They are all right. They are simply not answering the same question.

This series has argued earlier that the value of certified product carbon intensity (CI) today is about access rather than premium — the number is becoming a condition of sale, not (yet) a surcharge on it. That leaves an obvious question hanging. Access to what, and demanded by whom?

Four drivers are pushing a CI number into purchase orders. Each requires a different standard of proof. And the mix is about to change in a way most of the market is not yet built for.

The four drivers, and their standards of proof

Before diving into the drivers, an explanatory note on where they sit. An earlier post in this series set out the six forcing functions turning measurement into a market. This post zooms into force no.5: buyer demand and procurement. The four drivers below are not a new taxonomy. They are four carbon-related drivers influencing a business’s purchase order.

See the full and live Standards Stack diagram and details in my Github repo here.

Driver 1 -Compliance. Regulatory rules make the CI number a condition of sale, and then a cost. In turn, this driver has two axes. The first axis is access: can I sell at all? Importers under the EU’s CBAM border tariff, importers of gas and crude under the EU Methane Regulation from 2027, shipowners under FuelEU and the maritime extension of the EU ETS, bidders into public procurement with embodied-carbon thresholds. And the shape of that access requirement has just become clearer: in July 2026 the Commission confirmed that the Methane Regulation does not require physical tracing of molecules, and that trace-and-claim and third-party certification are valid routes for an importer to demonstrate equivalence — subject to criteria against fraudulent or duplicate issuance. That is a regulator putting its name to a certificate-based compliance route. The second axis is cost: how much must I surrender? Here, the arithmetic weighs in. Because of how the EU’s border tariff is designed, the relationship between emissions intensity and certificate obligation is non-linear — recent analysis of hot-rolled steel coil suggests a reduction of roughly a tenth in intensity can cut the CBAM certificate requirement by around a third. Required standard of proof: whatever satisfies the rule, at acceptable cost and manageable risk.

Driver 2 — Disclosure. A reporting obligation makes the CI number a data problem. But it matters enormously which number is being reported.

At entity level the pressure is softer than it looks. European sustainability reporting does not require primary supplier data for a corporate Scope 3 inventory — only that you disclose what proportion rests on primary data and what on secondary averages. Spend-based averages remain entirely permissible, and the assurance ceiling is now permanently limited rather than reasonable. What pressure exists is comparative: your split, read against your peers’. Standard of proof: auditable and consistent.

But look at product level and the same word is far stricter. A declared product carbon footprint — an EPD under a product category rule, a battery carbon footprint declaration, a digital product passport entry — cannot be assembled from spend-based averages. It needs facility-specific foreground data, and it attaches to a physical good rather than a legal entity. The EU battery rules show where this leads: declaration first, then a performance class, then a maximum threshold that keeps the product off the market altogether. That is a disclosure obligation converting itself into a compliance one. Standard of proof: conformance to the applicable product rule, verified.

The consequence matters later, and it cuts both ways. Entity-level disclosure buyers are structurally slower and less demanding than they are usually assumed to be, because the rules let them use averages. Product-level disclosure buyers are not — but there are far fewer of them today, and the regimes that will create more of them arrive in the same window as the compliance shift.

Driver 3 - Commitment. A voluntary target makes the number a self-imposed condition — 3^rd party validated net-zero targets or transition plans, buyer coalitions, corporate pledges. This is the driver most often described in purely reputational terms, and that description is incomplete.

A commitment is also leverage. It provides a buyer a defensible rationale for a commercial ask. Demand certified CI from your suppliers and you can take the value in contractual terms rather than in price: priority allocation, longer duration, exclusivity, tighter delivery windows, better product specs, etc. That is how commitment-driven buyers routinely obtain certification without paying a premium for it — and it is the mirror image of the seller who offers a certified number in exchange for contract term. Standard of proof: the most defensible claim available, because reputational downside dominates cost.

Driver 4 - Cost of capital and insurance. Here the CI number moves the price of money and risk. Sustainability-linked loans and bonds tie margin to verified performance, typically in the range of five to twenty-five basis points either way. Trade-finance providers have begun offering concessions on certified cargoes. Insurers have started setting conditions on underwriting. Standard of proof: whatever a lender or an underwriter will actually credit.

The point of this little taxonomy is this: the same certificate is worth different things to different buyers because the driver sets the standard of proof. That is why the question has no single answer, and why two well-informed people can give opposite ones.

The buyer types, briefly

Most buyers are being influenced by more than one driver. Indeed, the driver combinations explain more than the buyer industry categories. Hyperscalers sit on commitment and disclosure, with unusual reputational exposure. Automakers add competitive positioning as their fleets pivot. Shipowners are close to pure compliance, on both axes. Importers exposed to the border tariff are compliance plus cost of capital where the exposure is financed, and they are the fastest-growing group. Producers financing transition capital expenditure are the odd case: they buy verification of their own number, to move a spread rather than to satisfy a customer.

The useful pattern is what happens when two or more drivers stack. The standard of proof is set by the highest driver, not the average. A buyer that is both compliance- and commitment-driven will demand what its commitment requires, not what the rule permits. Which is why the most demanding buyers in this market today are not the most heavily regulated ones — and why that picture is about to invert.

Why some buyers get to write the specification

Not every buyer can set a standard. The ones that do are those whose commitment serves to underwrite capacity that would not otherwise be built. Think of it like an anchor-tenant: a twenty-year offtake agreement that makes a supply project financeable is also buying the right to dictate terms, and measurement terms will be part of those.

This is how Europe’s natural gas import infrastructure was built. The arrangement between the Soviet gas ministry and West Germany’s Ruhrgas in the 1970’s — twenty-year delivery contracts, large-diameter pipe, and a Western bank credit line — financed the pipelines before a molecule shipped. Long-term contractual commitment was the instrument that made the capital work. I have heard commodity traders point out that there is no obvious reason the same mechanic cannot underwrite fuel decarbonisation, and there is some evidence that it is indeed starting to do so. That is a better explanation for procurement specification-setting than “these buyers are large and move quickly,” and it makes a prediction: the buyers who set the next specifications will be whoever is underwriting new supply.

Which is why energy supply to data centres is the clearest current case. Data centres consumed roughly 415 terawatt hours in 2024, about 1.5% of global electricity; the International Energy Agency’s base case has that reaching around 945 terawatt hours by 2030, just under 3%, with sensitivity cases either side depending on efficiency, uptake and grid bottlenecks. Growth of roughly 15% a year, around four times faster than electricity demand from everything else — projections contested in both directions, and which the argument here does not depend on. That growth is being underwritten with long-dated commitments — nuclear and clean-firm agreements at scale — and increasingly by buying the developers outright, as Alphabet did with Intersect Power in December.

The exception that explains the rule

The last post argued that certified CI earns little or no premium. But hyperscalers look like a counter-example. They are better understood as the exception that explains the mechanism.

The test is not how big the input is. It is how big the premium on the input is, measured against the buyer’s own economics — and whether that buyer has something reputational at stake.

Take the two sides of the same commodity. Electricity is around 40% of the cost of producing primary aluminium, and smelter economics depend on long-term power contracts at low fixed prices. A premium on power there is not a marketing decision; it is a threat to competitiveness. But the automaker buying that aluminium is in a different position entirely: the metal is a modest share of the cost of a vehicle, and the brand stake is considerable. Which is why the premium shows up at the automaker’s end of the chain and not at the smelter’s.

The data-centre case works the same way, and it is worth being accurate about it, because “power is trivial for hyperscalers” is a claim people make and it is not true. Power is a major cost. On one detailed estimate for a one-gigawatt AI facility, servers account for roughly 60% of annualised cost and energy for closer to 7% — and if you strip out depreciation, electricity becomes the largest cash operating line there is. These buyers negotiate hard on power.

What they are relatively insensitive to is the attribute premium sitting on top. A certified low-methane or hourly-matched claim adds a few per cent to an energy bill that is itself a minority of total cost — basis points of the whole. So the same organisation can be ruthless about the price of a megawatt hour and relaxed about the certificate attached to it. That is not inconsistency. It is the arithmetic.

So: commitment-driven buyers pay, or extract the equivalent in contract terms. Compliance-driven buyers do not.

What changes when compliance takes over

Here is the part I find uncomfortable, and I want to underscore that the composition here is inference, not observation.

Today, most new demand for certified carbon intensity comes from commitment-driven buyers. They want defensibility, and they set a high standard of proof because their exposure is reputational and reputational risk is asymmetric.

As the EU CBAM phases in, as methane import equivalence takes effect from 2027, and as fuel-intensity limits tighten, most new demand for certified CI will come from compliance-driven buyers instead. They will want sufficiency — the cheapest thing that satisfies the rules. On methane, note that the enforcement timetable is now softer than the obligation timetable: in July the Commission recommended that Member States not apply penalties to information failures due in 2027, 2028 and 2029. The duties still apply. The teeth arrive later.

The standard of proof falls as volume rises. Quality competition gives way to cost competition. The premium that exists at the top of this market may not survive contact with scale. And the certifiers and data providers built for defensibility will meet buyers who want a box ticked and an auditor satisfied, in that order.

Whether that becomes a race to the bottom depends almost entirely on whether regulators write their criteria tightly enough. Which is where the independence requirements I wrote about last time stop being a matter of principle and start being a matter of commercial structure.

There is one counterweight that does not depend on regulators, and it is the driver I set out fourth. Compliance-driven decarbonisation is capital-intensive — SAF, e-fuels, DRI-EAF steel, CCUS, low-methane gas infrastructure, HFO-to-LNG switching in shipping — and none of it gets built without lenders, insurers and long-dated offtake. A compliance buyer needs a number that satisfies this year’s rule. A lender is exposed for twenty years, over a period in which that rule is designed to tighten. It therefore cannot accept a number that is merely sufficient today. That is the stacking rule again, applied on the supply side rather than the demand side: a higher standard of proof than the buyer’s, set privately, and growing with precisely the volume that threatens to erode the other one.

So the more likely outcome is not a single race to the bottom but a split: a financed tier where proof is expensive, durable and forward-looking, and a compliance tier where it is cheap and annual.

The honest ceiling

Three things to concede.

One, the fuel tonnage in play here is small. Data centres are around 1.5% of global electricity today and under 3% on the central 2030 projection. Against global energy emissions this is not the main event. Two, these buyers are adding emissions, not only certifying them. Behind-the-meter gas turbines exist because the grid cannot deliver electrons fast enough. The buyer tendering for certified low-methane gas needs it precisely because it is burning gas. And three, a low-methane certificate rewards better operators; it does not, by itself, abate anything.

All three are true. But I think the argument survives them, because the specification effect will exceed the tonnage effect. A CI requirement written into an anchor buyer’s purchase order propagates to every supplier that wants the business, and from there into what the market starts to treat as the cost of doing business.

Where this goes

The buyers underwriting twenty-year supply are precisely the buyers who will, before long, stop asking for a CI flat number and start asking for a CI curve — a contracted, declining carbon intensity profile rather than a flat figure taken once and quoted thereafter.

Which raises the question the next post takes up: what kind of instrument is a CI number actually becoming?

For now, the practical version of all this is short. Before you ask what a certificate is worth, ask which driver is asking — and what standard of proof that driver will accept.

Which driver(s) are yours? I would be interested to know whether they have changed in the last year.

Subscribe now

Share

Sources & notes.

Demand and drivers: CBAM certificate obligation arithmetic — analysis of hot-rolled coil indicating a ~10% cut in emissions intensity can reduce certificate requirements by ~30% (Fastmarkets). EU Methane Regulation: on 20 July 2026 the Commission adopted two Recommendations rather than reopening the Regulation — one confirming that the Regulation does not require physical tracing of molecules and that trace-and-claim and third-party certification are valid routes to demonstrate importer equivalence, subject to criteria against fraudulent or multiple issuance; the other recommending that Member States not apply penalties to certain information failures due in 2027, 2028 and 2029. Recommendations have no binding force under Article 288 TFEU and the underlying obligations continue to apply (European Commission; Linklaters).

Disclosure: ESRS E1 requires disclosure of the proportion of primary versus secondary data rather than mandating primary data; limited assurance applies from the first reporting year, and the Omnibus I revisions removed the option to step up to reasonable assurance, with the Commission’s deadline for adopting limited-assurance standards moved to 1 July 2027.

Product-level disclosure: environmental product declarations under EN 15804 and the applicable product category rules require facility-specific foreground data; Regulation (EU) 2023/1542 on batteries provides for a carbon footprint declaration, then performance classes, then maximum lifecycle thresholds — a sequence whose dates are contingent on delegated and implementing acts that have repeatedly slipped and that defer the obligations if not adopted; the Ecodesign for Sustainable Products Regulation provides for digital product passports on a rolling product-group basis.

Cost of capital: sustainability-linked instruments typically apply margin adjustments of 5–25 basis points against verified performance targets. Cost structures: electricity is commonly put at around 40% of primary aluminium production cost (estimates range roughly 30–50%), with smelter economics resting on long-term contracts at low fixed prices; for a one-gigawatt AI data centre, one detailed annualised breakdown attributes roughly 60% of cost to servers and around 7% to energy, with electricity becoming the largest line once depreciation is excluded (Epoch AI). The inference drawn from these — that buyers are price-sensitive on power and comparatively insensitive on the attribute premium — is the author’s.

Anchor-buyer precedent: the 1970 “gas-for-pipe” arrangement — twenty-year Soviet–Ruhrgas delivery contracts, West German large-diameter pipe and Western bank credit — financed pipeline construction (Chatham House; Schattenberg, Journal of Contemporary History, 2022).

Data-centre demand: IEA Energy and AI — ~415 TWh in 2024 (~1.5% of global electricity); Base Case ~945 TWh by 2030 (just under 3%), with Lift-Off, High Efficiency and Headwinds sensitivity cases; ~15% annual growth, more than four times faster than all other electricity demand (IEA). These are projections and are labelled as such.

Vertical integration: Alphabet’s acquisition of Intersect Power (December 2025).

Buyer specifications: corporate RFPs for certified low-methane gas (2026); Buy Clean thresholds requiring facility-specific verified environmental product declarations.

The four-layer map is the live Standards Stack (monobermudez-svg.github.io/ghg-standards-stack). The shift from defensibility to sufficiency, the commitment-as-leverage observation, the anchor-buyer framing, and the argument that lender and insurer tenor sets a higher and privately enforced standard of proof than compliance demand are the author’s reading and are labelled as such in the text.

Then look at what is happening around them. Industrial software firms — Context Labs and Cognite among them — are building measurement systems that produce high-trust, batch-level carbon intensity for fuels or materials from real time production and operational data. Price-reporting agencies — e.g. S&P Global Platts and Argus — are folding carbon intensity into their assessments. Exchanges and registries — Xpansiv and registry operators such as Evident — keep emerging and rebuilding the plumbing for attribute trading. Certification schemes — OGMP 2.0, ISCC, CertifHy — are being written into regulation. None of these organisations are naïve and, while many would certainly like one, none of them are waiting for a carbon intensity premium — because none of them are selling the commodity.

Both things are true at once, and reconciling them is the point of this piece.

The thing is, the premium is the wrong thing to measure. Certified carbon intensity is becoming a condition of sale, not a surcharge on it — and the value it creates is captured almost entirely by the businesses around the number, not by the molecule or material itself. Premia are beginning to appear at the edges — European green steel and low-carbon aluminium both carry published differentials now — but they follow the compliance price rather than lead it and, for now, they remain the exception.

The premium is the wrong question

Here is the reframe. When a carbon-intensity specification becomes a prequalification requirement in a tender, the economics stop being about price and start being about access. If you cannot produce the CI certificate, your revenue on that contract is not lower — it is zero. If you can, you compete for the whole thing. The value of the certificate is the option to bid at all, which is to say the entire contract margin you would otherwise have forfeited.

Think of this: nobody pays a premium for meeting a sulphur specification, or a Wobbe index band for pipeline gas, or the LME Grade A purity standard for copper cathode. You simply cannot sell without it. Certified carbon intensity is on its way to becoming the same kind of thing — a licence to bid.

Once you see it through this lens and look for it, the evidence is not subtle.

Volvo Cars contracted near-zero-emissions steel from SSAB for serial production; Mercedes-Benz, Porsche and Scania signed offtakes with Stegra’s hydrogen-based plant. Those were won on specification. Meta has run tenders for certified low-methane gas to cover the upstream emissions of its data-centre power — a buyer with no direct regulatory obligation writing an intensity requirement into a purchase order. Public buyers are doing it by rule: Buy Clean California has required facility-specific, third-party-verified environmental declarations beating a maximum global-warming-potential limit since the start of 2025, and that limit ratchets down.

And the EU’s border tariff (CBAM) has quietly made intensity a matter of arithmetic. Because of how the mechanism is designed, the relationship between the emissions intensity of imported goods and the importer’s certificate obligation is non-linear: analysis of hot-rolled coil suggests a reduction of roughly a tenth in emissions intensity can cut the certificate requirement by around a third. That is not a premium. It is avoided cost, it is calculable, and it accrues to whoever measured well. Steelmakers outside Europe are responding accordingly — Hyundai Steel has pointed to rising demand for carbon-reduced plate from European automakers as the tariff takes full effect, with exporters prioritising lower-carbon routes precisely to preserve market access. Across the globe, exporters to Europe of EU CBAM-covered commodities have been scrambling to get ready for this.

So demand is where the value is created. It shows up as revenue protected rather than revenue gained, which is why it is nearly invisible in any analysis hunting for a green premium.

Where the value is captured

See the full Standards Stack diagram and details in my Github repo here.

Value created is not the same question as value captured, and conflating them is a mistake I see frequently. The producer captures market access. Almost everything else is captured by five pools sitting around the intensity number.

Three of them can be described quickly.

Methodology is a commodity. The frameworks themselves — the ISO standards, the GHG Protocol, the sector protocols — are free or nearly free, and the merger of ISO 14067 with the GHG Protocol’s product standard will push further in that direction. Owning the method confers enormous influence but little revenue — the EU’s own MRV framework is built substantially on OGMP 2.0, a protocol the industry developed and gave away. It is prestige, not profit, and that is broadly as it should be.

Certification and assurance is a trust toll — and it has just been handed a defensible moat. When the European Commission published its methane-regulation recommendations in July 2026, most commentary (including my own) saw only softened penalties. The more consequential detail sits underneath: certificate-based compliance is recognised for complex, commingled supply systems, and compliance providers are expected to be structurally independent of the industry they certify. Read against current operative policy rather than against the proposal and (as kindly pointed out to me by MiQ’s CEO Georges Tijbosch) that is a tightening — and it is regulation deciding who is allowed to collect this toll. Independence, in other words, has been converted from a virtue into a barrier to entry. Note what the Commission actually did: it published criteria, not names. Fourteen of them, covering certificate content, registry integrity, third-party audit and — the decisive one — that a provider be legally and functionally independent of any energy producer, supplier or importer, with separate accounts and no financial interest in the outcome. That single criterion disqualifies producer-affiliated certification in one go. Schemes already built to that specification are now publishing their own mappings against the criteria — MiQ has published its own assessment of compliance — while those that were not are likely reconsidering theirs. The timing sharpens it. EUMR reporting obligations began in 2025, verification requirements for importers arrive in 2027, and intensity limits follow in 2030 — but the binding constraint is not stringency, it is uncertainty. Firms can adapt to rules; what they cannot do is sign a fifteen-year supply contract while the interpretation of those rules is still moving. Certification is how that ambiguity gets converted into something a counterparty can underwrite.

Market infrastructure is a toll road, still under construction. I briefly covered the state of it in the last post — a daily methane-certificate benchmark launched and then withdrawn for thin liquidity, alongside block trades in the millions and a recent clearing arrangement between MiQ and CBL/Xpansiv. Registries and exchanges are where a mature attribute market would collect its rent. That market is not mature yet. The contrast with metals is instructive: price reporting agencies now publish standing low-carbon differentials for aluminium and green steel — Northern European green flat-rolled was assessed at a premium of roughly €100–170 a tonne early this year — because a compliance driver sits underneath them. Gas certificates have no equivalent, and no equivalent price.

The two remaining pools deserve longer.

Data and intelligence

This is the largest business in the stack, and the most defensible — but not for the reason usually given.

The obvious version of the argument is that somebody has to produce the number, and that producing it well requires instrumentation, sampling, engineering models and continuous data rather than some arithmetic of factors pulled from a database. True enough, and the gap between the two is wide: average emissions factors are, by definition, not measurements.

The stronger argument is that CI data has a second life. The same instrumentation that produces a measured and defensible carbon intensity figure — sensor data off compressors, flow and flare measurement, equipment-level process digital twins — is also an asset-optimisation dataset. It provides new lenses into production efficiency, predictive maintenance, prescriptive optimisation, and overall cheaper delivery of decarbonisation targets. I have watched operating teams request access to emissions data built for certification because it turned out to enhance the picture of how their plant actually runs.

That changes the investment case entirely. A business whose value depends on carbon policy holding its nerve is a fragile business. A business whose output improves plant performance and happens to produce a certifiable carbon number is not. If you are looking for where to build in this market, the test I would apply is whether the product would survive a decade of weak carbon prices. This pool passes it.

Cost of capital

The quiet one, and the pool most consistently underestimated.

Carbon performance is already priced into the cost of debt. Sustainability-linked loans and bonds tie margin to verified performance against agreed targets, with emissions the most commonly used indicator; the ratchets typically run in the range of 5–25 basis points either way. Trade-finance providers have begun offering concessions on cargoes backed by verified intensity certification.

Basis points sound trivial next to the promise of a green premium. They are not. A handful of basis points applied to billions of outstanding debt is a larger and far more reliable number than a per-tonne premium that, outside a few metals markets, mostly does not exist. It arrives annually, it compounds, and it does not depend on any counterparty agreeing to pay more for the same molecule. For a capital-intensive producer, this is currently the most bankable financial return on measuring well — which is a strange sentence to write about carbon, and a useful one to sit with.

The caution is real, though: these instruments have had a difficult few years, margin ratchets disappeared from parts of the market after some deserved criticism about weak targets, and the discipline now expected of the underlying metrics is considerably higher. That is a reason to expect the pool to reward credible measurement, not a reason to dismiss it.

The same logic is appearing on the risk-transfer side. Insurers have begun setting methane-management conditions for oil and gas underwriting — Chubb was the first major carrier to require leak detection and repair programmes of its upstream clients — and sustainability-linked policies tie premiums to verified performance, as in AXA’s programme with Enel. Note the direction of travel: the condition is about access to cover, not a discount on it. The same pattern as the tender.

Book-and-claim is a bridge, and electricity markets are the dry run

Most of what trades today is an unbundled attribute — a certificate separated from the physical molecule, sold on its own. It carries all the objections that have been levelled at renewable energy certificates (RECs), and it deserves some of them. Anyone who recalls the 2021 vogue for “carbon-neutral” LNG cargoes or “carbon-compensated” crude, where a hydrocarbon shipment acquired a green label through offset credits, should be wary of a market that separates claims from physical reality too casually. Regulators are now drawing that line themselves. The Commission’s July recommendations confirm that the methane regulation does not require physical tracing of molecules, deliveries or cargoes — meaning certificate-based compliance is acceptable — but only where certificates match the country of origin of gas actually imported into the EU. Global book-and-claim is out; origin-matched book-and-claim is in. That is a regulator specifying which version of the bridge will be allowed to bear weight.

But the useful thing about that comparison is that the power market has already run this experiment, and it is roughly a decade ahead of molecules.

Annual RECs attracted exactly this critique, and the response was not abandonment but a move towards granularity. EnergyTag now maintains a standard for hourly, time-stamped granular certificates, the first issuers have been accredited against it, and hourly matching underpins the 24/7 carbon-free energy goals that large electricity buyers like Google have set. The architecture did not change — it is still an unbundled attribute — but the claim became specific enough in time and place to overcome the most salient shortcomings of annual RECs.

That is the trajectory I would expect for molecules and materials: from annual averages, to supplier-level figures, to per-cargo and per-delivery certified intensity that a buyer can underwrite and an auditor can sign. The destination is not a certificate market. It is financial-grade carbon intensity — measured, recorded, verified independently, and carried with the product.

The convergence is already visible in a single buyer. A hyperscaler like Meta matching data-centre consumption hour by hour to carbon-free generation, while separately tendering for certified low-methane gas to cover the molecules that generate the rest, is doing both halves of this at once. Electrons and molecules, same problem, same direction of travel.

The honest ceiling

I should be straightforward about the limit of all this, because it is the strongest objection to my own argument and I made it myself in the last post.

Attribute certificates are, in an important sense, derivatives — of a carbon price, a tax, or any enforceable constraint on emissions. Where one of those sits underneath them, they can be valued, hedged and traded. Where one does not, a trader cannot price the option and will not pay for it — and that remains the condition across much of the world, including most of the region I now work in. Carbon pricing revenues reached around $107 billion in 2025 and now cover roughly 29% of global emissions, which is real, but it also means 71% of emissions carry no direct price at all.

So these pools are deep where compliance is real and shallow where it is not. Value capture follows the carbon price with a lag. Anyone selling you a business plan that assumes otherwise is selling you the derivative without the underlying.

Where this goes

The short version: methodologies will stay commodities, certification has just been handed a regulatory moat, registries are building a toll road that is not finished, and the two businesses I would want to own today are the data layer — because it earns its keep even if carbon policy stalls — and, less obviously, the financing advantage that accrues to whoever can prove their numbers.

And there is a further destination in view. Once intensity is certified per cargo rather than averaged per company, it stops describing a company and starts grading a product. That is how a commodity de-commodifies: not through a premium, but through differentiation that buyers, lenders and regulators can all verify. Gas priced on its carbon as well as its calories is not here yet, and I would label it a thesis rather than a forecast — but every layer of the stack is being built as though it were coming.

Who ultimately captures the value when that happens — the producers, the platforms, the certifiers or the exchanges — is the question this series ends on, and I will not pretend to settle it here.

What would you pay for, in this stack — and what would you expect to get for free?

Next: Who’s buying? Why the fastest-growing new buyer of low-carbon molecules is not a government or an oil major, but a data centre.

Subscribe now

Share

Sources & notes. Carbon pricing scale: World Bank, State and Trends of Carbon Pricing 2026 — 87 direct carbon-pricing instruments (47 carbon taxes, 40 emissions trading systems) covering ~29% of global GHG emissions, with revenues of ~US$107bn in 2025 (ETSs ~$87bn; carbon taxes ~$20bn), data cutoff 1 April 2026. CBAM non-linearity: analysis of hot-rolled coil indicating a ~10% cut in emissions intensity can reduce certificate requirements by ~30% (Fastmarkets); same source for Hyundai Steel on European automaker demand and market access.

Buyer specifications: SSAB–Volvo Cars (SSAB Zero, serial production); Stegra offtakes with Mercedes-Benz, Porsche, Scania; Meta RFPs for MiQ-certified low-methane gas (May and June 2026); Buy Clean California (facility-specific verified EPDs and maximum GWP limits from 1 January 2025).

Certification and the EU Methane Regulation: Commission recommendations of 20 July 2026 on the application of Article 33 and on optional model contract clauses (energy.ec.europa.eu); these are soft-law instruments and do not alter the Regulation’s underlying obligations.

Market infrastructure: Platts/Xpansiv daily Methane Performance Certificate assessments launched 2021 and discontinued July 2025 on delisting for thin liquidity; 3.5 million MiQ certificate block trade settled on Xpansiv CBL, announced March 2026; MiQ–CBL clearing arrangement. Cost of capital: sustainability-linked instruments typically apply margin adjustments in the region of 5–25 basis points against verified performance targets, with emissions the most common indicator; ratchet structures contracted in parts of the market after criticism of weak target-setting.

Granular certificates: EnergyTag standard for hourly, time-stamped certificates and first accredited issuers; hourly matching for Google’s 24/7 carbon-free energy goal.

The four-layer map is the live Standards Stack (mauriciobermudezneubauer.github.io/ghg-standards-stack). EU Methane Regulation timeline and interpretation: reporting obligations from 2025, importer MRV requirements from 2027, methane-intensity requirements from 2030, first review 2028; non-compliance attracts proportionate penalties rather than an automatic import ban — Margriet Kuijper, “The EU Methane Regulation: From Voluntary Action to Mandatory Accountability” (June 2026). The EU MRV framework draws substantially on OGMP 2.0, developed by industry.

Insurance: Chubb methane criteria for oil and gas underwriting (2023); AXA–Enel sustainability-linked insurance programme.

Carbon-differentiated price assessments: Fastmarkets low-carbon aluminium differentials (Europe P1020A, 4 tCO2e/t Scope 1+2 threshold; extended to Asia and the US) and daily CBAM assessments; Argus Carbon (launched April 2025); ICIS carbon analytics; S&P Global Platts CASP green-steel methodology, with the Northern European green flat-rolled differential assessed at roughly €100–170/t in January 2026. The Commission’s Recommendation on optional model clauses (20 July 2026) sets out, in its Annex, criteria for recognising providers of compliance solutions (including provider independence) and confirms at recital 9 that the Regulation does not require physical tracing of molecules, deliveries or cargoes. MiQ has published its own mapping of its certification programme and CIRIS framework against the Commission’s criteria (miq.org, thought leadership) — a self-assessment by the scheme concerned. Judgements about where value pools, and the de-commodification thesis, are the author’s reading and labelled as such.

The week I write this made the point for me. On Friday 17 July 2026, the European Commission proposed to slow its own carbon market: softer annual cuts to the emissions cap, permanent removals and international credits allowed in, billions in extra free allowances for industry out to 2038, and a correspondingly gentler ramp for the carbon border tariff. Three days later it went further, softening enforcement of its own methane-import rules — proportionate, phased penalties and optional model contract clauses to ease compliance. If regulation were the only lever, that would be the market quietly losing power. It isn’t. The ETS revision is only a proposal — the European Parliament sets its position by year-end — and the methane easing is a soft-law recommendation, not a binding rule; either way, the market they were meant to build carries on forming around them.

Product level carbon measurement is being turned into a market by six forces — and only two of them are regulatory. Which is why softened carbon rules in Europe change the speed, not the direction.

The useful question is not “will regulation force this?” It is “what, exactly, is turning a voluntary disclosure number into a business metric you cannot trade without?” My answer is a system — six forcing functions, working on the same commodities and products from different directions.

The six forces

See the full Standards Stack diagram and details in my Github repo here.

Think of the top of the Standards Stack as six levers on the same price, each working from a different direction. Ranked, roughly, by how directly each one moves it:

1. Compliance pricing. Border tariffs, emissions-trading schemes, carbon taxes, methane-import rules, fuel-intensity limits — the instruments that attach a legal cost to a tonne of carbon. This is the most visible lever — and the one the EU just eased on two fronts at once: its ETS carbon cap and its methane-import rules.

2. Disclosure regulation. Mandatory reporting — Europe’s CSRD, the ISSB’s IFRS S2 now adopted or mandated across twenty-plus jurisdictions, the coming product passports — that forces a firm to measure and assure its value-chain emissions before anyone puts a price on them.

3. Capital-market and stakeholder commitments. Validated net-zero targets, science-based pathways, transition plans, investor pressure — soft law that behaves like hard law once a board has signed it, because a target commits a firm and cascades to its suppliers.

4. Cost of capital and insurance. Sustainability-linked loans and bonds that move a spread on a verified carbon intensity number; insurers pricing physical and transition risk into premia and cover for production assets. This is carbon entering the price of money itself.

5. Buyer demand and procurement. Large, well-capitalised customers writing carbon attributes into their purchase orders — often ahead of any regulator.

6. Market infrastructure and price discovery. The exchanges and registries that turn an attribute into something with a clearing price and an audit trail.

Only the first two are regulation in the strict sense. The other four are the market building its own plumbing. Miss that, and Friday’s news reads as a retreat. See it, and Friday’s news is one lever eased while the other five hold.

The transmission mechanism

Here is the part many observers leave out: how the pressure travels. Today it travels mostly through Scope 3 — the value-chain emissions that dominate almost every industrial company’s footprint. The mechanics are crude but powerful. Each firm quantifies the emissions of the whole chain it sits in, largely on average (i.e. not measured) factors; a large buyer’s disclosure obligation becomes a demand for real data; that demand lands on suppliers as a specification they must now meet. Your carbon exposure becomes your supplier’s problem, and then your supplier’s supplier’s, and so on.

A cleaner rail is emerging alongside it. Ledger-based accounting — the “e-liability” method of Kaplan and Ramanna, now being taken up by the industry-backed Carbon Measures coalition launched in October 2025 (ExxonMobil, ADNOC, BlackRock’s GIP, BASF, Nucor and others, with first draft intensity standards for power and steel due in 2026) — measures embedded emissions once, at source, and hands them on with the product at each transaction, inherited from direct suppliers like a financial liability. Same pressure, a tidier mechanism: assigned once, auditable, supplier-specific, rather than every firm independently estimating everyone else’s footprint. It is early and contested — on boundaries, on double-counting, on who governs it — so treat it as a promising challenger, not a settled answer. But the direction is telling. Both routes are trying to do the same thing: move a credible carbon number up a value chain without it degrading into a guess.

Two routes, one destination

Watch the pressure crystallise into an actual price and you see it arriving from two directions at once.

Bottom-up, from buyers. Volvo’s near-zero steel, Microsoft’s SAF certificates, Maersk’s green methanol, Meta’s low-methane gas. All of these large buyers with no direct regulatory obligation to do so are using their purchasing power to write a carbon-intensity spec into their supply chains. That is demand manufacturing standards.

Top-down, from regulation. The EU’s carbon border tariff (CBAM)— now in its compliance phase, with certificate prices tracking EU allowance auctions — reaches through steel, aluminium, cement, fertilisers and hydrogen and turns each importer’s supplier data into a compliance input. And it is no longer unique: the UK’s CBAM arrives in 2027, with Australia, Canada and Taiwan weighing their own border measures. In a draft published in May 2026, the Commission proposed letting high-integrity Article 6 credits offset up to a tenth of a CBAM bill — but Parliament’s environment committee stripped that out in July, pushing the question into the ETS revision instead. Either way, the intent is unmistakable: convert measured carbon intensity into a priced product attribute.

Both routes end in the same place — a carbon number a buyer will ascribe value to, or a regulator will levy a charge against. For now, that value rarely shows up as a price premium — desks report that certified intensity buys market access and contract stickiness, occasionally a few basis points on financing, far more often than a per-tonne premium. The premium may come; the access is already here.

The plumbing is real, but young

The sixth lever — market infrastructure — is where I would temper the excitement, because it is the newest and the most fragile. The honest picture is a mixed one. A daily methane-performance price benchmark launched by S&P Global Platts and Xpansiv in 2021 was quietly discontinued in July 2025 when the underlying contracts were delisted for thin liquidity — continuous price discovery, it turned out, was ahead of the market. And yet, in March 2026, a single block of 3.5 million MiQ methane certificates changed hands on Xpansiv’s exchange between a European supplier and an integrated major — one of the largest trades of its kind. The rails are being relaid, not abandoned: MiQ has since signed a clearing agreement with CBL/Xpansiv to deepen certificate liquidity — the same exchange, a second attempt at price discovery. So attributes do trade, in size; what does not yet exist is a deep, continuously-priced market for them. The registry and exchange layer is being built, torn up and rebuilt in real time. That is what an emerging market infrastructure actually looks like — not a smooth ascent.

There is a deeper objection, and it deserves a hearing. Attribute certificates are, in a sense, derivatives of a carbon price — and where no liquid compliance market sits beneath them, a trader cannot value the option and will not pay for it. That is precisely the condition across much of the world today. It is why the market-infrastructure lever is the last to fire, and why “access, not premium” is the honest description of where we are — not the destination, but the departure point.

The fair objection

The strongest counter to all this is that I am overcomplicating it — that regulation sits upstream even of the “voluntary” demand. Meta measures its gas partly because disclosure regimes and its own net-zero commitment oblige it to; buyer coalitions form in the shadow of coming rules. I’ll concede it: regulation is often the first cause. Disclosure and commitments are themselves levers two and three.

But two things follow, and they are the reason this is worth mapping. First, the velocity has moved. The fastest-moving demand today — hyperscalers, first-mover coalitions, differentiated-commodity buyers — is running ahead of the regulation that seeded it, writing specs the rulebooks have not caught up with. Second, and this is the week’s real lesson: when the regulatory lever is eased, the others do not let go. Two levers eased in a single week — a carbon cap and methane import rules — and yet a softer EU carbon cap lowers one price signal; it does nothing to hyperscalers’ data-centre problem, an insurer’s transition-risk model, or an investor’s fiduciary mandate. A system with six levers is far harder to break than one held up by a single lever — which is precisely why betting the whole transition on compliance prices, in either direction, has always been a category error.

Where this goes

So the number is becoming a price — pulled by buyers, levied by regulators, and slowly, unevenly, given markets to clear in. Which raises the question the next post takes up: if measurement is becoming a market, where in the stack does the value actually pool — and who captures it? My short answer is that methods tend towards commodity while data, certification and the registry itself become businesses. But that deserves its own piece. And there is a further destination in view: once carbon intensity is certified per cargo, it stops being a footnote and starts to grade the product itself — the first step toward pricing gas by its carbon, not just by its calories.

One thing to carry out of this one: the next time a carbon rule is loosened and someone declares the whole agenda dead, count the levers. Last week one of six was eased. Watch what the other five do next.

Watch this: whether the European Parliament, in setting its ETS-reform position by year-end, restores any of the ambition the Commission proposed to trim — and how the CBAM–Article 6 question, now pushed into the ETS revision, is finally settled — and whether the Commission’s move to soften methane-penalty enforcement (Article 33 recommendation, July 2026) blunts the import rules in practice.

Next: where value pools — and why methodologies are commodities while certificates, data and registries are businesses.

Subscribe now

Share

Sources & notes. EU ETS / CBAM reform: European Commission proposal to revise the EU ETS and Market Stability Reserve, tabled 17 July 2026 — slower cap decline, removals and international credits admitted, additional free allocation to industry to 2038, slower CBAM phase-in, and a new Industrial Decarbonisation Bank; a legislative proposal, with the European Parliament to set its position by end-2026 (Carbon Pulse; ICAP). CBAM compliance phase since Jan 2026, certificate value linked to EU ETS auction prices (EU Taxation & Customs). CBAM × Article 6: draft implementing rules (13 May 2026) capping Art. 6.2/6.4 credits at 10% of reported emissions — not adopted, ; removed from CBAM by Parliament’s ENVI committee (8 July 2026) and referred to the ETS revision (S&P Global; Fastmarkets). Buyer demand: Meta RFPs for MiQ-certified low-methane gas, 19 May and 25 June 2026 (Meta Sustainability). Market infrastructure: Platts/Xpansiv daily Methane Performance Certificate assessments launched 2021, discontinued July 2025 on delisting for thin liquidity (S&P Global); landmark 3.5 million MiQ Certificate block trade settled on Xpansiv CBL, announced 25 March 2026 (Xpansiv). Ledger accounting: Carbon Measures coalition, launched Oct 2025, e-liability method after Kaplan & Ramanna; draft power/steel intensity standards due 2026 (carbonmeasures.org) — proposed and contested. Disclosure: CSRD/ESRS and ISSB IFRS S2. EU Methane Regulation: Commission Recommendations on Article 33 and optional model contract clauses (20 July 2026) — soft-law easing of methane-import enforcement (energy.ec.europa.eu). The four-layer map is the live Standards Stack (mauriciobermudezneubauer.github.io/ghg-standards-stack). Claims of velocity and value capture are the author’s reading, flagged as such.

I have spent the better part of twenty-five years advising companies and governments on carbon measurement, value and markets, and the most common thing I meet — on trading desks, in boardrooms and in sustainability teams alike — is the wishful desire for the one right standard: the belief that if everyone would simply agree on a single, clear method, the confusion would end. My reading is that it probably won’t for years. The standards are not converging on one. They are stacking.

There is no single ‘standard’ for the carbon intensity of energy and industrial products. There is a stack — four layers deep — and once you can see the layers, the fragmentation resolves into something closer to a market map.

The fragmentation is real — and it is admitted

Start with an awkward fact. The two most-cited product-carbon rule-sets — ISO 14067 and the GHG Protocol Product Standard — are not the same document and do not fully agree. They overlap, they cross-reference, and a practitioner routinely has to choose between them or reconcile them by hand. Beneath them sits the horizontal life-cycle grammar (ISO 14040 and 14044). Above them sit sector-specific methods — the oil-and-gas emissions estimation guidance, the steel and cement GHG inventory protocols — then the certification schemes, the border tariffs, the disclosure regimes, the procurement guidelines. Map it out honestly and a single class of products or commodities is touched, by my count, by several dozen distinct standards, methods and instruments.

Not surprisingly, this may look like a mess that begs tidying: a call for harmonisation, a plea for convergence towards “one global standard,” or at least an industry- or product-specific one. The instinct is understandable and — I will argue across this series of posts— largely beside the point. The mess is not a bug to be fixed. It is a structure to be read (and yes, also improved).

Read it as a stack, not a list

Here is the reframing that makes the apparent confusion more tractable. Stop lining these standards up as competitors on a single shelf, and start reading them as a stack of layers, each answering a different question.

1. The grammar. At the base sit the horizontal, product-agnostic standards. Life-cycle assessment (ISO 14040/44) is the parent method — and a multi-impact one, of which carbon is a single category. Product carbon footprinting (ISO 14067, and the GHG Protocol Product Standard) is its greenhouse-gas application; the Type III environmental declaration (ISO 14025, EN 15804) is the verified format many footprints travel in. Between them they set the terms everything above is written in: what counts as the product, how to split emissions between co-products, and where the measurement boundary falls — cradle-to-gate, cradle-to-grave, or even sector-specific bookends such as well-to-wake. That boundary is a choice, declared or fixed by a product category rule; it is not a given.

2. The industry substrate. Above the grammar sits the sector-specific machinery that turns raw operations into activity data, and activity data into product-level inventories — deciding how emissions are measured at each value-chain step, then allocated and summed into a product’s carbon intensity or footprint: the oil-and-gas estimation compendia, the steel and aluminium life-cycle inventories, the cement CO₂ protocol. Call it the activity-data engine — unglamorous, indispensable, and increasingly where the data businesses live.

3. The product-attribute layer. This is where a molecule or a material acquires a purchasable specification — a methane-intensity grade, a low-carbon steel certificate, an environmental product declaration. It is the fast-moving layer and, as later posts will show, the one where the commercial action is.

4. The forcing functions. At the top sit the things that turn all of the above from voluntary bookkeeping into a requirement you cannot sell without: border tariffs, import rules, emissions-trading schemes, disclosure mandates — and, critically, also the part many commentators miss, the demand of large buyers for these attributes and the pricing of the capital markets themselves.

Cutting across the four is a process strand for carbon capture and storage, which follows its own logic and its own incentive systems. But the shape is the point: four layers, one on top of the other, each doing a distinct job.

See the full stack diagram and details in my Github repo here.

Once you can see the layers, the fragmentation arguments need to be understood through a different lens. What a mill reports as its ISO 14067 footprint, what a green-steel certificate attests, and what CBAM uses as the basis to calculate liability on the same shipment are three different things — a number, a threshold, and a tariff base — not three attempts at one number. Seeing the stack turns the naïve — “which standard is the right one?” — into a better one: “which question are you actually answering?” And the bigger point: the whole stack is being wired into markets.

One clarification worth holding onto, because it is where most of the confusion lives. Two of the layers — the grammar and the substrate — are not rival standards so much as two halves of one calculation, and the genuine disagreement there is narrow: how you allocate between co-products, where you draw the boundary, whose data you trust. The layers above don’t re-answer that question; they ask a different one. Many disagreements about “the right number” are really that muddle — a methodological choice mistaken for a difference of purpose.

Why a map, and why now

I built the stack because an overall map of the board felt missing. Plenty of people know one layer intimately — the LCA and PCF consultants, the EAC traders, the CBAM compliance teams — but the layers are rarely drawn together, and the linkages between them are where both the risk and the opportunity sit.

The timing is not incidental either. Between now and 2030 the top layer fires almost at once: the EU’s carbon border tariff enters its definitive phase, its methane import rules begin to bite, shipping is folded into the EU ETS, and aviation and shipping face their first attempts at global lifecycle-fuel standards. Other jurisdictions may soon follow: the UK's CBAM starts in 2027, with Australia, Canada, Brazil and Taiwan weighing their own, and Türkiye exploring one under pressure as an EU exporter. Each of those quietly converts a voluntary method somewhere lower in the stack into the cost of market access. That is the real subject of this series, and hopefully the reason this map earns your time: the stack is being wired into markets. Carbon intensity measurement is becoming a market.

I want to be clear that I am not saying that standards fragmentation is not an issue. Indeed, in late 2025, ISO and the GHG Protocol — the two bodies that defined product-carbon accounting for a decade — announced they would jointly build a single, unified product standard, merging ISO 14067 with the GHGP Product Standard. You do not merge two things unless the split was costing someone real money.

And it is not only the base that is moving — the whole stack is in flux. At the bottom, the grammar is converging (that merger) and being challenged at the same time, as a rival, ledger-based school of accounting emerges to contest it. In the middle, the substrate and product-attribute layers keep expanding into new sectors — steel, aluminium, cement, chemicals — and new products — hydrogen, ammonia, marine and aviation fuels — each arriving with its own methods and certificates. At the top, the forcing levers multiply and tighten (and sometimes, loosen) year on year. This is not a diagram you draw once and hang on the wall. It is a dynamic map you have to keep up to date— which is rather the point of doing it in the open.

Where this goes

Over the next several posts I will walk up the stack and out into the market it is becoming: how measurement turns into a tradable market and which forces are driving it; where, precisely, the value is pooling; why the AI data centre may quietly become one of the most important new buyers of clean molecules; and who, in the end, captures the value when the carbon intensity of everything becomes a number a buyer can underwrite, a regulator can enforce and an auditor can sign.

For now, one idea is worth carrying out of this piece, and it is the simplest. The next time two carbon standards seem to disagree, don’t ask which one is right. Ask which question each of them is answering.

What would change in your business if you could see all four layers at once?

Next: how measurement becomes a market — the six forces turning a voluntary number into the price of entry.

Share

Sources & notes.
Standards. ISO 14040/14044 — life-cycle assessment, the multi-impact method; ISO 14067:2018 — product carbon footprint, its GHG application (iso.org); the GHG Protocol Product Standard, functionally parallel to ISO 14067 and not fully aligned with it; ISO 14025 / EN 15804 — Type III environmental product declarations.
The merger. ISO and the GHG Protocol announced a partnership (Sept 2025) and a joint working group (Oct 2025) to build one unified product-level standard (ghgprotocol.org).

The map. The full four-layer stack — several dozen standards, methods and instruments, sixty-plus at latest count — is live at The Standards Stack.

Border tariffs beyond the EU. The UK's CBAM is legislated to begin from 2027 (Finance Bill 2025-26; some sources describe implementation as 2027–28). Australia (carbon-leakage review completed March 2025, decisions expected in the safeguard-mechanism review July 2026 – June 2027), Canada (federal-backstop integration, targeting 2027–29), Brazil and Taiwan are considering their own; Türkiye is exploring one, driven by its exposure as an EU exporter rather than by a domestic proposal (WEF; IISD; Carbon Market Watch).

Pricing versus gating. The two mechanisms differ and the distinction matters commercially. CBAM prices market access — an importer surrenders certificates against embedded emissions rather than being refused entry. The instruments that gate access outright are the EU Methane Regulation’s intensity limits (from 5 August 2030), product-level maximum thresholds such as those provided for under the EU battery rules, and embodied-carbon thresholds in public procurement. The expectation that these arrive on the dates currently legislated is the author’s reading of published timetables, not a certainty; the methane enforcement timetable in particular was softened by Commission Recommendation on 20 July 2026, which leaves the obligations intact while recommending Member States withhold penalties for information failures due in 2027–29.

Buying on carbon intensity. Airlines are signing long-term SAF offtake agreements (RMI SAF Outlook); CF Industries reported its first sales of low-carbon ammonia at a premium in 2025 (FY2025 10-K).

“Twenty-five years” refers to the author’s career across carbon markets, MRV and decarbonisation.