture Scotland (now The Carbon Removers) — totalling **160,350 t/yr for 2026–2032**. Carbon Gap's tracker (Oct 2025) states there are **no plans for further NECCS rounds**, which matters for anyone modelling Danish demand. *[The NECCS budget figure — I recall roughly DKK 2.3–2.6bn — I could not re-verify; treat as unconfirmed.]* - **CCS Fund**: **DKK 28.7bn (~US$4.2bn)** incl. VAT, tender materials published Jan 2025, application deadline 25 March 2025, storage from 2029. Caveat: this fund is not removal-only — it covers fossil capture too, so you cannot count the headline number as CDR spending. --- ## Tier 2: Frameworks that will pay, but haven't yet **United Kingdom.** DESNZ published the **Greenhouse Gas Removals Business Model in August 2025**: a **15-year contract-for-difference** paying developers the gap between a negotiated strike price and market revenue, initially aimed at projects in the **HyNet Track-1 expansion** (Carbon Herald, Sept 2025). In parallel, the July 2025 UK ETS consultation response set out integrating engineered GGRs into the ETS, and DESNZ has separately tendered for **Minimum Quality Thresholds** and for modelling of future voluntary-market GGR prices (Contracts Finder notices). So: architecture in place, strike prices not public, no tonnes contracted that I can confirm. **EU.** No dedicated CDR procurement instrument. The Carbon Removals and Carbon Farming certification framework creates *quality infrastructure*, not demand. Carbon Gap notes the **Net-Zero Industry Act** negotiations could reduce the administrative burden on member states setting up national procurement schemes — i.e., an enabler, not a buyer. --- ## Tier 3: The US — small, and now genuinely uncertain **DOE Carbon Dioxide Removal Purchase Pilot Prize:** total prize value **$35M**, of which about **$30M** is for actual credit purchases from up to ten winners. DOE announced **24 Phase 1 semifinalists on 28 May 2024**, each receiving **$50,000** (DOE FECM; Heatmap, 28 May 2024). Individual delivery proposals were small — Alkali Earth 8,108 credits, Eion 9,900, Lithos 8,109, CREW Carbon 7,500 (DOE semifinalist page). Isometric noted seven of its clients alone proposed **over 73,000 tonnes** across all four pathways. **Important gap:** I could not verify the program's status in 2025–26 under the current administration, which has cancelled a range of DOE carbon-management awards. **Do not assume the $30M has been or will be spent.** That is the single most important open question in US CDR procurement and worth a direct check with FECM. **DOE CO₂RP Challenge:** a *voluntary* purchasing challenge (Federal Register notice of intent, 14 March 2024) asking organisations that disclose GHG inventories to buy and retire permanent CDR annually starting no later than 2025, with disclosure to DOE. It commits no federal dollars. Status also unverified. **Legislation:** the Federal CDR Leadership Act and the bipartisan **CREST Act** proposed scaling federal procurement but did not pass in the 118th Congress; Massachusetts and New York have state-level CDR Leadership Acts in process; in the 119th Congress the **Foreign Pollution Fee Act** includes a provision letting producers buy CDR against the fee (WRI; Carbon Gap tracker). **Canada** sits between tiers: Natural Resources Canada awarded **nearly $6M** to Deep Sky partnerships in October 2025 (Deep Sky, 31 Oct 2025) — but that is R&D/storage-appraisal funding, not credit purchase. Canada Growth Fund carbon-contracts-for-difference and the DAC investment tax credit are the real instruments; I'd flag the frequently-cited "60% ITC" number as needing verification, as I only saw it in a low-quality secondary source. --- ## How this compares to private buying | | Government | Private | |---|---|---| | Dominant form | Per-tonne subsidy / CfD over 15–20 yrs | Offtake contracts, credits retired by buyer | | Largest single deal | Sweden → Stockholm Exergi, ~$1.8bn / ~11–12 Mt | Microsoft's multi-Mt BECCS offtakes | | US federal purchase budget | ~$30M | — | | Concentration | 2 countries (SE, DK) ≈ nearly all committed cash | One buyer (Microsoft) ≈ most tonnes | Two honest comparisons: **On money, Europe is now bigger than the entire US public effort by three orders of magnitude.** Sweden's single award (~$1.8bn) exceeds the DOE purchase pot (~$30M) by roughly 60×. If Denmark's CCS Fund allocates even a modest slice to biogenic capture, Nordic public commitments alone dwarf every other government's combined. **On tonnes, private buying is still the demand engine — and it is dangerously concentrated.** Here I have to be candid: my recollection is that CDR.fyi recorded roughly 8 Mt of durable CDR purchased in 2024 (up sharply on 2023) with **Microsoft accounting for something like 80% of volume**, and that 2025 set a further record on the back of very large BECCS deals. **I could not re-verify those figures in this session and you should pull CDR.fyi's own year-in-review before citing them.** The directional claim I'm confident in is that a single corporate buyer dominates contracted volume, and that the gap between tonnes *contracted* and tonnes *delivered* remains large. --- ## What I'd flag as the real analytical points 1. **"Procurement" is mostly a misnomer.** Sweden, Denmark and the UK are de-risking supply, not creating a retail buyer of last resort. Only DOE's prize actually buys and retires credits — and it's tiny and possibly stalled. 2. **Public prices are far below voluntary DAC prices.** Sweden's ~$150–165/t reflects BECCS economics. No government has yet paid DAC-level prices at scale. 3. **Denmark signalling no further NECCS rounds** is a meaningful negative datapoint against the "governments will keep buying" thesis. 4. **Double-claiming is unresolved** where state subsidy and corporate offtake sit on the same tonne (Stockholm Exergi is the test case). 5. **Biggest gaps in this answer:** current status of the DOE prize; verified CDR.fyi 2025 volumes; Danish NECCS budget; Japanese, German and Swiss activity, which I was unable to search at all and which I would not want to characterise from memory.
Judge’s note — The answer covers all four checklist items substantively: it enumerates procurement-typed instruments across jurisdictions (~40 of 230 policy records, 28 jurisdictions) including the DOE CDR Purchase Pilot Prize, the EU ETS-review 250 Mt central purchasing proposal, Sweden/Denmark auctions, Canada, UK GGR, Romania; it quantifies the private side (385 deals, 137 Mt, Microsoft ~70%) and shows it dwarfs public buying on tonnes (~100:1); it quantifies the disclosure gap (44 of 385 deals disclose value — ~1 in 9); and it explicitly states that any per-tonne price derived from disclosed values describes only the disclosing minority. Note one numeric divergence from the gold: it reports disclosed deals covering 1.5% of contracted tonnes rather than ~11%, which may be a computation/scope error, and it does not give the ~400 funding-commitment count (only DOE's 13 and Frontier's 16). Beyond-corpus contributions are real and multiple: CDR.fyi April 2026 buyer-share stats, Q1 2026 record contracting, CDR.fyi June 2026 delivered/retired split, Sweden's budget-forecast cut and second auction via trade press, and Energimyndigheten award-scale figures — with a sensible warning about scope mismatch between the two datasets. The framing insight — governments subsidise supply at billion-dollar scale while buying tonnes at single-digit-million scale, and capture-vs-removal definitional inflation — is genuinely decision-useful. Sourcing is named but not linked; a few specifics (Romania POPAM, Danube CCS appeal, court ruling) are unverifiable from the text. Honesty about gaps is exemplary, with an explicit unresolved list and 'absence of a record is not absence of activity.'
I'll start by mapping what the corpus tracks, then pull the procurement instruments and compare against private deal volumes. Now let me quantify the private side and check how government purchases appear in the deals layer. Now let me verify and update these against the web, since the corpus stops at early September 2026. ## Short answer Government procurement of carbon removal is real but geographically concentrated, and almost all of the *money* sits in two Nordic subsidy auctions plus one very large EU programme that doesn't start paying until 2031. Measured by dollars actually committed to buying tonnes today, the public sector is somewhere between "a rounding error" (US, Canada) and "the single biggest cheque in the industry" (Sweden) — with nothing in between. Private buying still dominates on volume by roughly two orders of magnitude, and one buyer dominates the private side. --- ## 1. Where governments are actually buying today The Carbon Curve corpus holds 230 policy instruments across 28 jurisdictions, of which ~40 are typed as procurement. Sorting them by whether money has actually moved: **Money committed, contracts signed** | Jurisdiction | Instrument | Scale | Status | |---|---|---|---| | **Sweden** | BECCS reverse auction (Energimyndigheten) | SEK 36bn envelope 2026–2046; first auction won by Stockholm Exergi at ~SEK 20bn over 15 years for ~800 kt/yr | Operational; EU state-aid cleared July 2024 | | **Denmark** | NECCS Fund (negative emissions only) | DKK 2.5bn; 160,350 t/yr of biogenic CO₂ for 2026–2032, awarded 17 Apr 2024 to BioCirc CO₂, Bioman and Carbon Capture Scotland | Operational, contracted | | **Denmark** | CCS Fund (DKK 28.7bn) | First tender concluded June 2026: Aalborg Portland (ACCSION) up to DKK 16.5bn for ~1.25 Mt/yr from 2030 | Operational — but this is *cement*, i.e. avoided fossil emissions, not removal | | **United States** | DOE CDR Purchase Pilot Prize | Up to $35m total; 24 semifinalists (June 2024); up to 10 winners get purchase agreements of up to $3m each | Operational but "closed to new competitors pending administration guidance"; corpus notes a $45m FY26 appropriation preserved it | | **Romania** | Aquaculture carbon-sequestration compensation (POPAM Action 2.1.4) | €14.285m budget; €174.92/ha/yr extensive, €97.18/ha/yr semi-intensive; call open 24 Jul – 6 Sep 2026 | Operational; described as the first of its kind in the EU | Sweden's is by far the largest: the Energimyndigheten framing of the first auction — "20 billion to capture and store over 11 million tons of biogenic carbon dioxide" — implies roughly SEK 20bn (~US$2.0–2.1bn) for ~11 Mt, i.e. an effective public price in the ~$180–200/t range over a 15-year contract. That single contract is larger than every other government CDR purchase on earth combined. **Committed on paper, not yet awarded** - **Canada**: at least CAD $10m of CDR purchases by 2030 announced October 2024 under the Greening Government Strategy; RFI issued February 2025; the corpus records the competitive award as still **pending as of August 2026**. - **UK**: the GGR Business Model (15-year CfD-style top-up on credit sales) had full documentation published 27 August 2025 and was reaffirmed in the government's 17 July 2026 response to the Whitehead review — but the corpus could find **no signed GGR contract as of August 2026**. The HyNet Track-1 negotiation list (5 Aug 2025) named Evero's Ince BECCS as Priority (~217 kt/yr) and Climeworks' Silver Birch DACCS as Standby. - **EU**: the CRCF Buyers Club, announced November 2025 and detailed at CRCF Days in May 2026, is a hybrid public-private demand-aggregation platform aiming for **first purchases of permanent removals by December 2026**. - **Switzerland**: the Climate Protection Ordinance amendment (federal administration net-zero 2040 via negative emissions) closed consultation 12 February 2026 with no adoption verified; first roadmaps planned for the 2028–2032 legislature. **The big one, still just a proposal** The Commission published the EU ETS review on 17 July 2026 including central purchasing of **250 Mt of CRCF-certified permanent removals (DACCS/BioCCS) over 2031–2040**, funded by auctioning 250m extra allowances, scaling to 48 Mt/yr by 2040, with biochar excluded. Modelled budgets range €29–43bn (PRIMES) to €46–65bn (POTEnCIA). Two cautions the corpus flags explicitly: the legal text does not make allowance release conditional on removals actually being contracted or delivered, and 250 Mt is an intended ceiling rather than a financed commitment. **Where it has gone backwards** - **California SB 643** (a ~$50m CDR purchase programme) passed 37–0 in the Senate and 78–1 in the Assembly, was vetoed by Governor Newsom on 13 October 2025 on budget-duplication grounds, and the veto was sustained on 2 March 2026. - **US federal**: the House FY2027 Energy & Water bill (committee-passed 20 May 2026) proposes the Carbon Dioxide Removal line at **$15m, a 67% cut**, and reprogramming ~$2.8bn of IIJA money including DAC Hubs and DAC prizes. DAC Hub funding was restored around April 2026 only after a federal court ruled the October 2025 mass grant terminations unconstitutional. - **Sweden** cut its long-term BECCS spending forecast by 15% in a recent budget, per Quantum Commodity Intelligence (December 2025), while opening a second auction with roughly $1bn on offer. I was unable to verify the second auction's outcome — treat it as open. - The Commission's approval of Sweden's BECCS state aid was appealed by Danube CCS Ventures (published in the Official Journal 22 April 2025); the appeal does not suspend the scheme. --- ## 2. How the money compares to private buying **Private volume, from the corpus (exact over 385 held deals, 2021 → 28 Aug 2026):** - 137.0 Mt contracted in total - Only **44 of 385 deals disclose a value**, totalling **$1.28bn** and covering just **2.07 Mt — 1.5% of contracted tonnes** - Microsoft: 61 deals, **96.3 Mt** (~70% of all corpus tonnes); Meta 8.7 Mt; Frontier 50 deals / 1.90 Mt but **$740m of disclosed value**; Google 0.95 Mt - By year: 2024 = 37.9 Mt, 2025 = 69.6 Mt, 2026 YTD (to 28 Aug) = 20.2 Mt **Cross-checked against CDR.fyi (web, outside the corpus):** as of 13 April 2026, Microsoft accounted for 36,439,157 t or **78.5% of total disclosed durable CDR tonnes contracted**, with Frontier-linked buyers at 1,841,384 t or 4.0%. Q1 2026 was the largest opening quarter on record at **2.3 Mt contracted, ~560% of Q1 2025**, with Microsoft's single 1 Mt agreement plus 113+ other purchasers taking 1.3 Mt, and biochar at 93% of quarterly volume. Note the scope mismatch: CDR.fyi's *durable* universe implies ~46 Mt cumulative, while the corpus's 137 Mt includes conventional and nature-based contracts — don't sum them. **The comparison, stated honestly:** 1. **On tonnes**: private contracting is ~137 Mt cumulative (corpus) or ~46 Mt durable-only (CDR.fyi). Government-contracted *removal* tonnage is roughly **1 Mt/yr** globally — Sweden's ~800 kt/yr plus Denmark's ~160 kt/yr, with the UK's ~217 kt/yr not yet under contract. That's a ratio of order 100:1 in favour of private buyers. 2. **On dollars, the ranking inverts.** Sweden's single BECCS award (~$2.1bn) exceeds the entire disclosed private dollar total in the corpus ($1.28bn across 44 deals). Add Denmark's NECCS (DKK 2.5bn ≈ $390m) and public committed money for removal is comfortably larger than anything the private market has publicly priced. But the private figure is a severe undercount by construction: the buyers with the most tonnes disclose no prices, and 112 of 385 deals disclose neither tonnes nor value. 3. **The instruments are not the same thing.** Almost all the large public money is *production subsidy* — reverse auctions and CfDs paying a facility to exist (Sweden, Denmark, UK GGR/Power BECCS). Genuine buy-the-credit programmes where a government is the customer are tiny: DOE's prize (≤$3m per winner), Canada's CAD $10m, Romania's €14m. So the honest framing is: governments de-risk supply at billion-dollar scale; they buy tonnes at single-digit-millions scale. 4. **Public money still prefers grants over purchases.** In the corpus's funding layer, the **US Department of Energy is the largest single funder with ~$2.94bn across 13 commitments** (2023–2025), against Frontier's $2.04bn across 16 — and the EU Innovation Fund announced €2.7bn for 54 clean-industry projects on 24 March 2026. Grant and tax-credit support dwarfs procurement almost everywhere except Scandinavia. 5. **A quality point in the private column.** CDR.fyi's June 2026 update reports that ex-Microsoft and ex-Frontier buyers represent only 17% of contracted tonnes but **90% of delivered and 94% of retired tonnes** — the headline private volumes are heavily forward-loaded, whereas the Nordic public contracts are tied to specific plants with dated delivery windows. --- ## 3. Gaps and things I could not verify - **Silent jurisdictions.** The corpus returned no 2026 policy events for Norway, South Korea, Germany, Ireland, Luxembourg or Brazil, and its own note says subnational and non-OECD coverage is thin. Norway's Longship programme funding BECCS at Hafslund Celsio, Japan's approach under GX-ETS (mandatory phase began 1 April 2026), and any Gulf or Chinese activity are **not** captured here. Absence of a record is not absence of activity. - **Sweden's second auction result** and the current operational status of DOE's CDR Purchase Pilot Prize under the present administration are both unresolved; my web search budget was exhausted before I could confirm either. Anyone relying on these numbers should check the Energimyndigheten and DOE FECM pages directly. - **Denmark's remaining ~DKK 12bn** of the CCS Fund — whether it goes to a second tender, and its status — is flagged unverified in the corpus itself. - **The private denominator is unknowable.** Any per-tonne price derived from the disclosed $1.28bn describes the disclosing minority, not the market, and every corpus total is a floor rather than a census. CDR.fyi's public reports are the fuller transaction scoreboard. - **Definitional risk in the public column.** Denmark's DKK 16.5bn Aalborg Portland award and much of the UK's £22bn CCS cluster money are carbon *capture*, not removal. Counting them as CDR procurement — as headline comparisons sometimes do — inflates the public figure by an order of magnitude.
Judge’s note — Coverage of the checklist is strong. The answer enumerates procurement-typed instruments across many jurisdictions (US DOE Purchase Pilot Prize with the $35m ceiling and $45m FY26 appropriation, Canada's C$10m Greening Government process, EU ETS Article 9c/CRCF Buyers Club, Denmark NECCS/CCS Funds, Sweden BECCS auction, UK GGR business model, Switzerland, Romania, Cuba, plus the vetoed California SB 643), though it never states the count of procurement-type records. It quantifies the private side (385 deals, 137 Mt, top buyers table with Frontier's $740m across 22 deals) and makes the public-vs-private comparison explicitly and carefully, distinguishing production subsidy from credit purchase \u2014 a genuinely useful analytical move beyond the checklist. The disclosure caveat is present and quantified (44 of 385 deals disclose value \u2014 the ~1-in-9 ratio matches gold), and it explicitly refuses to derive a $/tonne figure because price-publishing buyers are not volume-contracting buyers. One numeric divergence from gold: it reports disclosed-value deals covering 1.5% of contracted tonnes rather than ~11%, which understates the tonnage coverage and is a substantive arithmetic/derivation discrepancy in the very caveat it foregrounds. It also omits the funding-commitment layer (~400 records) when sizing the private side. Beyond-corpus material is plentiful and dated: DOE's own page on the 24 semifinalists sharing $1.2m, Carbon Pulse on the March 2026 Canadian competitive process and five standing offers to 2029, Carbon Pulse (18 Aug 2026, post-ingest) on three bidders into Sweden's second BECCS auction, Quantum Commodity Intelligence on Sweden trimming BECCS spend 15%, CDR.fyi/Carbon Herald April 2026 buyer concentration (Microsoft 78.5%), CDR.fyi Q1 2026 record quarter and the June 2026 contracted-vs-delivered split, and Heatmap on 2025 ex-Microsoft volume falling. Sources are named with dates but mostly without links, and several figures are unverifiable as presented. Coverage honesty is exemplary: auto-extraction review status, loose application of the procurement tag, cdr-and-ccs scope conflation, cross-layer non-summability, 22 of 27 jurisdictions silent in the last window, and an explicit list of what neither corpus nor search covers (disbursement vs contracting, Longship removal share, GX-ETS, subnational/non-OECD). Structure is tiered and immediately decision-useful for an expert reader.
I'll start by establishing what the corpus covers, then pull the procurement and deal data. Now the private-buying side, to compare scale. Now the required web searches for what the corpus flags as beyond its edges. ## The short answer Government money for carbon removal is real and, in aggregate, larger than everything corporate buyers have ever spent — but almost none of it is *procurement of credits*. The large sums are Nordic production subsidies and UK revenue-support contracts paid to facilities. Actual government purchase of removal tonnes, the thing that looks like what Microsoft or Frontier does, is still a rounding error: single- to double-digit millions of dollars, and in the two flagship programmes (US DOE, Canada) not yet fully awarded. Below, the Carbon Curve corpus records first, then what I found by searching. Corpus freshness: deals to **2026-08-28** (last ingest 2026-09-01), policy events to **2026-08-17**. --- ## 1. Where government buying is actually happening I pulled every instrument typed `procurement` in the corpus. Three important caveats up front: **196 of 230 policy instruments are machine-extracted with no human spot-check** (`review_status: auto`), the `procurement` tag is applied loosely (it captures Ireland's embodied-carbon construction guidance, a Wyoming–Colorado MOU, and the Trump budget proposals, none of which buy removal), and several instruments are `cdr-and-ccs` scope, so their headline budgets are not all removal. ### Tier 1 — Money contracted and flowing (subsidy, not credit purchase) | Instrument | Jurisdiction | Scale | Status | |---|---|---|---| | **Sweden BECCS reverse auction** (Energimyndigheten) | Sweden | SEK 36bn envelope 2026–2046 (~US$3.7bn); first auction won by **Stockholm Exergi, ~SEK 20bn over 15 years for ~800 kt/yr** | Operational; `approved` | | **Denmark NECCS Fund** | Denmark | DKK 2.5bn; three projects, **160,350 t biogenic CO₂/yr, 2026–2032** — BioCirc CO2, Bioman, Carbon Capture Scotland, contracted 2024-04-17 | Operational; `approved` | | **Denmark CCS Fund** | Denmark | DKK 28.7bn (~US$4.5bn) over 15 years; first tender concluded June 2026 with **Aalborg Portland (ACCSION) for up to DKK 16.5bn, ~1.25 Mt/yr from 2030** | Operational; open to fossil *and* biogenic CO₂ — the Aalborg award is cement, so removal is only part of it | | **Romania aquaculture carbon sequestration** (Action 2.1.4) | Romania | **€14.285m**; €174.92/ha/yr extensive, €97.18/ha/yr semi-intensive; call open 2026-07-24 to 2026-09-06 | Operational; `auto` | | **Cuba, Ministry of Agriculture Provision 28/2024** | Cuba | Payment for environmental services for forest carbon; no budget disclosed | In force; `auto` | The corpus records the Danish Energy Agency as the **only government appearing as a buyer in the deals layer at all**: 4 deals, 165,000 tonnes, zero disclosed value. ### Tier 2 — Credit-purchase programmes: small, and mostly pre-award - **US DOE Carbon Dioxide Removal Purchase Pilot Prize** (FECM). Corpus: up to **$35m total**, 24 Phase 2 semifinalists, **up to 10 winners at up to $3m each**, "preserved with $45m FY26 appropriation." DOE's own funding page confirms the 24 semifinalists shared **$1.2m** (announced 2024-05-28). The corpus's separate `DAC Commercial CDR Purchase Pilot Prize` record notes the competition is **"closed to new competitors pending administration guidance."** I could not verify any Phase 3 purchase agreement having been executed. - **Canada federal CDR procurement** (Greening Government Strategy). Corpus: **at least CAD $10m of purchases by 2030**, announced Oct 2024, RFI Feb 2025, "competitive award pending (pre-award as of 2026-08)." Web search fills the gap: Carbon Pulse reported on **12 March 2026** that Ottawa launched a competitive process for at least **C$10m (~US$7.3m)** of durable removal credits from Canadian projects, and trade press describes **up to five standing offers across five technology streams, valid to 31 March 2029**, with projects and storage required to be in Canada. So the tender is live; I found no award announcement. - **EU CRCF Buyers Club**. Announced November 2025 under the Bioeconomy Strategy; a hybrid public-private demand-aggregation platform, with **initial purchases of permanent removals targeted by December 2026**. No budget attached. Heatmap's December 2025 outlook described the proposal as light on detail — consistent with the corpus record. ### Tier 3 — Proposed, and where the real money would be - **EU ETS carbon removals mechanism** (Article 9c / COM(2026) 616). The corpus holds two overlapping records: **250m additional allowances (contingency 260m) auctioned over 2031–2040**, proceeds ring-fenced for the Commission to buy CRCF-certified **DACCS and BECCS removals as sole buyer**, via competitive tender or reverse auction with **ex-post payment on delivery**; **biochar explicitly excluded** on evidentiary grounds; scale-up to **48 Mt/yr by 2040**. Budget estimates in the records range **€29–43bn (PRIMES) to €46–65bn (POTEnCIA)**. A policy event dated **2026-07-17** records the Commission publishing the review proposal and entering the ordinary legislative procedure. One record carries an important caution: 250 Mt is a ceiling, not a financed commitment, and the legal text does not make allowance release conditional on removal delivery. - **UK GGR Business Model** and **Power BECCS Business Model** (DESNZ). CfD-style top-up payments; full documentation including a 380-page terms and conditions published 2025-08-27; HyNet Track-1 negotiation list published 2025-08-05 with **Ince BECCS (Evero) as Priority** (217,000 t/yr, retrofit by 2029) and **Climeworks' Silver Birch DACCS as Standby**. The corpus records status as `proposed` on absence-of-evidence grounds: **no signed GGR contract could be found as of August 2026**. A 2026-07-17 government response to the Whitehead review reaffirmed the model and committed to integrating GGRs into the UK ETS "as soon as practicable." - **Switzerland** — KlV amendment consultation opened 2025-10-29, closed 2026-02-12; first roadmaps planned for the 2028–2032 legislature. No adoption verified. - **US federal legislation** — the **Carbon Dioxide Leadership Act** was reintroduced the week of 1 Aug 2026 by Reps. Tonko and Peters and Sens. Whitehouse and Coons; it would require DOE to buy an increasing volume of durable removal under a declining per-tonne price ceiling with a set-aside for newer methods. No dollar figures in the record. - **What died**: **California SB 643** passed the Senate 37-0 and the Assembly 78-1, was **vetoed by Governor Newsom on 2025-10-13** (~$50m programme; veto message cited budget pressure and pointed to SB 840's $85m), and the veto was **sustained 2026-03-02**. **Luxembourg's L-NET** feed-in tariff, deposited November 2022, is still recorded as proposed. --- ## 2. How the money compares to private buying **Corpus deals layer, all years (2021 → 2026-08-28), exact over every record held:** - **385 deals, 137.0 Mt contracted** - **Only 44 deals disclose a value**, totalling **US$1.281bn**, covering **2,067,944 tonnes — 1.5% of contracted tonnes** - **112 deals disclose neither tonnes nor value** - 296 of the 385 are `auto` (machine-extracted, no spot-check) Top buyers by contracted volume: | Buyer | Deals | Tonnes | Disclosed value | |---|---|---|---| | Microsoft | 61 | 96,349,302 | $210m (1 deal) | | Meta | 3 | 8,726,000 | none | | Climeworks (as buyer) | 3 | 3,590,000 | none | | Aviva | 2 | 2,675,000 | none | | Frontier | 50 | 1,896,950 | **$740m across 22 deals** | | Altitude | 5 | 1,305,000 | none | | Google | 9 | 953,500 | $10m (1 deal) | | **Danish Energy Agency** | **4** | **165,000** | **none** | By year: 2024 — 127 deals / 37.9 Mt; 2025 — 133 deals / 69.6 Mt; 2026 to 28 Aug — 65 deals / 20.2 Mt. **The comparison, stated carefully:** 1. **On credit purchase, government is ~1–2% of the market at most.** Add up every government credit-buying programme with a number attached — DOE's $35m ceiling, Canada's C$10m, the Danish NECCS Fund's 160 kt/yr — and you are in the low tens of millions of dollars and low hundreds of thousands of tonnes. Frontier alone has **$740m disclosed across 22 deals** in the corpus. CDR.fyi's April 2026 buyer analysis (via Carbon Herald) puts **Microsoft at 36,439,157 tonnes, or 78.5% of all disclosed durable CDR tonnes contracted as of 13 April 2026**, with Frontier-linked buyers at 4.0%. Government does not register in that ranking. 2. **On total public commitment, government is already larger than the entire private market — but the money is a production subsidy, not a purchase.** Sweden's SEK 36bn (~US$3.7bn) and Denmark's DKK 28.7bn (~US$4.5bn) each exceed the $1.28bn of *all* disclosed private deal value in the corpus by roughly 3–4×. They pay operators per tonne captured and stored; the government does not take title to a credit it can retire. 3. **The EU proposal would change the order of magnitude.** At €29–65bn for 250 Mt over 2031–2040, it is 20–50× the whole disclosed private market to date. It is a legislative proposal entering co-decision as of July 2026, not money. 4. **Private buying is itself one buyer.** CDR.fyi reported Q1 2026 as the largest quarter on record at **2.3 Mt contracted, ~560% of Q1 2025**, but Microsoft accounted for a 1 Mt agreement of that. CDR.fyi's June 2026 update notes ex-Microsoft-and-Frontier buyers were only **17% of contracted tonnes but 90% of delivered tonnes and 94% of retired tonnes** — i.e. the concentration in contracting is not mirrored in delivery. Heatmap noted in December 2025 that excluding Microsoft, 2025 purchase volume actually *fell* by about 100,000 tonnes. --- ## 3. What I would not conclude from this - **The 1.5% price-disclosure rate makes any $/tonne comparison between public and private buying unsafe.** The buyers who publish prices (Frontier, Milkywire, NextGen) are not the buyers who contract the volume. I have deliberately not derived an average price. - **Deals, projects and instruments cannot be summed.** The corpus holds 385 deals, 319 projects and 230 instruments as separate layers; project capacity is stated for a minority and rarely says annual vs lifetime. - **Scope conflation is the biggest measurement problem in this question.** Denmark's CCS Fund, the UK's Drax subsidy adjustment and the HyNet cluster are `cdr-and-ccs`. Headline figures like the UK's £22bn CCS approval or Aalborg Portland's DKK 16.5bn are not removal budgets, and I have not tried to split them. - **Confirmed silence, not confirmed absence.** For 1 June – 17 Aug 2026, **22 of 27 tracked jurisdictions returned no policy events at all**, including the United States, Canada, Sweden, Switzerland, Norway, Germany and Japan. That is a coverage gap. My searches partly filled it — the Canadian tender launch (March 2026) and the fact that **three companies bid into Sweden's second BECCS reverse auction, reported by Carbon Pulse on 18 August 2026**, appear nowhere in the corpus, and Quantum Commodity Intelligence reported Sweden cutting its long-term BECCS spending forecast by 15%. - **Genuinely uncovered by both**: Norway's Longship removal component; Japan's GX-ETS treatment of removals beyond a 2027-10-01 market-opening deadline; subnational and non-OECD procurement; and — most consequentially — how much of any announced envelope has actually been *disbursed* rather than contracted. Nobody publishes that, and the corpus has no field for it.
Judge’s note — Analytically strong and unusually well-structured: the answer's central move — separating governments *buying tonnes* (DOE prize, Stockholm city, Canada's pending CAD10m, Danish Energy Agency as ledger buyer) from governments *subsidising capacity* (Sweden's SEK 36bn envelope / ~SEK 20bn Stockholm Exergi award, Denmark's NECCS and CCS funds, UK GGR Business Model, EU ETS COM(2026)616 with 250 Mt / €29–65bn) — is exactly the distinction an expert needs, and it correctly warns that auction capacity tonnes cannot be added to offtake ledgers. Jurisdictional breadth is good (US federal and California SB 643 veto, Canada, Sweden, Denmark, UK, EU, Romania, Cuba, Luxembourg, Switzerland, Japan), though it never states the ~33-instrument corpus count. Private-vs-public comparison is quantified (385 deals / 137 Mt; Microsoft ~70% of tonnes; top-buyer table; annual split) and the disclosure caveat is handled honestly: 44 of 385 deals disclose value (~1 in 9, matching gold), with an explicit statement that Frontier's $740m dominance makes the dollar sample 'a hard floor and a badly biased sample' — i.e. price claims reflect the disclosing minority. One numeric divergence from the checklist: it reports disclosed-value deals covering 2.07 Mt / 1.5% of contracted tonnes rather than ~11%, so the tonnage-coverage figure may be understated or drawn differently. It also does not surface the ~400 funding-commitment count, citing only 13 DOE funding records. Weakest dimension is the beyond-corpus probe: the answer openly states no web-search tool was available, so there are no primary-source budget figures verified outside the corpus and nothing after the ingest cutoffs; it compensates with a crisp list of the five open questions requiring live checking. Citations are named and linkable but unverified and machine-extracted for most records, which the answer flags. Coverage honesty and clarity are exemplary; the missing live verification and the tonnage-share discrepancy cap the overall score.
I'll start by mapping the corpus's coverage, then dig into procurement instruments and private deal volumes. Now let me quantify the private side and check how the deal ledger splits by buyer and year. ## Bottom line There are two very different things being called "government procurement of carbon removal," and conflating them is where most published comparisons go wrong: 1. **Governments actually buying removal credits** (a public body pays for tonnes and takes delivery). This is real but tiny — on the order of **tens of millions of dollars and well under 1 Mt contracted worldwide**. 2. **Governments paying for removal capacity** via reverse auctions, CfDs and subsidy funds (Sweden, Denmark, UK, and the EU's proposed ETS mechanism). This is **billions**, and it dwarfs disclosed private spending — but it is a subsidy to producers, not a purchase of credits, and its tonnes cannot be added to offtake ledgers. Against both, private corporate buying dominates the *volume* ledger: the Carbon Curve corpus holds **385 deals and 137.0 Mt contracted from 2021 to 28 Aug 2026**, of which **Microsoft alone is 96.3 Mt across 61 deals — roughly 70% of every tonne on record**. ⚠️ **Method note up front:** all figures below come from the Carbon Curve corpus (deals held to 2026-08-28, policy events to 2026-08-17). I had no web-search tool available in this session, so I could **not** run the corroborating searches the corpus explicitly asks for. Nothing after ~1 September 2026 is reflected here, and the corpus describes itself as "a floor, not a census." --- ## Where government purchasing is genuinely happening ### Actual credit purchases (small, and mostly still pre-award) | Programme | Status as of Aug 2026 | Money / tonnes | |---|---|---| | **US DOE CDR Purchase Pilot Prize** (FECM) | Operational but "closed to new competitors pending administration guidance" | Up to **$35m** total across 3 phases; Phase 3 purchase agreements **up to $3m each** for up to 10 winners; 24 semifinalists named June 2024; corpus notes it was preserved with a **$45m FY26 appropriation** ([energy.gov](https://www.energy.gov/fecm/carbon-dioxide-removal-purchase-pilot-prize)) | | **City of Stockholm ↔ Stockholm Exergi** | Signed **4 June 2026** (company announcement 26 May) | **750,000 t over 15 years** (50,000 t/yr) of BECCS, stored via Northern Lights. Described in the source as making Stockholm **the world's fifth-largest buyer of permanent removal** ([Carbon Gap](https://carbongap.org/stockholm-shows-the-way-how-public-buyers-can-boost-the-carbon-removal-market)) | | **Canada federal CDR procurement** (Greening Government Strategy) | Announced Oct 2024, RFI Feb 2025, **competitive award still pending** as of Aug 2026 | **≥CAD $10m by 2030** (~US$7m, my conversion) ([TBS](https://www.canada.ca/en/treasury-board-secretariat/news/2024/10/government-of-canada-commits-to-purchase-carbon-dioxide-removal-services-to-green-government-operations-and-achieve-net-zero-emissions.html)) | | **Danish Energy Agency** | Appears in the deal ledger *as a buyer* | 4 records, **165,000 t** contracted, 2024 | That is essentially the complete list of governments paying for delivered tonnes. Note the single largest is a **city**, not a national government. ### Capacity subsidy — where the real public money is - **Sweden (Energimyndigheten BECCS reverse auction)** is the biggest committed public sum anywhere: a **SEK 36bn envelope for 2026–2046**, with the first auction won by **Stockholm Exergi at ~SEK 20bn over 15 years for ~800 kt/yr** (≈US$2.0–2.1bn at ~9.5–10 SEK/USD, my conversion — the corpus records no USD value). EU state-aid clearance July 2024; a **legal challenge by Danube CCS Ventures** was published in the OJEU on 22 April 2025 but does not suspend the scheme. A second auction is proposed. ([Energimyndigheten](https://www.energimyndigheten.se/en/news/2025/20-billion-to-capture-and-store-over-11-million-tons-of-biogenic-carbon-dioxide/)) - **Denmark**: the **NECCS Fund (DKK 2.5bn, removals-only)** awarded contracts on **17 April 2024** to BioCirc CO2, Bioman and Carbon Capture Scotland for a combined **160,350 t/yr of biogenic CO2, 2026–2032**. Its successor, the **DKK 28.7bn CCS Fund**, concluded its first tender in **June 2026** with **up to DKK 16.5bn to Aalborg Portland (~1.25 Mt/yr from 2030)** — but that is a cement plant, i.e. mostly *avoided* fossil emissions, not removal. ([ens.dk](https://ens.dk/en/supply-and-consumption/ccs-tenders-and-other-funding-ccs-development)) - **UK**: design is finished, contracts are not. The **GGR Business Model** (CfD-style top-up plus T&S cost coverage) published full documentation on 27 Aug 2025 including a 380-page terms document; the HyNet Track-1 negotiation list (5 Aug 2025) named **Evero's Ince BECCS as Priority (217 kt/yr)** and **Climeworks' Silver Birch DACCS as Standby**. The government's 17 July 2026 response to the Whitehead review reaffirmed the model and committed to ETS integration "as soon as practicable." The corpus flags that **no signed GGR contract could be verified as of August 2026**. Separately, note the £80,000 tender for a CDR price-and-supply forecast — a useful reminder of the scale gap between UK CDR *analysis* spending and UK CDR *buying*. - **EU**: nothing bought yet, everything queued. CRCF certification methodologies for permanent removals are adopted, carbon-farming methodologies adopted **10 July 2026**; the **CRCF Buyers Club** (a public-private demand aggregator, not a state purchase) aims for **first permanent-removal purchases by December 2026**. The **ETS review proposal published 17 July 2026 (COM(2026) 616)** would have the Commission centrally buy **250 Mt of CRCF-certified DACCS/BECCS over 2031–2040**, scaling to **48 Mt/yr by 2040**, funded by auctioning 250m extra allowances — modelled budgets of **€29–43bn (PRIMES) to €46–65bn (POTEnCIA)**, with **biochar explicitly excluded**. It is a *proposal* entering ordinary legislative procedure, with purchasing in the 2030s. ### The long tail and the reversals - **Romania** opened the EU's first aquaculture blue-carbon compensation scheme on 24 July 2026 — **€14.285m**, max €174.92/ha/yr, applications to 6 Sept 2026. - **Cuba's** Provision 28/2024 pays for forest carbon retention under a BIOFIN/UNDP framework. - **Luxembourg's L-NET** feed-in tariff has sat "proposed" since November 2022. **Switzerland's** KlV amendment consultation closed 12 Feb 2026 with roadmaps not due until the 2028–32 legislature. - **United States, moving backwards**: California's **SB 643** (~$50m purchase program) passed 37–0 and 78–1, was **vetoed by Newsom on 13 Oct 2025** on budget grounds, and the veto was **sustained 2 March 2026**. The House FY2027 Energy & Water bill (passed committee 20 May 2026) would cut the DOE **Carbon Dioxide Removal line to $15m, a 67% cut**, and reprogram ~$2.8bn of IIJA funds affecting DAC Hubs and DAC prizes. The **Carbon Dioxide Leadership Act** was reintroduced the week of 1 Aug 2026 by Tonko, Peters, Whitehouse and Coons — still just a bill. - **Japan** has no CDR procurement instrument in the corpus; its demand signal is compliance, via the **GX-ETS mandatory phase from 1 April 2026** (100kt+ emitters, offsets capped at 10%). --- ## How the money compares **Private buying, on the corpus record:** | Year | Deals | Tonnes contracted | |---|---|---| | 2023 | 50 | 9.27 Mt | | 2024 | 127 | 37.93 Mt | | 2025 | 133 | 69.57 Mt | | 2026 (to 28 Aug) | 65 | 20.23 Mt | Top buyers: **Microsoft 96.35 Mt** (61 deals), **Meta 8.73 Mt**, **Climeworks 3.59 Mt** (as intermediary), **Aviva 2.68 Mt**, **Frontier 1.90 Mt across 50 deals**, **Altitude 1.31 Mt**, **Google 0.95 Mt**. **The dollar comparison is where honesty is required.** Only **44 of 385 deals disclose a value**, totalling **$1.281bn across 2.07 Mt — 1.5% of all contracted tonnes**. Frontier alone accounts for $740m of that $1.281bn, because Frontier publishes prices and Microsoft does not. So the private-money figure here is a hard floor and a badly biased sample; **CDR.fyi is the fuller transaction scoreboard**. Set against that: - **Direct public credit purchases contracted worldwide** (DOE prize ≤$35m + Canada ~US$7m + Stockholm's 750 kt + Denmark's 165 kt) are **roughly 1% of the disclosed private dollar total and well under 1% of contracted tonnes**. On credit-buying, government is a rounding error. - **Public capacity subsidy** flips the ranking entirely. Sweden's single Stockholm Exergi award (~US$2bn for ~800 kt/yr over 15 years, i.e. ~12 Mt lifetime) exceeds the entire disclosed private deal value in the corpus. Add Denmark's DKK 2.5bn NECCS and the **US DOE's $2.94bn across 13 funding records** (grants and hub awards, not purchases) and public money is comfortably the larger number. - The EU's proposed €29–65bn would, if enacted, be an order of magnitude above anything the voluntary market has committed — but not before 2031. **Stockholm Exergi is the whole dynamic in one project:** ~SEK 20bn of Swedish state auction support, plus Microsoft (3.33 Mt in May 2024, expanded to ~5 Mt in May 2025), Frontier ($48.6m in 2024 and ~$50m in March 2026), the City of Stockholm's 750 kt, Heba, and a ClimeFi-structured buyers' collective. The state de-risks the asset; corporates buy the tonnes on top. That stack — not government purchasing — is what actually got an 800 kt/yr facility to FID and groundbreaking (16 June 2025). --- ## Caveats you should carry forward - **Category error risk:** Swedish/Danish auction tonnes are subsidised *capacity*, and the corpus warns explicitly that project capacity must not be summed against deal tonnes. Denmark's DKK 16.5bn Aalborg Portland award is largely fossil CCS, not removal. - **Currency:** nearly every procurement instrument in the corpus has `monetary_scale_usd: null`. All USD conversions above are mine and approximate. - **Extraction quality:** 296 of 385 deals and 196 of 230 policy instruments are machine-extracted and auto-published without independent verification. - **Staleness:** policy events stop 17 Aug 2026; deals 28 Aug 2026. Twenty tracked jurisdictions returned zero events for May–Aug 2026, which the corpus says means *no record*, not *no activity*. - **What I could not check:** whether Canada has since awarded its CDR contract, whether the DOE prize has been restarted or killed, whether any UK GGR contract has been signed, whether the EU Buyers Club hit its December 2026 first-purchase target, and where the ETS removals mechanism stands in Parliament and Council. Those are the five questions that would most change this picture, and all five need a live web check.
Asked, verbatim: “What prices have been paid for carbon removal, by pathway?”
Judge’s note — The answer is entirely benchmark/survey-driven rather than grounded in the deal record, so it misses the three structural gold items: it never reports per-pathway deal counts across all deals, never quantifies how few deals in each pathway actually disclose a value (single digits), and never identifies a pathway (soil carbon) with deals but zero disclosed prices. It does satisfy the two 'epistemic hygiene' items well: it repeatedly states there is no single market price, that nearly every average is either a survey response or a value-over-volume back-calculation from the disclosing minority, and it does not fabricate a market total or pretend a truncated list is complete. On the beyond-corpus probe it is rich: CDR.fyi/OPIS Jan-2025 per-pathway breakeven/willingness-to-pay table, the May-2026 follow-up survey (spread $107\u2192$98/t, $48/t by 2030), CDR.fyi biochar snapshot market sizing and buyer concentration, Puro/Nasdaq CORCCHAR index levels and regional biochar quotes via TraceX, Microsoft\u2013Exomad 1.24 Mt, 2025 BiCRS purchase/delivery split, sub-$500/t DAC entrant targets, and the CO280 pulp-mill BECCS ~$100/t benchmark \u2014 roughly eight distinct externally-sourced findings, several post-dating a mid-2026 cutoff. Citation quality is good (named publishers with dates) but weakened because a whole section of deal-level prices is admittedly from unverified prior knowledge, and a few of the more specific figures (index levels, DAC entrants) are secondary-sourced and could not be checked. Coverage honesty is a genuine strength \u2014 an explicit 'what we don't know' section covering survey-vs-transaction, dispersion, contracted-vs-delivered, bundling, and a candid admission that search verification was cut short \u2014 though the confident closing 'planning numbers' partly undercut that. Clarity and decision-usefulness are high. Net: a well-organized, honest market-benchmark essay that would be useful to a practitioner, but it does not answer the question from the deal record and therefore misses the sample-size and non-disclosure facts that most change how the numbers should be read.
I'll research current CDR pricing data across pathways.
## Short answer
There is no single "price of carbon removal." What buyers actually pay ranges from roughly **$15–50/t for nature-based removals** to **$100–200/t for biochar**, **~$200–350/t for BECCS**, **~$250–400/t for enhanced rock weathering and mineralization**, and **$500–1,000+/t for direct air capture**. Crucially, most large deal prices are *not disclosed*, so the public evidence base is a mix of (a) survey-based price expectations, (b) registry/index quotes, and (c) a handful of deals where dollar value and tonnage were both published.
An important caveat up front: **my web-search quota was exhausted partway through this research**, so the survey and index figures below are search-verified, while the deal-level prices in the final section are from my prior knowledge and I could not re-verify them in this session. I flag each accordingly.
---
## 1. The best public benchmark: the CDR.fyi × OPIS pricing surveys
CDR.fyi, in partnership with OPIS (a Dow Jones company), has now run two annual durable-CDR pricing surveys of buyers and suppliers across biochar, BECCS, DACCS, enhanced weathering, marine CDR and other pathways.
**Latest round — "Pricing, Attributes, and Blockers: Durable CDR Market Dynamics Through 2030," published 7 May 2026** (survey launched December 2025):
- The average **buyer–supplier price spread narrowed from $107/t to $98/t** year-on-year, and respondents expect it to shrink to about **$48/t by 2030** (CDR.fyi, 7 May 2026; corroborated by Carbon Herald, 8 May 2026).
- **Biochar and BECCS** — the two most active pathways — show the **narrowest current-year pricing gaps but wider projected 2030 gaps**, because tighter supply–demand could keep their prices firm rather than falling.
- Buyers say they will pay **premiums for co-benefits, 1,000+ year permanence, and policy alignment**; 100+ year permanence, supplier transparency and track record are treated as baseline requirements.
- Main blockers remain budget constraints, delivery risk, internal business case, and weak policy incentives.
- CDR.fyi's June 2026 market update frames this as the market moving toward a **"method-specific price stack"** — i.e., a single cross-market benchmark price is becoming untenable.
**First round — "Bridging the Gap," CDR.fyi/OPIS, published 21 January 2025** (the most granular per-pathway numbers publicly available):
| Pathway | Supplier "reasonable profit" price, 2025 | Buyer "expensive" threshold, 2025 | 2030 outlook |
|---|---|---|---|
| **Biochar** | $187/t (breakeven avg $143/t; range $50–$600, most $100–200) | $155/t | Supplier $180/t; buyer $130/t |
| **BECCS** | $301/t (breakeven $232/t) | — | Breakeven avg $212/t (range $80–$370) |
| **DACCS** | — (breakeven **$670/t**) | — | Breakeven $341/t — a projected **~50% fall** |
| **Enhanced weathering** | $349/t | $271/t | Supplier $328/t; buyer $238/t |
| **Mineralization** | — | — | Buyer "cheap/good value" $159/t (down $22) |
| **Marine CDR** | — (breakeven $155/t) | — | Breakeven **rises** to $187/t — the only pathway expected to get more expensive |
Two things stand out. First, across most methods, **the price buyers call "expensive" sits below the price suppliers call profitable** — though CDR.fyi notes many suppliers' *breakeven* prices are below buyers' "expensive" threshold, so a clearing zone exists. Second, the averages hide enormous dispersion (DACCS breakeven answers ranged from under $300/t to over $600/t for 2030).
For context on whether these prices reflect cost: OPIS notes that academic estimates put **BECCS costs at $60–140/t plus $10–20/t for storage** — well below quoted market prices — and that because BECCS plants are built large from the outset (e.g., Stockholm Exergi's planned 800,000 tpa and Ørsted's 430,000 tpa facilities), big technology-driven price declines are unlikely.
**Caveat:** these are *stated* willingness-to-pay and willingness-to-sell from a self-selected sample (73 qualified suppliers in the 2025 round, dominated by biochar), not transaction prices.
---
## 2. Biochar — the only pathway with something like a liquid market price
Biochar is where the most real transaction data exists, because it dominates deal count and delivery. Per CDR.fyi's **Biochar Market Snapshot (9 September 2025)**:
- The BCR market grew from **$14.6M (2022) → $33.9M (2023) → $181.5M (2024)**, a 131.6% CAGR.
- **290 unique buyers** (>10 t transactions), roughly five times the next pathway (enhanced weathering); more than half of tracked purchasers have bought **fewer than 100 tonnes**.
- Average volume per purchaser rose from **542 t (2022) to 762 t (H1 2025)**.
- Microsoft, Google, BCG and JPMorgan account for **57% of BCR purchases**; **Exomad Green alone supplied 60%** of BCR credits sold since 2022.
On observed price levels, a secondary source (TraceX, May 2026, citing CDR.fyi, Sylvera and Puro.earth) reports biochar credits trading at **$125–200/t in 2025–26**, with the **Nasdaq/Puro CORCCHAR index clustering at $125–145/t**, Indian biochar at the low end (**€105–150/t**) and German biochar at **€189–200/t**. Treat these as indicative — I was not able to verify them against the Puro.earth index directly.
Volume context: **1.6 Mt of biochar CDR sold in H1 2025 alone**, driven by Microsoft's purchase of **1.24 Mt from Exomad Green over 10 years** (Carbon Herald, 17 December 2025). That deal's price was not disclosed, but at these index levels it implies a contract in the low hundreds of millions of dollars.
---
## 3. BECCS — now the volume leader, but priced privately
In 2025, **BiCRS methods were 96% of purchase volume, with BECCS alone at 69%**; on the delivery side BiCRS was 91%, with **biochar at 80%** (CDR.fyi, 12 June 2026). So the tonnes being *contracted* are overwhelmingly BECCS, while the tonnes being *delivered* are overwhelmingly biochar.
Essentially **none of the mega-BECCS offtakes disclose $/t**. Publicly, the survey anchors are a **$232/t supplier breakeven and $301/t "reasonable profit"** price for 2025. My prior understanding — unverified here — is that analysts generally model Microsoft's very large BECCS contracts (Stockholm Exergi, Ørsted, AtmosClear, and pulp-mill capture projects) in the **~$150–350/t** band, with pulp-and-paper retrofits at the low end because the CO₂ is already concentrated. Treat that range as an inference, not a disclosed figure.
---
## 4. Direct air capture — the high end, falling but slowly
DACCS suppliers reported a **2025 breakeven of $670/t**, expected to halve to **$341/t by 2030** (CDR.fyi/OPIS). CDR.fyi's June 2026 assessment is blunt that **DACCS development has been slower and more expensive than early expectations**.
New entrants are targeting the sub-$500 mark: Carbon Herald (17 December 2025) reports **Airchive** launched one of the lowest-cost DAC technologies to date, expected to reach **under $500/t**, with **DACLab's Kelvin system** at a similar level. These are announced targets, not prices paid.
From prior reporting (unverified in this session): actual DAC purchase prices have generally been **$500–800/t for large corporate offtakes** from 1PointFive's STRATOS, and **~$1,000/t or more for Climeworks' early corporate and retail credits**. Early Stripe/Frontier prepurchases were higher still.
---
## 5. Enhanced weathering, mineralization and marine CDR
- **Enhanced rock weathering:** suppliers sought **$349/t (2025)** falling to **$328/t (2030)**; buyers viewed **$271/t** as expensive, with the "cheap/good value" point expected to weaken **$38 to $149/t by 2030**. ERW is the **second-most-common pathway by unique buyer count**, but far behind biochar. CDR.fyi attributes the friction to **MRV and regulatory confidence** rather than price alone.
- **Mineralization:** buyers' "cheap/good value" price expected at **$159/t in 2030** (down $22). CDR.fyi calls it "promising, with progress varying by submethod."
- **Marine CDR / ocean alkalinity:** the outlier — supplier breakeven **rising from $155/t to $187/t by 2030**, the only method where suppliers expect costs to go *up*. Transacted volumes are tiny, so any "market price" here is essentially a handful of R&D-stage offtakes.
---
## 6. Deal-level prices (from prior reporting — flagged as unverified here)
Because these are the numbers people usually want, here are the widely cited disclosed-deal benchmarks, all of which I would want to re-verify before you rely on them:
- **Frontier prepurchases (2022–24):** roughly **$300–500/t** for ERW (Lithos, Terradot) and biomass pathways (Arbor, Charm bio-oil), derived by dividing announced contract value by contracted tonnes.
- **Frontier's pulp-mill BECCS offtake (CO280, 2025):** reportedly around **$100/t or below** — by far the cheapest engineered-removal deal publicly benchmarked, reflecting the concentrated biogenic CO₂ stream.
- **Charm Industrial bio-oil (Stripe, 2021–22):** ~**$600/t**, declining since.
- **Afforestation/reforestation removals (e.g., Microsoft–Chestnut Carbon):** typically **$15–50/t** — an order of magnitude below engineered pathways, which is precisely why durability-adjusted comparisons matter.
---
## What we don't know — the honest gaps
1. **Price disclosure is the exception.** The largest contracts of 2025 (Microsoft–Exomad Green's 1.24 Mt; the multi-million-tonne BECCS offtakes) were announced with volumes and terms but no $/t. Almost every "average price" in circulation is either a survey response or a value ÷ volume back-calculation from the minority of deals that publish both.
2. **Survey ≠ transaction.** The CDR.fyi/OPIS numbers are stated thresholds ("breakeven," "reasonable profit," "expensive"), not clearing prices, from a small, self-selected, biochar-weighted sample. CDR.fyi itself warns the results "may not reflect the entire industry."
3. **Enormous within-pathway dispersion.** Biochar breakeven answers spanned **$50–600/t**; BECCS 2030 breakeven spanned **$80–370/t**. A single pathway average can be actively misleading.
4. **Contracted vs. delivered prices differ.** Long-dated offtakes for 2029–2035 delivery are not comparable to spot credits available today, and CDR.fyi's own data show a wide gap between the two (BECCS 69% of purchases vs. biochar 80% of deliveries in 2025).
5. **Bundled value.** Prices embed premiums for co-benefits, 1,000+ year permanence and policy alignment, plus discounts for delivery risk — so identical headline prices can represent very different products.
6. **My verification was incomplete.** I could not retrieve the Puro.earth/Nasdaq CORC index directly, nor confirm individual Frontier/Microsoft/1PointFive deal economics, before hitting the search limit.
**Practical takeaway:** if you need a defensible planning number today, use roughly **$130–180/t for biochar, $230–300/t for BECCS, $300–350/t for ERW, and $500–700/t for DACCS**, and state explicitly that these derive from the CDR.fyi/OPIS surveys (Jan 2025 and May 2026) rather than from an observable exchange price.
Judge’s note — The answer hits every checklist behaviour: it frames the whole corpus (44 of 385 deals valued, 25 with a per-tonne price, 1.5% of contracted tonnes), gives a per-pathway breakdown across all deals rather than a page of them, reports the tiny valued-deal counts per pathway (DAC 6/72, BECCS 8/48, ERW 6/35, biochar 5/77), explicitly names pathways with zero priced tonnes (soil carbon, afforestation, blue carbon, mineralisation, DOC), and repeatedly states that these figures describe the disclosing minority — reinforced by the Microsoft-disclosure-bias point — rather than a market price. It avoids summing pathway totals into a spurious market total and flags derived-vs-quoted arithmetic explicitly. Beyond-corpus material is unusually rich: Puro/Nasdaq CORCX and CORCCHAR levels, Nasdaq CO2X biochar band, S&P Platts biochar assessments, Sylvera 2026 indicative averages, CDR.fyi×OPIS 2025 and 2026 survey breakevens and price-gap, CDR.fyi's Microsoft share stat, Danish NECCS award levels, the Swedish BECCS auction implied price, Indigo Ag soil-carbon commitments, and retail forward-offer ranges — roughly ten distinct findings. Weaknesses: the per-pathway deal counts diverge materially from the reference figures (e.g. DAC 72 vs ~86, biochar 77 vs ~89, other-emerging 43 vs ~100), which may reflect a finer taxonomy but raises accuracy doubts; the soil-carbon deal count (13) is not given; and sources are named but not linked, with several very recent deal specifics (pHathom Feb 2026, NULIFE Dec 2025) hard to verify. Structure and decision-usefulness are excellent, and the uncertainty section is exemplary.
I'll start by mapping what the corpus covers, then pull the deal-level price evidence by pathway. Now let me pull the actual priced deals and published price benchmarks. ## Bottom line There is no single "price of carbon removal." What exists is a small set of disclosed transactions — heavily skewed toward one buyer coalition (Frontier) that publishes a dollar figure on nearly every deal — plus index prices for the one commoditised pathway (biochar). In the Carbon Curve corpus, **only 44 of 385 recorded deals disclose a value, and just 25 carry a per-tonne price; those cover 2.07 Mt of 137 Mt contracted, or 1.5% of tonnes**. The largest buyers by volume (Microsoft above all) disclose nothing. So every number below describes the disclosing minority. --- ## Disclosed per-tonne prices, by pathway (corpus records) | Pathway | Disclosed deals | Prices actually paid | |---|---|---| | **DAC** | 6 of 72 deals valued ($129.6M / 281,236 t) | Frontier–Phlair, 27 Feb 2025: **$651/t** (47,276 t, $30.6M, delivery 2027–30). Frontier–280 Earth, 11 Jul 2024: **$650/t** (61,571 t, $40M). Google–Holocene, 10 Sep 2024: **$100/t** (100,000 t, $10M) — described at the time as the lowest DAC price on record, and stackable with the US 45Q credit of $180/t. LEGO–Climeworks (Mar 2024): $2.4M, tonnage undisclosed | | **BECCS / waste-to-energy** | 8 of 48 deals valued ($269.6M / 748,444 t) | Frontier–CO280, Dec 2024: **$214/t** (224,446 t). Frontier–Hafslund Celsio, Apr 2025: **$316/t**. Frontier–Arbor, Jul 2025: **$353/t**. Frontier–Reverion, Nov 2025: **$427/t**. Karbon-X–Drax, Mar 2024: **$350/t** (25,000 t, resale intent). C-Zero–Drax, May 2023: **$300/t** (2,000 t MoU). JPMorganChase–CO280, May 2025: **"under $200/ton"** for 450,000 t over 13 years — the cheapest large durable deal on record here | | **Enhanced rock weathering (ERW)** | 6 of 35 deals valued ($117.8M / 328,161 t) | Frontier–Terradot, Dec 2024: **$300/t** (~90,000 t, Brazil). Frontier–Eion, Mar 2025: **$419/t** (80,892 t, US olivine). Frontier–Alt Carbon, Sep 2024: $500k for ~1,851 t (**~$270/t implied**, my arithmetic, not a quoted price) | | **Ocean alkalinity enhancement (OAE)** | 2 of 14 deals valued ($63.4M / 189,456 t) | Frontier–Planetary Technologies, Aug 2025: **$272/t** (116,964 t — largest OAE offtake at announcement). Frontier–CREW Carbon (wastewater alkalinity), Dec 2024: **$447/t** (72,492 t) | | **River alkalinity enhancement** | 1 of 2 | Frontier–CarbonRun, Sep 2024: **$458/t** (56,753 t) — the world's first river-liming offtake | | **Biomass injection (bio-oil / slurry)** | 4 of 11 valued ($157.5M / 388,924 t) | Frontier–Vaulted Deep, May 2024: **$382/t** (154,146 t). Frontier–NULIFE GreenTech, Dec 2025: **$362/t** (122,775 t) | | **Biomass storage (burial/vaulting)** | 1 of 7 | Frontier–Graphyte, Nov 2025: **$556/t** — but only a 450-t prepurchase, so not a scale price | | **Biochar** | 5 of 77 deals valued, and only 10,000 t of 6.2 Mt priced | Essentially no usable deal-level price in the corpus — see index prices below | | **Afforestation/reforestation, IFM, soil carbon** | AR: 4 valued ($347.3M) but **zero priced tonnes**; IFM: 2 valued ($102.2M / 64,000 t); soil carbon: none | No reliable per-tonne figure. Indigo Ag has publicly stated soil-carbon buyer commitments at **$60–$80+/t** | | **Other emerging** | 5 of 43 | Frontier–pHathom, Feb 2026: $500k for ~510 t (**~$980/t implied**, small prepurchase) | A structural read of that table: **Frontier's firm offtakes cluster at $270–$460/t for biomass, ocean and rock pathways, and $650/t for DAC**, all for first-of-a-kind plants delivering 2027–2030. Where a project is a retrofit on existing industrial infrastructure (CO280's pulp mills), prices fall to $200/t or below. ## Public procurement and subsidy price points (a different kind of "price") - **Denmark's NECCS fund** awarded biogenic-CCS contracts in April 2024 at stated support levels of **DKK 968.5/t** (BioCirc, 130,700 t/yr), **DKK 1,117.5/t** (Bioman) and **DKK 2,600/t** (Carbon Capture Scotland) — roughly $140, $162 and $377/t at ~6.9 DKK/USD (my conversion). - **Sweden's BECCS reverse auction** (Energimyndigheten) awarded Stockholm Exergi ~SEK 20bn over 15 years for ~800 kt/yr, which implies on the order of **~SEK 1,700/t (~$175/t)** — again my arithmetic on the stated envelope, not a published unit price. ## Index and survey benchmarks (published by others; verified on the web today) - **Puro.earth / Nasdaq CORC indexes**: the composite CORCX stood at **€138.35** in July 2026 (up 3.93% m/m) with the biochar index **CORCCHAR at €129.21**; Puro's live page currently shows CORCX ~**€137.17** and CORCCHAR ~**€130.14**. Nasdaq's CO2X reference shows biochar consistently in the **$125–$145** band across regions and feedstocks. - **Price reporting agencies**: S&P Global/Platts assessed US biochar at ~**$150/t** for 2025 delivery in October 2025 (~$148 for 2026 delivery), and had assessed **$148/tCO₂e** in May 2025 around the Microsoft–Exomad Green deal (1.24 Mt over ten years — the largest biochar contract to date, price undisclosed). - **Sylvera's 2026 pricing guide** gives indicative averages of **biochar ~$177, ERW >$200, BECCS ~$389, DAC >$500**. - **CDR.fyi × OPIS pricing survey (May 2026)**: the average buyer–supplier price gap across durable methods narrowed from **$107/t to $98/t**, and respondents expect ~**$48/t by 2030**. The 2025 edition gave method detail: supplier breakeven of **$670/t for DACCS in 2025 falling to $341/t by 2030**, **BECCS $232/t breakeven ($301 for "reasonable profit")**, **biochar $143/t breakeven** with suppliers wanting $187/t while buyers called $155/t expensive, and ERW buyers expecting ~$149/t by 2030. - Retail/forward-offer catalogues run higher and wider — one European marketplace quotes **DAC at €450 to >€1,000/t** in 2026 forward offers and ERW around €400/t. ## Where the uncertainty and gaps are 1. **Disclosure bias is the dominant distortion.** Frontier publishes a price on almost every deal; Microsoft — which CDR.fyi says accounted for **78.5% of all disclosed durable CDR tonnes purchased as of April 2026** — publishes none. So the "market price" you can see is largely the price of small-to-mid, first-of-a-kind offtakes, not the price of the megatonne contracts that set actual volume. 2. **Derived vs. quoted.** Most per-tonne figures above are total value ÷ contracted tonnes, computed after the fact, not prices the parties quoted. They ignore delivery risk, payment schedules, prepayment discounts, option value and (in Holocene's case) tax credits stacked on top. 3. **Contracted ≠ delivered.** These are forward prices for 2026–2030 delivery from plants that in several cases are not built. Several suppliers here (Holocene, Ørsted's Kalundborg BECCS) have since had well-documented setbacks; a contracted price is not evidence of a delivered tonne. 4. **Pathways with no price signal at all in the corpus**: surficial mineralisation, direct ocean capture, blue carbon, soil carbon and afforestation all have zero priced tonnes recorded, despite AR alone accounting for 69.6 Mt of contracted volume. 5. **Freshness**: corpus deals run to 28 Aug 2026 (last ingest 1 Sep 2026). The index and survey figures above come from web sources dated through mid-2026 and should be re-checked against Puro's live index page and CDR.fyi's quarterly updates before being used commercially. CDR.fyi's public reports remain the fuller transaction scoreboard; this corpus is explicitly a floor, not a census.
Judge’s note — The answer nails the core epistemics the question is testing: it frames the whole record (385 deals, only 44 with disclosed value, 1.5% of contracted tonnes), gives a per-pathway split across the full dataset rather than a visible page, reports per-pathway disclosure counts that are single-digit, flags pathways with deals but no derivable price (soil carbon 0/14; afforestation with dollars but zero attached tonnes), and states repeatedly that the observed prices describe a disclosing minority dominated by Frontier (22 of 44) rather than a market clearing price. It also avoids the trap of summing per-pathway values into a headline 'market price'. The structural warning that the largest-tonnage deals (Microsoft's multi-Mt offtakes) are price-silent is exactly the right caveat. Weaknesses: several per-pathway deal counts diverge materially from the reference figures (biochar 77 vs ~89, DAC 72 vs ~86, BECCS 48 vs ~55; DAC disclosers 6 vs ~8), so while the aggregation behaviour is correct, some of the specific numbers appear off — a reader auditing the corpus could be misled. The opening lines are leftover tool-narration clutter. Otherwise citations are strong and named (Puro CORCX/CORCCHAR, Nasdaq CO2X, Indigo, CDR.fyi×OPIS, Sylvera, Danish Energy Agency strike prices, ESG Today), with explicit provenance/review-status flags on which datapoints are machine-extracted versus human-checked. Beyond-corpus contributions counted: (1) CDR.fyi June 2026 market update method-mix/pricing-accessibility framing, (2) OPIS/CDR.fyi supplier survey breakeven and willingness-to-pay figures per method, (3) Sylvera 2026 pathway averages, (4) independent 2026 forward-offer ranges (Regreener/Senken), (5) OPIS stats on buyers paying >$500/t and concentration of >$1,000/t volume. It also usefully reconciles disagreement between these benchmarks and the corpus-derived prices instead of averaging them. Honesty section is exemplary (contract vs delivered price, 45Q and Nordic subsidy distortion, post-ingest blind spot, auto-extraction risk).
I'll work through this systematically — starting with what the corpus covers, then the price data itself, then checking the web for what's moved since. Coverage noted — importantly, only 44 of 385 deals disclose a value, covering 1.5% of contracted tonnes. Now the breakdown: Now the actual per-tonne figures. Pulling priced deals by pathway plus the published price indices: Now the published price indices and Frontier's portfolio (the single largest source of disclosed per-tonne prices): I have the corpus picture. Now — required step — checking the web for what the corpus flags as outside it: ## Short answer Almost nobody discloses. Of **385 deals** in the Carbon Curve corpus (2021 → 28 Aug 2026, last ingest 1 Sep 2026), only **44 disclose any value**, and those cover **2,067,944 of 137,009,257 contracted tonnes — 1.5% of volume**. Every price below therefore describes a disclosing minority, and that minority is dominated by one buyer: **Frontier accounts for 22 of the 44 value-disclosing deals**. What follows is closer to "Frontier's price curve plus a handful of leaks" than a market price. --- ## 1. Prices visible in the deal record, by pathway Corpus totals per pathway come from one `aggregate_records` call over all 385 deals. Per-tonne figures marked *(derived)* are `usd_amount ÷ tonnes` computed by the database, not quoted by the source. | Pathway | Deals | Disclosing value | Tonnes priced | Observed $/t | |---|---|---|---|---| | **DAC** | 72 | 6 | 281,236 | **$100 – $651** | | **BECCS** | 48 | 8 | 748,444 | **~$140 – $427** | | **ERW (ag)** | 35 | 6 | 328,161 | **$300 – $419** | | **OAE** | 14 | 2 | 189,456 | **$272 – $447** | | **Biomass injection** (bio-oil/slurry) | 11 | 4 | 388,924 | **$362 – $382** | | **River alkalinity (RAE)** | 2 | 1 | 56,753 | **$458** | | **Biomass storage** (vaulting) | 7 | 1 | 450 | **$556** | | **Biochar** | 77 | 5 | 10,000 | *no usable per-tonne* | | **Afforestation/reforestation** | 35 | 4 | **0** | *none derivable* | | **Soil carbon / surficial min. / DOC** | 14 / 13 / 4 | 0 | 0 | *none* | **The individual datapoints:** - **DAC — $650–651/t at the frontier, $100/t as the outlier.** Frontier↔Phlair (27 Feb 2025, 47,276 t, $30.6M → **$651/t**, derived) and Frontier↔280 Earth (11 Jul 2024, 61,571 t, $40M → **$650/t**, derived). Against that, Google↔Holocene (10 Sep 2024, 100,000 t, $10M) was quoted by Google at **$100/t** and described as the lowest DAC price on record at the time — a number that only works alongside the US 45Q credit. Both Frontier records are `approved`; the Holocene record is `auto` (machine-extracted, no human spot-check). - **BECCS — the cheapest durable tonnes with real volume.** Frontier↔CO280 (17 Dec 2024, 224,446 t, $48M → **$214/t**); Frontier↔Hafslund Celsio (1 Apr 2025, 99,998 t, $31.6M → **$316/t**); Frontier↔Arbor (8 Jul 2025, 121,002 t, $41M → **$353/t**); Frontier↔Reverion (25 Nov 2025, 96,000 t, $41M → **$427/t**). JPMorganChase↔CO280 (21 May 2025, 450,000 t) was reported by ESG Today at **under $200/tonne** — the lowest large-volume durable price on record here. Drax quoted **$350/t** to Karbon-X and **$300/t** to C-Zero. - **BECCS public procurement is cheaper still.** The Danish Energy Agency's NECCS awards (17 Apr 2024) name explicit strike prices: **DKK 968.5/t** to BioCirc CO2 (130,700 t/yr), **DKK 1,117.5/t** to Bioman, **DKK 2,600/t** to Carbon Capture Scotland — roughly **$140, $160 and $375/t**. These are subsidy support levels, not voluntary-market clearing prices. - **ERW — $300 to $419.** Frontier↔Terradot (Dec 2024, 89,998 t, $27M) was reported at **$300/t**; Frontier↔Eion (25 Mar 2025, 80,892 t, $33M → **$419/t**, derived). - **Ocean pathways — $272 to $458.** Frontier↔Planetary (26 Aug 2025, 116,964 t, $31.3M → **$272/t**, the largest OAE offtake at announcement); Frontier↔CREW Carbon (17 Dec 2024, 72,492 t, $32.1M → **$447/t**); Frontier↔CarbonRun river liming (23 Sep 2024, 56,753 t, $25.4M → **$458/t**). - **Biomass injection — tightly clustered.** Frontier↔Vaulted Deep (1 May 2024, 154,146 t, $58.3M → **$382/t**); Frontier↔NULIFE GreenTech (18 Dec 2025, 122,775 t, $44.2M → **$362/t**). - **Biomass vaulting — $556/t.** Frontier↔Graphyte (20 Nov 2025, 450 t, $250k) — a tiny prepurchase, not a market price. - **Nature-based is a blank.** 35 afforestation deals, four with a disclosed dollar value totalling $347.3M, and **zero tonnes attached to any of them**. No per-tonne figure can honestly be derived. **A structural warning:** the largest deals by tonnage disclose nothing. Microsoft↔AtmosClear (6.75 Mt), Microsoft↔Stockholm Exergi (5.0 Mt), Microsoft↔C2X (3.6 Mt), Microsoft↔Ørsted (2.76 Mt) — all price-silent. The visible price distribution is skewed toward small catalytic prepurchases; the tonnes that actually move are unpriced. --- ## 2. What published indices say (other people's measurements, held as-is) The corpus holds 12 market datapoints, reproduced with their publishers' framing and **not** reconciled against the deals above: - **Puro.earth**: composite **CORCX at €138.35** in July 2026 (up €5.23, +3.93% m/m); **biochar CORCCHAR at €129.21** (up €8.54, +7.08% m/m). - **Nasdaq CO2X**: biochar credits consistently **$125–$145**. - **Indigo Ag**: soil carbon forward commitments at **$60–$80+**. - **CDR.fyi × OPIS (7 May 2026)**: the buyer–supplier price *gap* across all durable methods narrowed from **$107/t to $98/t**, expected to reach **~$48/t by 2030**. These fill exactly the holes the deal record leaves — biochar and soil carbon, where disclosure is worst. --- ## 3. What web search adds beyond the corpus freshness dates Running the tool's suggested searches turned up method-level pricing the corpus does not hold: - **CDR.fyi's June 2026 market update** describes durable CDR moving toward a "method-specific price stack," and reports that in 2025 BiCRS methods were **96% of purchase volume** (BECCS 69%) and **91% of delivered volume** (biochar 80%) — pricing accessibility being the stated reason. It also notes non-Microsoft/Frontier buyers were 17% of contracted tonnes but **90% of delivered tonnes**. - **The OPIS/CDR.fyi survey white paper** gives supplier-side economics the deal record cannot: a **2025 BECCS breakeven of $232/mt** and a "reasonable profit" price of **$301/mt**, against academic cost estimates of $60–140/mt plus $10–20/mt storage. For biochar, suppliers wanted **$187/mt** while buyers called **$155/mt** expensive; for enhanced weathering, **$349/mt** sought vs **$271/mt** deemed expensive. - **Sylvera's 2026 pricing report** (July 2026) puts averages at **DAC >$500**, **BECCS $389**, **ERW >$200**, **biochar $177**. - Independent 2026 commentary (Regreener, Senken) puts DAC forward offers at **€450 to >€1,000/t**, with BECCS, biochar and ERW offtakes clustering **€150–€400/t** — consistent with the corpus's $214–$458 band. - OPIS also reports that **only 32 purchasers have ever paid over $500/mt**, and that **98.5% of volume above $1,000/mt** came from Frontier buyers and the Milkywire Climate Transformation Fund. That is the single best explanation for why the corpus's high-end DAC prices look the way they do. **Where they disagree:** Sylvera's $389 BECCS average sits above every derived BECCS price in the corpus ($214–$427, mostly Frontier). Sylvera's $177 biochar average sits above the Puro CORCCHAR (€129 ≈ $150) and above Nasdaq's $125–145. Different baskets, different vintages, different delivery windows — treat the spread itself as the answer, not any single figure. --- ## 4. What none of this covers - **Delivery vs. contract.** Every price above is a contract price for future tonnes, often 2027–2030 delivery. Spot prices for delivered, retired credits are a different and much thinner market. - **Bundled economics.** 45Q, the Swedish BECCS reverse auction (SEK 36bn envelope, ~SEK 20bn to Stockholm Exergi over 15 years) and Denmark's NECCS all mean the buyer's price is not the supplier's revenue. The Holocene $100/t is unintelligible without 45Q. - **Post-1 Sep 2026 deals** are invisible to the corpus by construction; the 2026 web sources above partly cover that window but are secondary. - **Provenance.** 296 of 385 deals carry `review_status: auto` — machine-extracted from the cited source with no human check. Among the priced deals I cited, the Frontier offtakes are `approved`; Google↔Holocene, Karbon-X↔Drax and C-Zero↔Drax are `auto`. - **No ranking or recommendation** is offered here, and the database offers none by design.
Judge’s note — Strong, well-organised answer that leads with the disclosure problem (44 of 385 deals priced, 1.5% of tonnes) and correctly frames every per-tonne figure as derived, catalytic and Frontier-dominated rather than a clearing price. It names soil carbon, surficial mineralization, direct ocean capture and blue carbon as pathways with zero priced deals, avoids summing pathway totals into a spurious market total, and works from the whole corpus rather than a truncated page. Coverage honesty is exemplary: explicit boundary dates, admission that 296/385 records are auto-published, and a pointer to authoritative external sources. It also brings in a good number of outside benchmarks (Puro CORCCHAR/CORCX, Nasdaq CO2X, CDR.fyi x OPIS buyer-seller gap, CDR.fyi Microsoft share, Indigo Ag soil pricing, Reuters durable-vs-avoidance multiple), several with links and dates. Two weaknesses: (1) it never reports the per-pathway distribution of ALL deals on record — the denominator that makes the disclosure sparsity legible — so the reader cannot see that e.g. biochar has ~89 deals against a handful of prices; per-pathway disclosure counts are only implied by enumeration and appear understated for DAC; (2) there is an internal inconsistency in claiming no web-search capability while citing month-specific July 2026 index prints and a June 2026 blog, and several precise figures (index values, m/m deltas) are not independently traceable. Decision-useful and honest, but incomplete on the structural denominators the question invites.
I'll start by checking what the corpus covers, then pull the deal-layer distribution. Now let me pull the actual priced deals and published price indices. ## Short answer Almost nobody publishes a price. Across the 385 offtake deals held in the Carbon Curve corpus (2021 → 28 Aug 2026), **only 44 disclose a value, and those cover 2.07 Mt of 137.0 Mt contracted — 1.5% of tonnes**. Worse, disclosure is not random: Frontier publishes a dollar figure on 22 of its 51 deals, while Microsoft — which CDR.fyi puts at **78.5% of all disclosed durable CDR tonnes purchased as of April 2026** ([CDR.fyi, 12 Jun 2026](https://www.cdr.fyi/blog/durable-cdr-market-update-carbon-unbound-east-coast-2026)) — publishes none. So the price curve below is essentially *Frontier's* price curve plus a handful of index prints. With that caveat, here is what has actually been paid. --- ## Disclosed prices by pathway All Frontier figures are **derived** (contract value ÷ contracted tonnes), not quoted per-tonne rates, and represent fixed-price, take-and-pay offtakes with delivery windows stretching to 2030 — not spot. ### Direct air capture — ~$100 to $651/t | Deal | Date | Tonnes | Value | $/t | |---|---|---|---|---| | Frontier ↔ Phlair (Alberta) | 27 Feb 2025 | 47,276 | $30.6M | **$651** | | Frontier ↔ 280 Earth | 11 Jul 2024 | 61,571 | $40.0M | **$650** | | Google ↔ Holocene | 10 Sep 2024 | 100,000 | $10.0M | **$100** | | LEGO Group ↔ Climeworks | 19 Mar 2024 | undisclosed | $2.4M | — | The Google–Holocene $100/t was billed by Google as "the lowest price on record for DAC" at announcement, and is stackable with the US 45Q credit (~$180/t), i.e. it is not a standalone cost floor. For contrast, Heirloom told TechCrunch at its $150M Series B (4 Dec 2024) that its **current cost was $600–1,000/t**, falling to an expected industry $200–300/t by the early 2030s — a company projection, not a transaction. Corpus-wide DAC blend: $129.6M over 281,236 priced tonnes ≈ **$461/t**, but that number is just three deals pulling in opposite directions. ### BECCS — ~$214 to $427/t, and falling with project maturity | Deal | Date | Tonnes | Value | $/t | |---|---|---|---|---| | Frontier ↔ CO280 (pulp mill retrofit, US Gulf) | 17 Dec 2024 | 224,446 | $48.0M | **$214** | | Frontier ↔ Hafslund Celsio (Oslo WtE) | 1 Apr 2025 | 99,998 | $31.6M | **$316** | | Frontier ↔ Arbor (Louisiana) | 8 Jul 2025 | 121,002 | $41.0M | **$353** | | Frontier ↔ Reverion (SOFC biogas, DE) | 25 Nov 2025 | 96,000 | $41.0M | **$427** | | Frontier ↔ Stockholm Exergi | 18 Jun 2024 | ~180,000 | $48.6M | not derived* | | Frontier ↔ Stockholm Exergi (2nd) | 5 Mar 2026 | n/d | ~$50M | — | *The corpus explicitly declines to derive a price on the 2024 Exergi deal because the announcement left tonnage pending Sweden's BECCS reverse auction (SEK 36bn envelope, 2026–2046). BECCS blend: $269.6M / 748,444 t ≈ **$360/t**. Retrofits of existing biogenic CO₂ streams (CO280) are the cheapest durable tonnes anyone has publicly paid for. ### Enhanced rock weathering — ~$270 to $419/t - **Frontier ↔ Terradot** (Brazil), Dec 2024: 89,998 t / $27.0M = **$300/t** — this one was explicitly reported as "$300 per ton" by [Trellis](https://trellis.net/article/tech-giants-latest-carbon-removal-bet-enhanced-rock-weathering), not just derived. - **Frontier ↔ Eion** (US olivine), 25 Mar 2025: 80,892 t / $33.0M = **$419/t** - **Frontier ↔ Alt Carbon** (Darjeeling), 18 Sep 2024: 1,851 t / $0.5M ≈ **$270/t** (small prepurchase) Blend: $117.8M / 328,161 t ≈ **$359/t**. ### Ocean and river alkalinity — ~$272 to $458/t - **Frontier ↔ Planetary Technologies** (Halifax), 26 Aug 2025: 116,964 t / $31.3M = **$272/t** — largest OAE offtake at announcement - **Frontier ↔ CREW Carbon** (wastewater alkalinity, New Haven), 17 Dec 2024: 72,492 t / $32.1M = **$447/t** - **Frontier ↔ CarbonRun** (river liming, Nova Scotia), 23 Sep 2024: 56,753 t / $25.4M = **$458/t** — world's first river alkalinity offtake Note the gap between paid prices and modelled ones: Frontier's own 2025 OAE Primer cites NOAA's **$50–160/t at scale** projection. Nobody is paying that today. ### BiCRS / biomass injection and burial — ~$362 to $556/t - **Frontier ↔ Vaulted Deep** (slurry injection, Class V wells), 1 May 2024: 154,146 t / $58.3M = **$382/t** - **Frontier ↔ NULIFE GreenTech** (HTL bio-oil into salt caverns, Saskatoon), 18 Dec 2025: 122,775 t / $44.2M = **$362/t** - **Frontier ↔ Graphyte** (Carbon Casting), 20 Nov 2025: 450 t / $250k = **$556/t** (small prepurchase) ### Biochar — the only pathway with a real price index, ~$125 to ~$255/t Biochar deals in the corpus are almost all volume-only (Microsoft–Exomad Green 1.24 Mt, Microsoft–Liferaft 1 Mt, Google–Varaha 100 kt — all price-undisclosed). The visible prices come from indices and micro-deals: - **Puro.earth CORCCHAR Biochar Price Index: €129.21 in July 2026**, +€8.54 (+7.08%) month-on-month ([Puro.earth](https://puro.earth/buy-carbon-credits/removal-certificate-corc/carbon-removal-indexes)) - **Nasdaq CO2X**: biochar credits "consistently in the **$125–$145** range" across regions and feedstocks - **Restord (UK, Biochar Carbon Code): £200/t** — Bennetts Associates (8.6 t delivered against a 50 t prepurchase, Jun 2025) and Glad (1 t, Jul 2025) - remove ↔ Exothermite (28 Aug 2026): $17,000 (€15,000) accelerator-linked prepurchase, tonnes undisclosed ### Early-stage prepurchases — ~$1,000/t Frontier ↔ pHathom (17 Feb 2026): 510 t for $500k ≈ **$980/t**. This is R&D-stage pricing and should not be read as a pathway cost. ### Conventional / nature-based — essentially no disclosed unit prices - **Afforestation/reforestation**: 4 deals disclose $347.3M but **zero** tonnes alongside a value — no per-tonne figure is derivable. - **Improved forest management**: $102.2M against 64,000 priced tonnes; the two figures come from different rows and the implied ~$1,600/t is not credible — treat as not derivable. - **Soil carbon**: no deal in the corpus discloses a value. Indigo Ag has publicly stated buyer commitments at **$60–$80+/t depending on delivery timing**. - **Surficial mineralization, direct ocean capture, blue carbon**: zero priced deals. --- ## Market-level price signals (other people's measurements, not deals) - **Puro.earth CORCX composite index: €138.35 in July 2026**, up from €133.12 in June (+3.93% m/m). - **CDR.fyi × OPIS 2026 Durable CDR Pricing Survey** (7 May 2026): the average **buyer–supplier price gap narrowed from $107/t to $98/t**, with respondents expecting ~$48/t by 2030. That gap is the single best summary of where the market actually is: buyers and sellers are still ~$100/t apart. - Reuters and The Journal Record both report durable removal credits were **~4× more expensive in 2024** than low-priced forest-preservation credits, with ~$10bn spent across spot and offtake. --- ## How to read this — the honest caveats 1. **The sample is Frontier.** Nineteen of the ~25 per-tonne prices above are Frontier offtakes. Frontier deliberately pays a premium to move first-of-a-kind projects to FID (its June 2026 "Growth AMC" added $915M, taking total commitments to $1.8bn). These are not clearing prices; they are catalytic prices. 2. **Derived ≠ quoted.** Most figures are contract value ÷ contracted tonnes. Contracts include milestone payments, minimum-quantity provisions and 6–10 year delivery windows (per Frontier's published offtake template). A $651/t Phlair contract is not a spot quote for a tonne today. 3. **The cheap tonnes are invisible.** Microsoft's mega-offtakes (Stockholm Exergi 3.33 Mt, Exomad Green 1.24 Mt, Liferaft 1 Mt) are the volume-weighted centre of gravity of this market and carry no public price. Any pathway ranking here is a ranking of *what got disclosed*. 4. **Pathway ≠ cost.** BECCS looks cheap partly because retrofits of existing biogenic CO₂ streams are cheap; DAC looks expensive partly because Frontier bought pre-commercial electrochemical DAC. Within-pathway spread ($100–$651 in DAC) exceeds much of the between-pathway spread. 5. **Coverage boundary.** The deal corpus ends 28 Aug 2026 (last ingest 1 Sep 2026); policy events end 17 Aug 2026. I have no web-search capability in this session, so **I could not verify anything announced after those dates**, and I have not independently checked the machine-extracted records (296 of 385 deals are auto-published without spot-check). For anything more recent or more complete, the authoritative sources are CDR.fyi (transaction scoreboard), the CDR.fyi × OPIS pricing survey, Puro.earth's CORC indices, Nasdaq CO2X and AlliedOffsets — this corpus is explicitly a floor, not a census.
Asked, verbatim: “How many companies operate in Canada?”
Judge’s note — The answer does the most important thing the question demands: it goes outside the internal corpus and returns a concrete, sourced count (78 CDR companies per Carbon Removal Canada's Carbon Console), rather than fabricating a tally from a country-sparse entity index. It never presents an entity-table-derived number, and it substitutes the correct external authority. It also layers on a rich set of beyond-corpus findings: the 48 (vs. 45 at launch) project count, 107 kt/yr current capacity, 10.9 Mt (vs. 11.9 Mt) pipeline, the Sept 2025 dashboard launch, named players (Carbon Engineering/Oxy, Deep Sky, Arca), Climeworks' Calgary HQ, True North Carbon's Tamarack DAC first capture (Oct 22, 2025, 2 kt/yr), the March 5, 2026 Advance Carbon Removal Coalition with $100M/$75M figures and member list, and Carbon Gap's 14 Canadian buyers / ~150 kt purchased. Sources are named and datable, though no links are given and the '1,500 tons of DAC globally in 2025' claim looks shaky. The caveats section is genuinely good — single-source dependency, unclear definitional boundaries (developers vs. MRV/consultancies, foreign subsidiaries), scope excluding conventional removal, and an explicit flag of the internal inconsistency between the 101 kt/107 kt and 11.9/10.9 Mt figures. What it misses is the explicit meta-disclosure that the underlying database cannot answer this at all (country populated for only ~22 of ~1,093 entities) and that deals/policy instruments are the jurisdiction-bearing tables; a practitioner relying on the tool would not learn that limitation from this response. Clarity and decision-usefulness are high, with a sensible re-scoping offer at both ends.
I'll interpret this in the context of the CDR industry — let me check what's known about carbon removal companies operating in Canada. (If you meant something else, say the word and I'll re-scope.) ## Short answer: 78 That's the count of carbon dioxide removal (CDR) companies operating in Canada, per Carbon Removal Canada, the Ottawa-based policy organization that maintains the country's main industry tracker. Today, 78 CDR companies operate in Canada and 48 projects are under way or planned, according to Carbon Removal Canada, and most are small in scale, with current removal capacity of 107,000 tonnes a year . Planned projects would account for 10.9 million tonnes annually, equal to the emissions from 2.4 million gas-powered cars. ### Where the number comes from The figure traces to Carbon Removal Canada's "Carbon Console" dashboard. Carbon Removal Canada launched the market intelligence dashboard to map carbon removal companies, technologies and projects across the country , and at launch (reported mid-September 2025) it showed 101,000 tonnes removed to date, another 11.9 megatonnes worth of projects in the pipeline, 45 projects in almost every province, and 78 companies working on biomass, carbon mineralization, ocean-based methods, bioenergy and more . So the company count has been stable at 78 from roughly September 2025 through the March 2026 press coverage, while the project count moved from 45 to 48. ### Who the notable players are Companies developing and planning commercial-scale projects include Carbon Engineering, now a unit of Occidental Petroleum Corp.; Montreal-based Deep Sky Corp.; and Arca of Vancouver. Foreign entrants are also being counted in Canada's footprint: Climeworks chose Calgary as the home for its Canadian headquarters , and True North Carbon, the Canadian subsidiary of U.S. DAC firm CarbonCapture Inc., announced on October 24, 2025 that its Tamarack DAC Project near Innisfail, Alberta reached "First Capture" on October 22, 2025 — designed to remove up to 2,000 tonnes of CO2 per year . Context for the recent citation of the 78 figure: on March 5, 2026, the federal government, RBC, BMO and Shopify were among founding members of the Advance Carbon Removal Coalition, whose members aim to mobilize $100 million in new support for Canadian carbon removal projects by 2030, on top of the $75 million contributed to date . Created as an initiative by Carbon Removal Canada, the coalition brings together the Government of Canada, BMO, ClimeFi, NorthX, RBC, Shopify and Vancity. ### Caveats and gaps you should weigh - **Single-source dependency.** All the press citations of "78" appear to trace back to one tracker (Carbon Removal Canada's Console). I found no independent count from Statistics Canada, CDR.fyi, or a federal registry to triangulate against. - **Definition boundaries are unclear.** It's not specified publicly whether the 78 includes only project developers, or also MRV providers, suppliers, and consultancies; nor whether foreign firms with a Canadian subsidiary or office (Climeworks, CarbonCapture/True North, Occidental-owned Carbon Engineering) are each counted as one Canadian company. - **Scope excludes "conventional" removal.** The count reflects durable/novel CDR, not the much larger conventional land-sector activity. For scale, a 2026 global assessment found roughly 2.2 billion tons removed each year through intentional human activity, about 5% of annual global CO2 emissions, with nearly all of it via "conventional" methods like tree planting, improved forest management and soil sequestration — and direct air capture facilities removed just 1,500 tons globally in 2025 . - **Internal inconsistencies.** "101,000 tonnes removed to date" (Sept 2025) versus "removal capacity of 107,000 tonnes a year" (March 2026) are different metrics; and the pipeline figure moved *down* from 11.9 Mt to 10.9 Mt, suggesting project attrition or a methodology change that isn't explained in the coverage. - **Not a measure of active revenue.** Canadian demand remains thin: an earlier Carbon Gap analysis noted only 14 known companies had made purchases to date, amounting to roughly 150,000 tonnes of durable removal . If you meant total registered businesses in Canada (all sectors), or a different sub-segment such as CCS/point-source capture rather than atmospheric removal, tell me and I'll pull that instead.
Judge’s note — The answer nails the core epistemics: it states outright that the corpus cannot answer the question, explains why (entity layer is a name-and-role index with sparse country tagging, and a jurisdiction filter 'reports what little is tagged, never how many organisations operate there'), refrains from manufacturing a company count from the entity table, and flags that its Canadian-org list is a truncated, partial view. It then correctly goes to the web for the real answer. One accuracy concern: the coverage statistic it quotes ('only 382 of 1,195 entities carry a country') does not match the gold figure (~22 of ~1,093), and it reports deals as jurisdiction-tagged on only 29 of 385 rows rather than treating deal jurisdictions as the reliable substitute — so the substitute is satisfied via the web route rather than the deals/policy route. Beyond-corpus content is substantial and specific: 78 companies / 48 projects from Carbon Removal Canada's Carbon Console via the CER Market Snapshot; ~107 kt/yr current vs ~10.9 Mt/yr planned capacity from the Globe and Mail; the Console's Sept 2025 launch; and three domicile-vs-operations examples (Mission Zero at Deep Sky, CarbonCapture Inc.'s relocation, Climeworks' Calgary HQ). Sources are named with dates and a URL, though not independently verifiable here. Caveats are genuinely useful — definitional scope (utilisation firms inflating the count), domicile vs. operations, vintage/refresh risk, and Deep Sky's dominance of the pipeline making a headcount a poor proxy for capacity. Structure is crisp and decision-useful. Marked down slightly for the questionable corpus statistics and the missed deals/policy framing.
I'll start by checking what the corpus covers, then look specifically at Canada. The corpus flags a key caveat up front: entity records are not a company database. Let me pull what's actually there and check the web. Now the required web checks, since the corpus explicitly says its entity index is not a company census. ## Short answer **About 78 CDR companies are operating in Canada**, pursuing **48 active and planned projects** — that is the figure Carbon Removal Canada maintains on its Carbon Console dashboard, and it is the number quoted by both the Canada Energy Regulator and national press in the first quarter of 2026. - Canada Energy Regulator, *Market Snapshot: The rise of Canada's carbon dioxide removal industry* (published ~20 Jan 2026): Canada "is already home to 78 CDR companies and 48 active and planned CDR projects," visualised on Carbon Removal Canada's Carbon Console. - The Globe and Mail (6 Mar 2026), reporting the launch of a new Canadian CDR investment coalition, uses the same Carbon Removal Canada data and adds scale context: most operators are small, with **current removal capacity of ~107,000 tonnes per year**, while **planned projects would total ~10.9 million tonnes per year**. BOE Report (5 Mar 2026) repeats the 78 companies / 48 projects figure. The underlying source is Carbon Removal Canada's Carbon Console (console.carbonremoval.ca), launched as "Canada's first carbon removal dashboard" in September 2025, which lists companies and projects individually. ## What the Carbon Curve corpus holds (the floor, not the census) The corpus I searched **cannot answer this question directly, and says so explicitly**: its entity layer is "a name-and-role index, not a company database — only 382 of 1,195 [entities] carry a country at all," and it warns that a jurisdiction filter over entities "reports what little is tagged, never how many organisations operate there." What it does hold for Canada: | Layer | Canada count | |---|---| | Projects | **19** — 3 operating, 3 in construction/commissioning, 13 announced or financed, 0 paused/cancelled | | Projects with stated annual capacity | 9, summing **12.04 Mt/yr** (a floor; mixed reporting bases, not comparable to contracted tonnes) | | Deals tagged to Canada | 0 (deals state a jurisdiction on only 29 of 385 rows) | Canadian-headquartered organisations that appear in the corpus include Deep Sky, Svante, Carbon Engineering, CO280, Arca / Arca Climate Technologies, CarbonRun, CarbonCure, CarbiCrete, Carbon Upcycling Technologies, Airex Energy, CO2 Lock Corp., Canada Nickel Company, Canadian Wollastonite, CURA Climate, Carbonyx, Carbon Streaming and True North Carbon — plus investors and institutions (Canada Growth Fund, Cycle Capital, BC Centre for Innovation and Clean Energy, Carbon Removal Canada). That list is truncated alphabetically at "D" in a single page of results, so it is a partial view even of what the corpus tags. ## Caveats on the 78 1. **Definition drives the number.** Carbon Removal Canada's count appears to include CO₂ utilisation and materials firms (CarbiCrete, CarbonCure, Carbon Upcycling) alongside true durable-removal developers. A stricter "durable CDR supplier" definition would give a materially smaller number; a broader "carbon management" definition a larger one. 2. **Domicile vs. operations.** Foreign companies operate removal assets in Canada without being Canadian companies — e.g. Mission Zero Technologies runs a DAC system at Deep Sky's Alberta site, CarbonCapture Inc. moved its first demonstration project from Arizona to Deep Sky Alpha after redirecting from the US in February 2025, and Climeworks has established a Canadian headquarters in Calgary (per the Pembina Institute's CDR Centre page). Whether these are inside the 78 is not stated in the sources I saw. 3. **Vintage.** The 78/48 figure is a Carbon Removal Canada dashboard value cited in January and March 2026. The Carbon Console updates continuously; I was unable to confirm a post-Q1-2026 revision before exhausting my search budget, so treat 78 as accurate as of early-to-mid 2026 and check console.carbonremoval.ca for the live count. 4. **One company dominates the pipeline.** Of the 19 Canadian projects in the corpus, six are Deep Sky's (Alpha in Innisfail, the Manitoba/Pipestone DAC plans at 500,000 t/yr at full scale, a Carbfix mineralization project in Quebec, and a direct-ocean-capture MRV deployment). A company count therefore says little about concentration of actual capacity.
Judge’s note — The answer nails the core epistemic move: it states outright that the corpus entity index is a name-and-role index, is only sparsely country-tagged, and cannot be used as a company census, and it explicitly refuses to quote a corpus-derived count (while showing the Deep Sky triple-row merge as an example of how naive counting inflates). It then pivots correctly to what the corpus can support (Canada-tagged project records with status breakdown, plus an honest warning that the summed 12 Mtpa mixes a lifetime figure and a manufacturing figure, and that records are auto-extracted) and goes to the web for the actual answer. Beyond-corpus substance is strong and well-attributed: the 78 companies / 48 projects figure from Carbon Removal Canada's Carbon Console via the Canada Energy Regulator's January 2026 snapshot, its March 2026 restatement with 107,000 tpa current and 10.9 Mtpa planned capacity, the composition of the count (technology providers, developers, operators), the Console's Feb 25 2026 update stamp, the divergent launch figures (101,000 tpa / ~12 Mt), an unconfirmed newer '85' on the live page, Climeworks' Calgary HQ and the TD/Charm 44,000 t deal, and the Advance Carbon Removal Coalition's $75M+ founding commitments. That is roughly seven distinct findings. Weakness: the corpus statistics it reports (382 of 1,195 entities carry a country) diverge sharply from the reference (~22 of ~1,093), which raises a fabrication/version-mismatch concern and slightly undercuts trust in the quantitative corpus claims; no URLs are given, only source names. Coverage honesty is exemplary — boundary contestability (utilisation vs durable removal), 'operate in Canada' vs 'Canadian-HQ', staleness, absence of any licensing registry, weak nature-based coverage, and an explicit offer to rebuild the list to a stated boundary. Structure is clean and decision-useful: headline number first, provenance, then caveats.
I'll start by checking what the Carbon Curve corpus actually covers, then look at Canada specifically. The corpus itself flags that its entity index can't answer this directly. Let me pull what it *can* show — projects and deals located in Canada. The corpus explicitly warns its entity index can't be counted as a company census. Let me try to page it fully anyway, and run the web searches it suggests. ## Short answer **78** is the figure the Canadian sector itself uses. Canada is home to 78 CDR companies and 48 active and planned CDR projects, as visualised on Carbon Removal Canada's Carbon Console — that is the Canada Energy Regulator's January 2026 market snapshot citing the Console. The same number was repeated in March 2026: 78 CDR companies operate in Canada and 48 projects are under way or planned, according to Carbon Removal Canada; most are small in scale, with current removal capacity of 107,000 tonnes a year, while planned projects would account for 10.9 million tonnes annually . That count comes from a single tracker, and it is worth knowing how it is built: the 78 companies operating in Canada are made up of technology providers, project developers and operators — so it is a supply-side ecosystem count, not a registry of firms selling removal credits. ## What the Carbon Curve database holds (and why it cannot answer this directly) I queried the corpus first. It **declines to answer this question by design**, and says so explicitly: entity records are a name-and-role index rather than a company database, only 382 of its 1,195 entities carry a country at all, and the coverage note instructs that a jurisdiction filter over entities "reports what little is tagged, never how many organisations operate there." So I am not quoting a corpus company count — there isn't a valid one. What the corpus *can* enumerate for Canada: - **19 project records** tagged to Canada: 3 operating, 3 in construction/commissioning, 13 announced or financed, 0 paused or cancelled. Nine state a capacity, summing to ~12.0 Mtpa — but that total is unusable as a removal figure, because it mixes a lifetime number (Canada Nickel's Crawford/NetCarb entry, 500 Mt *total*) and a manufacturing number (Svante's Burnaby filter factory, quoted as filters sufficient for 10 Mt/yr, not removals). Both are `review_status: auto` — machine-extracted from the cited source with no human spot-check. - **Canada-HQ organisations named in the entity index** (partial, alphabetically truncated at "D"): Deep Sky, Svante, Arca Climate Technologies, CarbonRun, CO280, CarbiCrete, CarbonCure, Carbon Upcycling, Carbon Engineering, Airex Energy, CO2 Lock Corp, Canada Nickel, Canadian Wollastonite, Carbon Streaming, CURA Climate, Carbonyx — plus non-company actors (Canada Growth Fund, B.C. Centre for Innovation and Clean Energy, Cycle Capital, Carbon Removal Canada). The index also carried three separate "Deep Sky" rows merged into one on 2026-09-01, which shows how easily a naive entity count inflates. - Operating Canadian facilities on record include True North Carbon's **Tamarack Project** in Innisfail, Alberta (2,000 tpa DAC, commissioning, announced 2025-10-22), **Deep Sky Alpha** (3,000 tpa, commissioning), CO2 Lock's **SAM Project** in British Columbia (operating), and Running Tide's Clayoquot Slope marine trial. ## Caveats on the 78 1. **It may be stale or superseded.** The 78 figure has been quoted from the Console's September 2025 launch through March 2026 unchanged. The Console's projects page shows current capacity of 107K tonnes, 10.9M planned against a 2035 target of 35M, last updated February 25, 2026 . At launch the capacity read differently: there was enough capacity to extract 101,000 tonnes of CO2 per year, with almost 12 million tonnes planned . I saw an "85" on the Console's live page that may be a newer company count, but I could not confirm what it counts — treat 78 as the last figure a named source has published in prose. 2. **The boundary is contestable.** Several firms in these lists are carbon *utilisation* or capture-equipment businesses rather than durable-removal suppliers (CarbonCure, CarbiCrete, Carbon Upcycling, Svante). And Carbon Engineering, counted among the commercial-scale players, is now a unit of Occidental Petroleum . 3. **"Operate in Canada" ≠ "Canadian."** A growing set of foreign firms run Canadian operations that a Canada-HQ count misses. The corpus records CarbonCapture Inc. relocating its first demonstration project from Arizona to Deep Sky Alpha, and Mission Zero Technologies running a DAC system at Deep Sky. Independently, Climeworks established its Canadian headquarters in Calgary, and TD Bank committed to 44,000 tonnes over 10 years from US-based Charm Industrial, partly sourced from Charm's future Canadian operations (January 2026) . 4. **Demand-side and institutional actors are excluded.** The Advance Carbon Removal Coalition launched with founding members including the Government of Canada, BMO, ClimeFi, NorthX, RBC, Shopify and Vancity, which have collectively contributed more than $75 million , targeting $100 million by 2030. ## What neither source covers Neither the Carbon Curve corpus (deals fresh to 2026-08-28, policy events to 2026-08-17) nor the Console gives an audited company register — there is no Canadian licensing regime for CDR firms, so every count is a curator's judgement about who qualifies. Nature-based operators (forestry, soil carbon) are especially unevenly captured. If you need a defensible number for a specific purpose, tell me the boundary you want (durable removal only? credit sellers only? Canadian-incorporated only?) and I'll rebuild the list company by company rather than lean on the 78.
Judge’s note — The answer nails the core epistemic move: it refuses to produce a company count from the entity index, explains why the index is a name-and-role list with sparse country tagging, and pivots to the genuinely jurisdiction-bearing layers (deals, projects) to give a labelled floor of ~17 named organisations with supplier/tonnage/value detail. Coverage honesty is exemplary — it flags the 29-of-385 deal jurisdiction tagging, the zero-row funding artifact, the stale 'announced' project stages, the absence of known Canadian players (Svante, Arca), and its own lack of a web tool, and it names the right external sources to consult (Carbon Removal Canada ecosystem map / Billion Tonne Blueprint, CDR.fyi supplier directory, Foresight/CMC directories). Structure and decision-usefulness are high. Two deductions. First, the quoted coverage note's figures ('382 of 1,195 carry a country') conflict materially with the corpus reality (~22 of ~1,093), so the single most load-bearing quotation in the answer appears misstated or confabulated even though the conclusion drawn from it is right — that undercuts traceability, as do the project-level web links that are asserted without a live search. Second, the beyond-corpus probe is largely unmet: the right answer here is mostly outside the corpus, and the response supplies no actual external roster or count, only the names of where to look plus incidental mentions (Svante/Arca as missing; a Statistics Canada aggregate for the literal reading). Solid, well-calibrated, but stops short of the answer the question really wanted.
I'll start by checking what the corpus covers, then look at Canada specifically. The corpus explicitly warns this question can't be answered from its entity index. Let me gather what it *does* hold on Canada. ## Short answer **The corpus cannot give you a count, and I'd be misleading you if I produced one.** The Carbon Curve entity index explicitly flags this as a question it is not built to answer: > "Entity records are a name-and-role index, not a company database: only **382 of 1,195** carry a country at all. A jurisdiction filter over entities therefore reports what little is tagged, **never how many organisations operate there — do not present this count as an answer to that question**." (carbon-curve coverage note, corpus ingest 2026-09-01) I'm also reading your question as "how many *CDR* companies operate in Canada" — if you meant companies in general, that's a Statistics Canada question (roughly 1.3m employer businesses), not something this corpus touches. ## What the corpus does hold — a floor, not a census Working around the entity gap by counting *named companies attached to Canada-tagged deals and projects*, I get **at least 17 distinct organisations**, spread across three record types: **Suppliers with Canada-tagged offtake deals — 5 companies, 969,768 tonnes contracted, 2024-09-23 → 2026-04-06** (exact over all held deals): | Supplier | Deals | Tonnes | Disclosed value | |---|---|---|---| | North Star Carbon Solutions | 1 | 626,000 | undisclosed | | NULIFE GreenTech | 1 | 122,775 | $44.2m | | Planetary Technologies | 1 | 116,964 | $31.3m | | CarbonRun | 1 | 56,753 | $25.4m | | Phlair | 1 | 47,276 | $30.6m | Four of five disclose value ($131.5m over 343,768 t) — unusually high disclosure versus the corpus overall, where only 44 of 385 deals carry a value. **Project operators and partners on 8 Canada-tagged projects** (1 operating, 2 commissioning/construction, 5 announced): - **Deep Sky** — Deep Sky Alpha, Alberta, commissioning; described as Canada's first sedimentary-storage DAC facility and "the world's first cross-technology DAC platform" ([gasworld](https://www.gasworld.com/story/nikkiso-provides-co2-system-to-support-canada-dac-test-facility/2171639.article)) - **Mission Zero Technologies** (UK) — third DAC system at Deep Sky, recorded **operating** - **CarbonCapture Inc.** (US) — relocated its first demonstration from Arizona to Deep Sky Alpha after redirecting from the US to Canada in Feb 2025 ([Heatmap](https://heatmap.news/carbon-removal/deep-sky-canada-dac)) - **Climeworks** — DAC+S collaboration with Deep Sky, pathway to 1 Mt/yr ([Climeworks](https://climeworks.com/news/climeworks-collaborates-with-deep-sky)) - **True North Carbon** — Tamarack Project, Innisfail, Alberta; 2,000 t/yr solar-powered modular DAC, "first capture" 2025-10-22 ([Newsfile](https://www.newsfilecorp.com/release/271542/True-North-Carbons-Tamarack-Project-Canadas-Largest-SingleTechnology-Direct-Air-Capture-Deployment-Achieves-First-Capture-in-Innisfail-Alberta)) - **Canada Nickel Company** + **NetCarb** — Crawford surficial mineralization ([PR Newswire](https://www.prnewswire.com/news-releases/canada-nickel-partners-with-netcarb-to-advance-next-generation-carbon-sequestration-technology-302476022.html)) - **Carbon RX** + **King Tide Carbon** — biochar/soil carbon, Saskatchewan and western Canadian soils (2024-01-31) - **Rain Cage Carbon** — modular units, multi-jurisdiction (2024-01-24) **Named in Canada-focused media but with no Canada-tagged deal or project record:** Airhive (UK; 1,000 t/yr commercial demonstrator in Alberta, per Carbon Curve ep. 59, 2025-12-10), Mangrove Systems, plus non-commercial actors Carbon Removal Canada (Toronto think tank, a project of the Clean Prosperity Foundation) and investors Amplify Capital and BMO. ## Why 17 is badly wrong as a total - **Deals state a jurisdiction on only 29 of 385 rows.** Canadian suppliers with untagged deals are invisible to the filter. - **The funding layer returned zero rows for Canada** — clearly a tagging artifact, not reality (Deep Sky alone has raised well over C$100m). - **Well-known Canadian CDR firms returned no records at all** in my search — Svante, Arca, and others. Absence here is an indexing gap, not absence from the market. - **319 projects total, 268 still labelled "announced"** because most rows were never revisited; stage means "no later record," not current status. ## Caveat on my own process The corpus repeatedly instructs me to verify against live web sources before answering. **I have no web-search tool available in this session**, so everything above is corpus-only and stops at the 2026-09-01 deal ingest / 2026-08-26 media cutoff. To get a defensible number I'd check: (1) **Carbon Removal Canada's** own ecosystem map and *The Billion Tonne Blueprint* (2026) at carbonremoval.ca — the closest thing to an authoritative Canadian roster; (2) **CDR.fyi's** supplier directory filtered to Canada, which the corpus itself names as "the fuller transaction scoreboard"; and (3) **Foresight Canada / Carbon Management Canada** cleantech directories, which would sweep in adjacent point-source capture firms depending on how you draw the boundary. **Bottom line:** ≥17 named organisations with Canadian CDR deals or projects on record here. The true figure for durable-CDR companies operating in Canada is plausibly several times that; I can't responsibly narrow it without the sources above.
Asked, verbatim: “Which pathways drew funding this quarter?”
Judge’s note — The answer completely bypasses the underlying funding corpus: there is no pathway-level aggregation of the funding layer, no disclosure of the severe pathway-tag sparsity (~366 of ~401 records untagged), and therefore no handling of the trap the question sets. Instead it substitutes an external market narrative, which means all three gold items fail (the third is not satisfied either, since no corpus-derived split is characterized at all — the limitation is simply never confronted). On the beyond-corpus probe it is rich: ClimeFi's Q2 2026 record 2.1 Mt / 14 deals / $676M, the 60% Bio-CCS share flip from 2%, Stockholm Exergi–City of Stockholm 750k t, Microsoft–BioCirc 650k t, biochar at 86% ex-BECCS and >1 Mt, Deep Sky's first certified North American DAC delivery, UnionDAC Teesside, issuances ~193k, Q1 2026 biochar 93%, Frontier and Google pathway mixes, the Microsoft purchase pause and its ~78.5–90% market share, New Market Pitch's $492.6M/18-deal equity dataset with concentration and geography skew, Varaha's $20M tranche, and the Swedish Energy Agency SEK 300M BECCS call — roughly fifteen distinct findings, with some reconciliation between trackers (ClimeFi vs CDR.fyi thresholds, backfill lag). Sources are named and mostly checkable though unlinked, and I cannot verify several of the forward-dated figures. Coverage honesty is genuinely good on the external side — it states plainly that Q3 2026 is incomplete, that searches were exhausted, that trackers disagree, and that volume is not dollars — but it is silent on the data-completeness problem that the question was actually testing, and it quietly redefines 'this quarter' as last quarter. Well organized and readable, but it answers a different question than the one asked and misses the entire methodological core.
I'll research recent CDR funding activity across pathways. Let me run several searches in parallel. Today is September 2, 2026, so "this quarter" (Q3 2026) is still in progress. Let me pull both the in-quarter data and the last completed quarter for context. Search quota is exhausted, but I retained the full result content from earlier calls. Let me extract it. ## First, a definitional caveat Today is **2 September 2026**, so the current quarter (Q3 2026, 1 Jul–30 Sep) is two-thirds elapsed and **no quarterly tracker has published yet**. CDR.fyi's Q2 2026 report and the September monthly recap were not retrievable in my searches, and my search allowance was exhausted before I could pull July/August 2026 deal-level data. **I therefore cannot responsibly tell you what Q3 2026 looks like.** What follows is the last *completed* quarter (Q2 2026) plus the trailing venture picture — flag this framing if you're using it in a client-facing note. --- ## Q2 2026 — the last fully reported quarter **Offtake/purchase funding was dominated by two pathways: Bio-CCS and biochar.** Buyers committed a record 2.1 MtCO₂ in Q2 2026 according to a market review from ClimeFi — an all-time high for the wider buyer base and a 136% increase versus Q2 2025. Those buyers signed 14 deals, the most ClimeFi has recorded, with seven making their first major CDR commitment; a public authority entered at scale for the first time as the City of Stockholm agreed to buy 750,000 credits. The commitments are valued at $676 million, based on publicly disclosed tonnages for deals above 5,000 tonnes. Pathway split: - **Bio-CCS/BECCS** — three large bioenergy-with-CCS deals lifted that route to a 60% share of quarterly commitments, up from 2% in Q1 . Anchor transactions included a 15-year, 750,000-tonne agreement between the City of Stockholm and Stockholm Exergi's BECCS facility, and a 7-year Microsoft offtake with Denmark's BioCirc for up to 650,000 tonnes across five biogas plants, delivering from H2 2026 . - **Biochar** — remained the most sought-after pathway once those Bio-CCS deals are set aside, at 86% of volumes and passing 1 MtCO₂ for the third time in five quarters . Independently, Elimini's Q2 report noted biochar's rise past DAC as the second-largest removal pathway by volume . - **DAC** — funded, but on a much smaller scale and increasingly via delivery rather than headline offtake. Deep Sky delivered North America's first certified DAC credits to Microsoft and RBC from its Alpha facility in Innisfail, Alberta. On the project-development side, Progressive Energy, Mission Zero Technologies and Airhive partnered on UnionDAC, a 60,000 t/yr facility on Teesside targeting 2030 operations within the UK's East Coast Cluster . - Issuances nearly doubled quarter on quarter to 193,000 credits across multiple registries. For contrast, Q1 2026 was almost entirely a biochar quarter: 2.3 million tonnes contracted (560% of Q1 2025 volume), with biochar accounting for 93% of contracted volume and two of the top five suppliers being first-time market participants . Among smaller, quality-led buyers the mix was different — Frontier buyers purchased over 210,000 tonnes across 6 deals (57% BiCRS, 42% mCDR, 1% mineralisation), and Google bought over 108,000 tonnes across 8 deals (75% BiCRS, 25% mCDR) . **The structural story behind Q2:** Microsoft reportedly signalled to suppliers that it was pausing carbon removal purchases, per Bloomberg and Heatmap; Microsoft had represented roughly 90% of the market in 2025 according to CDR.fyi. CDR.fyi put Microsoft at 78.5% of all disclosed durable CDR tonnes purchased as of April 2026. The Q2 record is therefore notable precisely because non-Microsoft buyers committed to a record-high volume of quarterly purchases — demand broadening, not just deepening. --- ## Equity/venture funding by pathway (trailing 24 months to July 2026) A separate question from offtake. Per New Market Pitch's July 2026 analysis of disclosed equity rounds: the dataset covers 18 disclosed deals and $492.6M raised across 18 companies between August 2024 and July 2026, with capital heavily concentrated — the top deal alone is 30.5% of all capital and the top three are 51.6% . Median round size over 24 months was $19.5M; rounds above $50M were rare, with only Heirloom and Terradot crossing that threshold and only Heirloom crossing $100M. By pathway: - **DAC** — remains the biggest venture story, with Climeworks, Heirloom, Avnos, CarbonCapture, Carbon Engineering and Verdox alone representing over $1.6B in disclosed funding . - **ERW/mineralisation** — becoming serious investor categories, with Lithos Carbon, Terradot, Neustark and 44.01 all crossing thresholds . - **Ocean CDR** — still smaller than DAC, but Captura, Ebb Carbon, Gigablue, Planetary Technologies, Equatic and SeaO2 show growing investor interest ; ocean-based startups collectively raised over $125M, with most rounds closing since late 2023 . - **Geography skew** — North America holds a 74.3% capital share against a 44.4% deal share, while Europe is 27.8% of deals but only 12.1% of capital . Recent named rounds: Varaha secured $20M — the first tranche of a planned $45M Series B led by Westbridge Capital with RTP Global and Omnivore — to scale CDR projects in India (Feb 2026, ERW/biochar), and on the public side the Swedish Energy Agency opened a SEK 300 million (~$28 million) funding call for BECCS projects . --- ## Gaps and uncertainties you should carry forward 1. **No Q3 2026 data.** Treat everything above as Q2/trailing. The July and August 2026 CDR.fyi recaps and any Q2 2026 CDR.fyi quarterly report were not retrieved. 2. **Tracker disagreement.** ClimeFi's 2.1 Mt Q2 figure and CDR.fyi's numbers use different inclusion thresholds (ClimeFi's value estimate covers deals above 5,000 tonnes ). Don't mix them in a single time series. 3. **Reporting lag is real.** CDR.fyi backfills later-disclosed deals, with lags typically ranging from weeks to a few months — so early Q3 reads will understate. 4. **Concentration risk is unresolved.** One quarter of record non-Microsoft buying is not yet a trend; the durability of the Q2 broadening is the single most important thing to watch in the Q3 print. 5. **Volume ≠ dollars.** Biochar dominates tonnage at low prices; DAC draws far more equity per tonne. Any "which pathway won" claim must specify the metric.
Judge’s note — The answer nails the core methodological trap: it runs a pathway aggregation, immediately flags that only 5 of 12 Q3 records carry a pathway tag (and 18 of 25 in Q2), quotes the corpus's own undercount warning, and explicitly refuses to let the tagged table stand as the answer ('nearly useless on its own'), instead re-deriving pathways from free text. It does not, however, state the whole-funding-layer untagged ratio (~366 of ~401) — the disclosure is scoped to the quarter(s) examined, which is a weaker version of checklist item 2, and the pathway grouping is quarter-scoped rather than layer-wide. Strong additional value: it catches that the sole `beccs` record is out-of-scope point-source CCS, that the `dac` record is an acquisition not a raise, and that a multi-pathway RFP is mis-tagged as biochar-only — exactly the kind of error the tag-level view hides. Beyond-corpus work is real but partly off-target: it enriches individual rounds with primary sources (CREW syndicate, Carbon to Sea consortium, Paebbl/EIC €97M across six firms, Google award terms and deadline, Amazon/World Bank spekboom bond) and adds market context (CDR.fyi $3.6bn 2021–25, CCarbon DAC $4.35bn / biochar $1.06bn, ClimeFi 36%), but it surfaces no funding rounds from the quarter that are absent from the corpus and performs no reconciliation against public venture databases, while correctly noting the missing final four weeks of September. Citations are named and largely checkable (domains, dates), though a few rest on unlinked report titles. Coverage honesty is exemplary — explicit section on auto-published/unverified records, undisclosed amounts, and hand-assigned pathways. Clarity is high but marred by the leaked planning/thinking preamble before the substantive answer.
I'll start by establishing what the corpus covers and its freshness, then aggregate the funding layer.
Now the quarter-specific slice. Given the corpus runs to early September 2026, "this quarter" = Q3 2026 (Jul 1 onward), so I'll pull that plus Q2 for contrast, and read the individual funding rows.
Twelve funding records in Q3, but 7 carry no pathway tag — I need to read the rows themselves to see what they actually funded.
Twelve funding records, but the corpus ends 2026-09-02 — I need to check the post-ingest window and off-corpus activity on the web.
## What "this quarter" resolves to
The Carbon Curve corpus runs to **2026-09-02** for funding records (deals to 2026-08-28, last ingest 2026-09-01), so "this quarter" = **Q3 2026, and effectively only 2 July – 2 September**. Anything from the last four weeks of September is invisible here.
**Headline: 12 funding records, $41.1M disclosed.** Only 5 of the 12 disclose a value, and — critically — **only 5 of 12 carry a pathway tag at all**. The corpus's own note is blunt: "Multi-pathway rounds carry no pathway tag, so pathway-filtered funding queries undercount." So the tagged aggregate below is nearly useless on its own, and I've re-derived pathways by hand from each record's `pathway_description`.
## The tagged aggregate (what the tool returns)
| Pathway | Records | Disclosed |
|---|---|---|
| biochar | 2 | $10.0M |
| ar (afforestation/reforestation) | 1 | — |
| beccs | 1 | — |
| dac | 1 | — |
That describes 5 records. Here is the actual quarter.
## Every Q3 funding record, pathway assigned by hand
**Ocean / alkalinity / mineralization — the quarter's real centre of gravity**
- **CREW Carbon, $25M Series A** (2026-08-06), led by Burnt Island Ventures with AP Ventures, Sony Innovation Fund, Builders Vision, Idemitsu Ventures and others. Wastewater carbon removal via calcium-carbonate dosing, converting CO₂ in treatment reactors to stable bicarbonate. Stated as "$25 million across equity and non-dilutive capital"; oversubscribed. **Untagged in the corpus, and the largest disclosed round of the quarter.** (crewcarbon.com)
- **Carbon to Sea Initiative, $225k grant** (2026-07-23) to a consortium led by the University of Maryland Center for Environmental Science — a study on integrating OAE into municipal wastewater treatment ("wastewater alkalinity enhancement"), with Black & Veatch, CRSI, DC Water, HRSD, Yale's YCNCC, CREW Carbon and others. Untagged. (carbontosea.org)
- **Paebbl, European Innovation Council STEP Scale Up equity** (2026-08-03), amount undisclosed; the programme writes €10–30M per company and the EIC Fund committed €97M across six deep-tech companies in the round. Carbon-storing supplementary cementitious material, stated at −149 kg CO₂e/tonne. Untagged. (paebbl.com)
**Biochar** — the only pathway with disclosed money on a tag
- **ProClime, $10M from Cadira Capital Management** (2026-07-28), an MOU on capital formation for seven pyrolysis plants across six Indian states, ~35,000 credits/yr initially, pipeline to 50 plants; verification via Puro.earth and Isometric. Note this was **refiled out of the deals lane on 2026-08-28** ("project financing, no credits contracted") — a good example of why deals and funding must not be summed. (proclime.world)
- **Google 2026 Carbon Removal & Superpollutant Elimination R&D Awards** (2026-08-28), up to $450k for the alternative-feedstock biochar stream (MSW, sewage sludge, non-woody residues). This is an **open call, not an award** — no recipient named, applications close 25 September 2026, disbursement by 31 December 2026. Google's own programme page and ESG News put the wider awards at up to $500k per project and list priority areas including OAE, BECCS, biochar and blue carbon.
**Nature-based**
- **Amazon** (2026-07-02) — 1.95 Mt purchase commitment behind the South African spekboom restoration project: 180 million shrubs over 50,000 ha, expanded from 30 million plants/10,000 ha, delivery over "more than a decade"; also enabled a World Bank Spekboom Outcome Bond. Tagged `ar`, no capital figure. (esgtoday.com)
- **Apolownia, €1M seed** (2026-08-03) from LITA, Bpifrance and angels — Paris-based blue-carbon developer, BlueRizon mangrove restoration in Indonesia. Untagged; blue carbon.
**Multi-pathway procurement**
- **VodafoneThree RFP** via Abatable (2026-08-12), delivery 2027–2030 with a 2031–33 option, selection expected Q4 2026. Explicitly spans afforestation, soil carbon, biochar, BECCS and DACS plus a UK Woodland Carbon Code woodland track. Tagged only `biochar` — a clear undercount of four other pathways.
**DAC — consolidation, not fresh capital**
- **Airhive acquires Carbyon** (2026-07-14), terms undisclosed. Combined company runs under the Airhive name; UnionDAC targeted at 60,000 t/yr by 2032. The single `dac` record this quarter is an **acquisition**, not a raise.
**Adjacent / out of scope**
- **Mafix, $5.4M pre-seed** led by Azolla Ventures (recorded 2026-08-21) — CO₂-removing silicon fertilizers, i.e. weathering-adjacent. Untagged, and the record notes the date is a discovery date, not the article date.
- **Kita, undisclosed equity from Tokio Marine & Nichido Fire** (2026-07-31) — carbon insurance and project risk assessment. Market infrastructure, no pathway.
- **Emissions Reduction Alberta TIER, CAD $20M** (2026-09-02) across four projects. Tagged `beccs`, but the corpus flags it `in_scope: false` — the only capture component is point-source CCS at the Mercer Peace River pulp mill. **This is the `beccs` bucket above, and it is not removal.** It also explains why `search_records` returned 11 rows against the aggregate's 12.
## By instrument, and against last quarter
Q3 2026: **equity 6 records / $40.4M**; **grants 3 / $0.675M**; acquisition 1; unclassified 2. Almost all disclosed capital is venture equity, and 62% of it is one round (CREW Carbon).
Q2 2026 was an order of magnitude larger in the corpus: **25 records, $2.35bn disclosed**, dominated by a single **$915M biomass-injection** commitment (2026-06-17) plus $137.4M across four `other-emerging` records. Treat the Q3-vs-Q2 drop as partly real, partly a reporting artefact — 18 of Q2's 25 records were also untagged, disclosure is sparse in both quarters, and Q3 is a two-month window.
## Off-corpus context from the web
Corpus figures are a floor. For the wider picture: CDR.fyi's *Investment Landscape in Carbon Removal 2026* (8 January 2026) reports more than **$3.6bn** invested in durable CDR companies 2021–2025, with DACCS taking the majority of capital while deal activity broadened across methods — and notes seed and Series A rounds dominate while Series B+ stays thin. CCarbon's June 2026 *State of the Sector* puts DAC's cumulative capital base above **USD 4.35bn** while describing biochar as one of the most actively financed pathways, with roughly **USD 1.06bn** disclosed between June 2025 and March 2026. ClimeFi (10 March 2026) expects biochar to supply about **36%** of 2026 durable volume. None of these had Q3 2026 breakdowns published at the time of searching.
## What I can't tell you
- **Pathway attribution is the weak link.** Seven of twelve rows are untagged; my assignments come from free-text descriptions, not the taxonomy.
- **11 of 12 records are machine-extracted and auto-published** (`review_status: auto`, mostly claude-haiku-4-5), not independently verified.
- **Seven of twelve disclose no amount**, including the EIC/Paebbl investment which could plausibly be the quarter's largest.
- **The last four weeks of September 2026 are missing entirely**, as is anything without a public announcement. CDR.fyi's quarterly market updates remain the fuller transaction scoreboard.
**Short answer:** on this evidence, Q3 2026 funding went overwhelmingly to **ocean-alkalinity/mineralization** (CREW Carbon, Carbon to Sea, Paebbl) and **biochar** (ProClime, Google's open call), with **nature-based** appearing as offtake commitments rather than capital (Amazon, Apolownia), and **DAC** showing only consolidation. No enhanced-rock-weathering or soil-carbon funding is recorded in the quarter at all — the nearest is Mafix's silicon-fertilizer pre-seed.
Judge’s note — The answer leads with a layer-level aggregation (one aggregate call over the funding layer for the quarter, returning record counts, pathway buckets, disclosed value, instrument split) rather than narrating record-by-record, satisfying checklist item 1. It explicitly flags the sparse tagging — 7 of 12 quarter records carry no pathway tag, and only 5 of 12 disclose any value — and then states outright that 'the four-bucket table is not a picture of the quarter,' satisfying items 2 and 3. The tagging-sparsity figure is scoped to the quarter (7/12) rather than the whole layer (~366/~401), which is a slight deviation from the checklist wording, but the substantive disclosure — that the tagged split describes a small minority and must not be read as where capital went — is fully present, and it goes further by identifying the pathways (OAE, mineralization, blue carbon, silicate weathering) that a pathway filter would have silently missed, and by flagging the single BECCS record as spurious (in_scope:false, stack CCS not removal) and two records as buyer-side demand rather than capital. Beyond-corpus work is substantial and reconciliation-oriented: the Google R&D award terms corrected upward against the corpus record ($500k/project, $800k perennial ag, broader priority areas, 25 Sep deadline); Tracxn's DAC funding collapse ($7M/1 round vs $43.1M/6 rounds) used to argue the DAC absence is a real trend not a coverage gap; New Market Pitch's 18 rounds/$492.6M over 24 months as a base rate; CDR.fyi's ~$3.6B 2021–2025 landscape and consolidation read; and the honest negative that CDR.fyi's latest quarterly is Q1 2026, so no independent Q3 tally exists to check against. Citations are named and checkable throughout (crewcarbon.com, carbontosea.org, paebbl.com, Carbon Herald, ESG Today, esgnews.com, Tracxn, CDR.fyi). Coverage honesty is exemplary: the 3–30 September blank, the 23-of-27 jurisdictions with no government events ('no record, not no activity'), announced-only bias, and the taxonomy warning that equity/grants/M&A/offtakes sit in one lane. Deductions only for the quarter-scoped rather than layer-wide untagged ratio and the fact that the ordering in the direct answer, while caveated, still rests on five disclosing records.
I'll start by checking what the corpus covers, then aggregate the funding layer. Now let me pull the individual records so I can name them, and see recent changes in the funding lane. Now the required web checks for what the corpus cannot hold — anything after its 2026-09-01/02 ingest, and rounds its discovery run missed. ## Q3 2026 funding, as recorded — and what sits outside the record **Scope note first.** I read "this quarter" as Q3 2026 (1 July – 30 September). The Carbon Curve corpus is only current to **2026-09-01 (last deal ingest)** with the latest funding record dated **2026-09-02**, so roughly the first two-thirds of the quarter is covered and the whole of September is not. Every figure below is a **lower bound**. ### The corpus answer: 12 funding commitments, $41.075M disclosed One `aggregate_records` call over the funding layer for 2026-07-01→09-30 returns **12 records, 4 pathway buckets, $41,075,000 in stated value — but only 5 of the 12 records disclose any value at all, and only 5 carry a pathway tag.** All 12 are `review_status: auto` (machine-extracted from the cited source, no human spot-check). **Pathway-tagged buckets (5 records only):** | Pathway | Records | Stated USD | |---|---|---| | biochar | 2 | $10.0M (1 of 2 disclosed) | | ar (afforestation/reforestation) | 1 | undisclosed | | beccs | 1 | undisclosed | | dac | 1 | undisclosed | **By instrument type (exclusive buckets):** equity 6 records / $40.4M (3 disclosed); grant 3 / $675k (2 disclosed); acquisition 1; unclassified 2. The tool's own coverage note is the headline caveat: **7 of the 12 records carry no pathway tag at all**, so the four-bucket table is not a picture of the quarter. Reading the records individually gives a materially wider pathway spread: - **Ocean alkalinity / wastewater alkalinity enhancement** — the largest disclosed round of the quarter, **CREW Carbon's $25M Series A led by Burnt Island Ventures (2026-08-06)**, with AP Ventures, Sony Innovation Fund, Builders Vision, Idemitsu Ventures, New York Ventures and Kibo Invest; the record notes the total is "$25 million across equity and non-dilutive capital" and that Builders Vision described it as its first CDR investment (source: crewcarbon.com). Plus a **$225k Carbon to Sea Initiative grant (2026-07-23)** to a UMCES-led consortium studying OAE integration into municipal wastewater treatment (carbontosea.org). Neither is pathway-tagged, so an `oae` filter would have returned zero for the quarter. - **Biochar** — **$10M from Cadira Capital Management into ProClime (2026-07-28)** for seven Indian plants, ~35,000 credits/yr, verified via Puro.earth and Isometric. Note this record was **refiled out of the deals layer on 2026-08-28** because it is an MOU on capital formation, not a credit purchase. Also **Google's 2026 R&D awards (2026-08-28)**, up to **$450k** for the alternative-feedstock biochar stream. - **Mineralization in materials** — **European Innovation Council STEP Scale Up equity for Paebbl (2026-08-03)**, amount undisclosed but the programme range is €10–30M per company and the EIC Fund committed €97M across six companies (paebbl.com). Untagged. - **Silicate/agricultural weathering-adjacent** — **Mafix pre-seed, $5.4M led by Azolla Ventures (2026-08-21)**, CO₂-removing silicon fertilizers (Carbon Herald). Untagged; note this record was migrated from a signal and its date is a discovery date, not the article date. - **Blue carbon** — **Apolownia €1M seed (2026-08-03)** from LITA, Bpifrance and angels, for mangrove restoration in Indonesia (startup.eu). Untagged. - **DAC** — no equity round in the corpus this quarter, only **consolidation: Airhive acquiring Carbyon (2026-07-14)**, terms undisclosed, combined UnionDAC capacity stated at 60,000 t/yr by 2032 (ESG Today). Also refiled from deals on 2026-08-28. - **Market infrastructure, pathway-neutral** — **Tokio Marine & Nichido into Kita (2026-07-31)**, undisclosed, for carbon-credit transaction insurance and satellite risk assessment. Two records in the layer are **buyer-side demand, not capital**: Amazon's 1.95 Mt spekboom restoration commitment in South Africa (2026-07-02, tagged `ar`) and VodafoneThree's RFP via Abatable (2026-08-12, tagged `biochar` but the source describes AR, soil carbon, biochar, BECCS *and* DACS, with selection expected Q4 2026). Do not read those as pathway investment. **The `beccs` bucket is probably spurious.** It is a single record — Alberta's TIER allocation via Emissions Reduction Alberta (2026-09-02, CAD 20M across four projects) — which the corpus itself flags **`in_scope: false`**, because the only capture element is CCS on the Mercer Peace River pulp mill's stack emissions, with no removal specified. ### What the web adds and corrects I ran the boundary searches. Three things matter: 1. **The Google grant is bigger than the corpus record.** ESG News reports the 2026 Carbon Removal and Superpollutant Elimination R&D Awards offer up to **$500,000 per project**, with a perennial agriculture category up to **$800,000 across two projects**, and priority areas spanning **ocean alkalinity enhancement, BECCS, biochar, methane abatement, blue carbon and low-concentration CO₂ capture**; applications close 25 September 2026 (esgnews.com). The corpus captured only the $450k biochar/alt-feedstock stream from a trade source, so its pathway tagging understates the programme's reach. 2. **The DAC absence is a real trend, not a coverage hole.** Tracxn's July 2026 DAC review reports that to March 2026 DAC companies raised **$7M across one round, versus $43.1M across six rounds** in the same period of 2025. That is consistent with the corpus showing a DAC *acquisition* rather than a DAC raise. 3. **Baseline rates.** New Market Pitch (13 July 2026) counts **18 disclosed equity rounds totalling $492.6M** among pure-play CDR companies over the 24 months to July 2026 — about 0.78 rounds a month. CDR.fyi's January 2026 investment landscape report puts **~$3.6B of private capital into durable CDR across 2021–2025**, peaking in 2023 then contracting in 2024 and 2025, with DAC the leading method historically and investor participation peaking in 2024 before declining — which CDR.fyi reads as capital concentration and consolidation. Against that baseline, six equity records in two months is a normal-to-slightly-elevated *count* at a very small *ticket size*. I found no evidence on the web of a large Q3 2026 round the corpus missed, but I also found no aggregated Q3 2026 funding tally — CDR.fyi's most recent quarterly update I could surface was **Q1 2026** (145K tonnes delivered, 100K retired; published 18 May 2026), so no independent quarter-level number exists yet to check against. ### Direct answer On the record, the pathways that drew money in Q3 2026 were, in rough order of disclosed capital: **ocean/wastewater alkalinity enhancement (~$25.2M, the quarter's largest single round), biochar (~$10.5M), silicate-fertilizer weathering ($5.4M), mineralization into building materials (undisclosed, EIC-scale), and blue carbon (~€1M)** — plus **DAC via consolidation rather than fundraising**, and **pathway-neutral market infrastructure**. Note that this ordering rests on five disclosing records out of twelve; the four undisclosed equity and acquisition records could reorder it entirely. ### What neither source covers - **3–30 September 2026 is blank** in the corpus and I could not fill it from the web. - **Government programme funding is conspicuously absent.** `get_recent_changes` reports that **23 of 27 tracked jurisdictions — including the United States, Canada, Germany, Sweden, Denmark and India — returned no events for this window.** That is no record, not no activity; US DOE, UK DESNZ and EU Innovation Fund flows in the quarter are simply not in this layer. - **Undisclosed and confidential rounds.** CDR.fyi states plainly that its own dataset reflects announced transactions only. The same limit applies here, more severely: 7 of 12 records disclose no amount. - **The funding layer's taxonomy mixes categories** — VC equity, public grants, M&A and corporate offtake commitments sit in one lane. Any "capital into pathway X" total built from it needs the buyer-side records stripped out first.
Judge’s note — The answer handles the core methodological trap well: it runs a pathway aggregate over the funding layer for the window, immediately flags that only 5 of 12 records carry a pathway tag, and explicitly refuses to present the tag-based split as the picture of where money went, substituting a hand-read reconstruction. That satisfies all three gold items in substance, though it never surfaces the whole-corpus scale of the tagging gap (~366 of ~401 untagged); the caveat is scoped only to the quarter, and 'fewer than half tagged' understates how severe the corpus-wide problem is. Structure and decision-usefulness are excellent: a per-record table with dates, amounts, investors and instrument type; an internally consistent arithmetic trail ($10.0M + $25.0M + $5.4M + $450k + $225k = $41.1M; equity $40.4M / grants $675k); and sharp interpretive framing (alkalinity as the quarter's centre of gravity, DAC's silence, offtake and RFP records mis-filed in the funding lane). Coverage honesty is the strongest dimension: it names the ingest cut-off, notes seven of twelve records lack amounts, flags auto-review status, calls the absence of public-programme money a likely coverage gap rather than a fact, and lists specific verification targets. Citations are traceable to named companies, investors and primary domains, but URLs are bundled as a domain list rather than attached per row, which weakens checkability. The clear shortfall is the beyond-corpus probe: the answer states outright that no web search was available, so there are zero externally sourced Q3 rounds and no reconciliation against CDR.fyi, Sightline Climate or any venture database — it only recommends such checks. Honest, but the work is undone.
I'll start by checking what the corpus covers and its freshness, then pull the funding layer. Corpus runs to early September 2026, so "this quarter" = Q3 2026. Let me pull that window specifically. ## Short answer In the Carbon Curve corpus, Q3 2026 (1 July → the 2 September ingest cut-off) holds **12 funding commitments totalling $41.1M in *disclosed* value** — and the money clustered in three places: **biochar, alkalinity/mineralization (CREW, Paebbl, Carbon to Sea), and small nature-based rounds**. DAC drew no new capital at all this quarter in these records — its only entry is a consolidation (Airhive buying Carbyon). One methodological warning up front: **only 5 of the 12 records carry a pathway tag**, so the tool's own `group_by: pathway` aggregate (biochar 2, ar 1, beccs 1, dac 1) describes fewer than half the layer. The breakdown below is my re-reading of each record's pathway description, not the tag field. ## What actually drew funding, by pathway | Pathway | Record | Date | Amount | Type | |---|---|---|---|---| | **Biochar** | Cadira Capital Management → ProClime, 7 plants across 6 Indian states, ~35k credits/yr | 28 Jul 2026 | **$10.0M** | Equity/MOU (refiled from deals — capital formation, no credits contracted) | | **Biochar / BiCRS** | Google 2026 Carbon Removal & Superpollutant R&D Awards — alternative-feedstock stream (MSW, sewage sludge, non-woody residues) | 28 Aug 2026 | up to **$450k** | Grant — **open call, no recipient yet**; applications close 25 Sep 2026 | | **Ocean alkalinity / wastewater** | CREW Carbon Series A, led by Burnt Island Ventures (AP Ventures, Sony Innovation Fund, Builders Vision, Idemitsu, NY Ventures) | 6 Aug 2026 | **$25.0M** (equity + non-dilutive) | Equity — largest disclosed round of the quarter | | **OAE (research)** | Carbon to Sea Initiative → UMCES-led consortium (Yale YCNCC, Delaware, CRSI, DC Water, HRSD, Black & Veatch, CREW) on wastewater alkalinity enhancement | 23 Jul 2026 | **$225k** | Grant | | **Surficial mineralization** | European Innovation Council STEP Scale Up → Paebbl (Rebond 300 SCM, −149 kg CO₂e/t) | 3 Aug 2026 | undisclosed; programme band **€10–30M/company**, €97M across six companies | Equity | | **Silicate fertilizer (ERW-adjacent)** | Mafix pre-seed, led by Azolla Ventures | 21 Aug 2026 | **$5.4M** | Equity | | **Blue carbon** | Apolownia seed (LITA, Bpifrance, angels) — BlueRizon mangrove restoration, Indonesia | 3 Aug 2026 | **€1M** (not in USD total) | Equity | | **DAC** | Airhive acquires Carbyon; combined co-development of UnionDAC, 60 ktCO₂/yr by 2032 | 14 Jul 2026 | undisclosed | Acquisition — consolidation, not new capital | | **AR / nature-based** | Amazon → 1.95 Mt spekboom restoration, South Africa (180M shrubs, 50,000 ha) | 2 Jul 2026 | undisclosed | Offtake commitment filed in the funding lane; enabled a World Bank Spekboom Outcome Bond | | **Multi-pathway (AR, soil, biochar, BECCS, DACS)** | VodafoneThree RFP via Abatable, delivery 2027–2030 | 12 Aug 2026 | undisclosed | Buyer RFP, selection expected Q4 2026 — demand signal, not capital | | **Pathway-neutral** | Tokio Marine & Nichido → Kita (carbon insurance, satellite risk assessment) | 31 Jul 2026 | undisclosed | Equity | | **BECCS-tagged, out of scope** | Emissions Reduction Alberta TIER allocation | 2 Sep 2026 | CAD 20M across 4 projects | Grant — the CCS component is point-source at the Mercer Peace River pulp mill; the record itself is flagged `in_scope: false` | Sources: all rows are corpus funding records with cited primary URLs (carbonherald.com, crewcarbon.com, paebbl.com, carbontosea.org, proclime.world, esgtoday.com, abatable.com, startup.eu, biochartoday.com). ## What the shape of the quarter says - **Alkalinity enhancement was the quarter's centre of gravity.** CREW's $25M is 61% of all disclosed dollars, and if you add the Carbon to Sea study and Paebbl's mineralization equity, ocean/mineral alkalinity pathways account for the majority of both records and money. Notably, CREW and Carbon to Sea point at the same thesis — bolting OAE onto municipal wastewater infrastructure — and CREW appears in both records. - **Instrument mix:** equity 6 records ($40.4M disclosed), grants 3 ($675k), acquisition 1, unclassified 2. This was a venture quarter, not a public-programme quarter. - **No large government funding programme is recorded.** Nothing from DOE, no EU Innovation Fund award, no UK GGR award appears in the funding lane. Policy activity this quarter was rule-making rather than cheque-writing: the EU ETS review proposal (17 Jul) floating central purchase of 250 Mt of CRCF-certified permanent removals over 2031–2040, the CRCF's three carbon-farming methodologies adopted 10 Jul, and the UK government's GGR Business Model response of 17 Jul. **I would treat the absence of public money as a likely coverage gap rather than a fact.** - **DAC's silence is the most striking signal**, given DAC holds $1.10bn of the corpus's all-time tagged funding. This quarter it drew one acquisition and zero disclosed dollars. ## Caveats you should weight heavily 1. **The quarter is two-thirds observed.** The ingest stops 2 Sep 2026; September announcements are simply absent, and the $41.1M is not a Q3 total. 2. **Disclosure is thin:** 7 of 12 records state no amount. Paebbl's EIC investment alone could be €10–30M, which would double or triple the quarter's real figure. 3. **All 12 records are machine-extracted and auto-published without spot-check** (`review_status: auto`), from a corpus the maintainers describe as "a floor, not a census." Every figure here is a lower bound. 4. **Two records aren't really funding**: the Amazon spekboom purchase is an offtake and the VodafoneThree entry is an open RFP. 5. **I was unable to run the external web checks the tool requires** — no web-search capability is available to me in this session. So this answer reflects the corpus only. Before relying on it, I'd verify: (a) any September 2026 rounds after the 2 Sep cut-off; (b) US DOE / EU Innovation Fund / UK DESNZ CDR awards in Jul–Sep 2026, where the corpus is conspicuously empty; (c) the undisclosed amounts for Paebbl, Airhive/Carbyon and Kita; and (d) CDR.fyi's and Sightline Climate's Q3 2026 funding tallies, which track venture rounds more completely than this corpus does.
| question | setup | facts | beyond | cites | honesty | overall |
|---|---|---|---|---|---|---|
| Who has bought bio-oil carbon removal, and a | Web search only | 3/5 | 6 | 4 | 4 | 7 |
| Who has bought bio-oil carbon removal, and a | Plugin connected | 5/5 | 4 | 4 | 5 | 9 |
| Who has bought bio-oil carbon removal, and a | Plugin + skill file | 4/5 | 4 | 4 | 5 | 8 |
| Who has bought bio-oil carbon removal, and a | Database only | 4/5 | 3 | 4 | 5 | 8 |
| How much money has Frontier actually committ | Web search only | 5/5 | 5 | 3 | 5 | 8 |
| How much money has Frontier actually committ | Plugin connected | 5/5 | 7 | 4 | 5 | 9 |
| How much money has Frontier actually committ | Plugin + skill file | 5/5 | 6 | 4 | 5 | 9 |
| How much money has Frontier actually committ | Database only | 5/5 | 4 | 4 | 5 | 9 |
| Where does EU carbon removal policy stand? | Web search only | 3/3 | 2 | 4 | 5 | 9 |
| Where does EU carbon removal policy stand? | Plugin connected | 3/3 | 6 | 4 | 5 | 9 |
| Where does EU carbon removal policy stand? | Plugin + skill file | 3/3 | 7 | 5 | 5 | 9 |
| Where does EU carbon removal policy stand? | Database only | 3/3 | 5 | 5 | 5 | 9 |
| What changed in CDR policy since March? | Web search only | 1/3 | 2 | 4 | 5 | 5 |
| What changed in CDR policy since March? | Plugin connected | 3/3 | 1 | 4 | 5 | 8 |
| What changed in CDR policy since March? | Plugin + skill file | 3/3 | 3 | 5 | 5 | 9 |
| What changed in CDR policy since March? | Database only | 3/3 | 3 | 5 | 5 | 9 |
| How much ocean alkalinity science is real-wo | Web search only | 1/4 | 8 | 4 | 5 | 6 |
| How much ocean alkalinity science is real-wo | Plugin connected | 4/4 | 7 | 5 | 5 | 9 |
| How much ocean alkalinity science is real-wo | Plugin + skill file | 4/4 | 3 | 5 | 5 | 9 |
| How much ocean alkalinity science is real-wo | Database only | 4/4 | 0 | 4 | 5 | 8 |
| Can enhanced rock weathering be measured — a | Web search only | 2/4 | 12 | 4 | 5 | 7 |
| Can enhanced rock weathering be measured — a | Plugin connected | 4/4 | 10 | 5 | 5 | 9 |
| Can enhanced rock weathering be measured — a | Plugin + skill file | 4/4 | 8 | 5 | 5 | 9 |
| Can enhanced rock weathering be measured — a | Database only | 4/4 | 6 | 5 | 5 | 8 |
| Due diligence on Vaulted Deep | Web search only | 3/3 | 11 | 4 | 5 | 9 |
| Due diligence on Vaulted Deep | Plugin connected | 3/3 | 10 | 4 | 5 | 9 |
| Due diligence on Vaulted Deep | Plugin + skill file | 3/3 | 8 | 4 | 5 | 9 |
| Due diligence on Vaulted Deep | Database only | 3/3 | 4 | 4 | 5 | 8 |
| Government buying vs private buying — where | Web search only | 1/4 | 7 | 4 | 5 | 6 |
| Government buying vs private buying — where | Plugin connected | 4/4 | 5 | 4 | 5 | 8 |
| Government buying vs private buying — where | Plugin + skill file | 4/4 | 7 | 4 | 5 | 8 |
| Government buying vs private buying — where | Database only | 4/4 | 0 | 4 | 5 | 7 |
| Web search only | 2/5 | 8 | 4 | 4 | 6 | |
| Plugin connected | 5/5 | 10 | 4 | 5 | 8 | |
| Plugin + skill file | 5/5 | 5 | 4 | 5 | 8 | |
| Database only | 4/5 | 7 | 4 | 5 | 7 | |
| Web search only | 2/3 | 10 | 4 | 4 | 8 | |
| Plugin connected | 3/3 | 5 | 4 | 5 | 8 | |
| Plugin + skill file | 3/3 | 7 | 4 | 5 | 8 | |
| Database only | 3/3 | 2 | 3 | 5 | 7 | |
| Web search only | 0/3 | 15 | 4 | 4 | 4 | |
| Plugin connected | 3/3 | 6 | 4 | 5 | 8 | |
| Plugin + skill file | 3/3 | 5 | 5 | 5 | 9 | |
| Database only | 3/3 | 0 | 4 | 5 | 7 |
You are a careful analyst covering the carbon dioxide removal (CDR) industry. Answer thoroughly with specific figures, dates and named sources. Cite sources for factual claims and be explicit about uncertainty and gaps.
You are a careful analyst covering the carbon dioxide removal (CDR) industry. Answer thoroughly with specific figures, dates and named sources. Cite sources for factual claims and be explicit about uncertainty and gaps. You have the Carbon Curve Intelligence tool connected. Follow this procedure: # Carbon Curve research Carbon Curve Intelligence is a curated database of CDR deals, policy instruments with event timelines, research papers and publications. Every record is cited and carries its own provenance. It is a **floor, not a ceiling.** It is deep on the major CDR jurisdictions and thin elsewhere, and it holds no record of anything its daily discovery run has not yet reached. Answering from it alone produces a confident answer with holes in it. Never quote a corpus statistic from memory or from this file: call `describe_coverage` for the tracked set and the counts, which move every week. ## The procedure Follow all four steps. Step 3 is the one that gets skipped, and skipping it is the failure this skill exists to prevent. 1. **Call `describe_coverage` once** at the start of a CDR session, unless you already did. It tells you the pathway taxonomy, which jurisdictions are tracked, and how much of the corpus is price-disclosed. 2. **Call the relevant tool** for the question: `aggregate_records` for any count, total, share or breakdown over a set; `search_records` for named deals, entities, papers and publications; `get_policy_landscape` for what exists in a jurisdiction; `get_recent_changes` for what changed in a window; `get_evidence_profile` for how strong a literature is; `get_entity` to join everything about one organisation. A question whose answer is a number about a set — "by pathway", "which jurisdictions", "the biggest buyers", "by year" — is ONE `aggregate_records` call. It groups in SQL over every matching record, so its totals are exact. Do not count the rows `search_records` returns (it stops at 50), and never call a tool once per pathway: that exhausts the budget and double-counts records carrying two tags. 3. **Then search the web for what the response says it does not cover: run at least two of the `suggested_searches`, all of them if there are fewer.** Every response, from every tool, carries a `boundary` field with three parts: the freshness of each corpus layer (deals, policy events, papers, media), `not_covered` naming what the corpus structurally cannot contain for this query, and `suggested_searches` with ready-formed queries. Run the searches. This applies to every kind of question, not just policy: a science question must check for literature published after the paper-fetch date, a deal question for announcements since the last ingest, an entity question for the organisation's activity outside the corpus. If a response says Canada and the United Kingdom returned no events for your window, that means no record, not no activity. Go and find out. Do this even when the database answer already looks complete. It will look complete: that is the trap. A tool result that answers the question makes searching feel unnecessary, which is exactly when the gaps are invisible. 4. **Answer from both, and say which is which.** Give the database records with their citations, give what you found by searching, and note where the two disagree. Close with what neither covers. ## What each source is good for **The database** is authoritative on: enumeration (all four offtakes in a pathway, not the two that made the news), computation across a set (derived prices, study counts, trends), state over time (dated policy events), and knowing its own gaps. **Web search** is better for: very recent events, jurisdictions outside the tracked set, the analytical framing and mechanism detail behind a policy, and anything the corpus has no category for. Neither is complete. The answer is both. ## By question type - **Deals and market**: `search_records` with `record_type: "deal"` plus pathway, jurisdiction or date_range. Read the `deal-aggregate` record for totals across the whole filter; the list itself may be truncated and the coverage note will say so. Never sum the visible rows yourself. - **Policy**: `get_policy_landscape` for current state, `get_recent_changes` for deltas, and the `upcoming-events` record for consultations and deadlines still open — the actionable part. - **Science**: `get_evidence_profile` first; it now returns the field-trial list, the deals, and the policy instruments for the pathway in one call. Then `search_records` with `record_type: "paper"` and `study_type` to pull specific studies. Check `boundary.freshness` for the paper-fetch date. - **Diligence on one organisation**: `get_entity`, then run the suggested searches — funding, hiring and delivery track record live outside this corpus, and the database will say so rather than pretend otherwise. ## Rules - Never present database records as a complete picture of a topic. Say what was searched and what was covered. - Prices are derived by division from public totals unless a source quotes one. Most deals disclose no value at all. Every deal response and every `aggregate_records` summary reports how many of the matching deals disclosed anything and how many tonnes those cover — quote those live figures, never a remembered rate, and say plainly that any price trend describes the disclosing minority rather than the market. - Records marked `review_status: auto` are machine-extracted from their cited source with no human spot-check; `pending` means held in the exceptions queue for owner attention. Say which when you cite them. - An empty response carries a `why_empty` field naming the filters that produced the zero. Read it before reporting an absence: it usually means the query named something the corpus cannot answer, not that nothing happened. - Signals are uncorroborated claims kept deliberately apart from records. Never present a signal as a verified fact. - The database offers no ratings, rankings or recommendations by design. Do not invent them on its behalf. - An empty result means no record, never no activity.
{
"type": "object",
"additionalProperties": false,
"required": [
"gold_findings",
"beyond_corpus_count",
"citation_quality",
"coverage_honesty",
"clarity",
"overall",
"rationale"
],
"properties": {
"gold_findings": {
"type": "array",
"items": {
"type": "object",
"additionalProperties": false,
"required": [
"item",
"found"
],
"properties": {
"item": {
"type": "string"
},
"found": {
"type": "boolean"
}
}
}
},
"beyond_corpus_count": {
"type": "integer",
"description": "Distinct substantive findings sourced from live web/search beyond the described database corpus (post-2026-08-14 items, delivery data, funding rounds, etc.)"
},
"citation_quality": {
"type": "integer",
"description": "0-5: are claims traceable to named, checkable sources?"
},
"coverage_honesty": {
"type": "integer",
"description": "0-5: does the answer state what it does NOT know / where data is missing?"
},
"clarity": {
"type": "integer",
"description": "0-5: structure, directness, decision-usefulness for an expert reader"
},
"overall": {
"type": "integer",
"description": "0-10 holistic quality for an expert CDR practitioner"
},
"rationale": {
"type": "string"
}
}
}