The term sheet is the scarce asset in corporate carbon

Four corporate carbon contracts landed within days of one another, each varying a different term in the same document: Google moved the delivery clock, the rice project moved the unit, GSK extended the tenure, Whirlpool changed the obligation. None is a technology bet, and the shared conclusion across the four is that engineering is the supplier's problem while the term sheet is the buyer's.

Two delivery clocks do most of the work in Google's largest carbon purchase, the contract with Terradot that puts methane abatement on a 2030 schedule and durable removal on a 2040 one — a decade of separation between two products that share a buyer, a counterparty and an announcement. The split is the part the next corporate buyer will copy, and the reason is practical rather than philosophical: a methane reduction can be counted inside a planning horizon, while permanent removal cannot, and a procurement team that blends the two into a single commitment is writing a promise whose second half it will not be able to keep on schedule.

Read closely, Google has bought two claims of different duration: the methane half is near-dated, deliverable and retirable well before the first clock runs out, while the durable half is long-dated and sits close enough to the horizon most large corporates now carry that folding it into the same line as a 2030 reduction would blur the difference between a reduction and a removal. Separate clocks let each half be measured against its own timeline, and they leave the developer with two cash-flow profiles to finance rather than one blended promise, with the near-dated piece built first.

The second Google contract makes the same argument in a different unit. The Brazilian rice project pairs superpollutant elimination with durable removal from the same ground, and the hectare, not the credit, is the number the next buyer will copy. A hectare is something an agronomist, a lender and a supply-chain auditor can each inspect, whereas a credit is an accounting construct that arrives downstream of whatever happened on the land; buying the hectare means buying the practice change and accepting the verified outcome that follows, a longer and less certain position than purchasing a ton off a registry but a more defensible one if the buyer's real exposure runs through agricultural supply chains rather than its own ledger.

There is a reason the unit matters as much as the clock: a buyer that commits to hectares is committing to land, equipment and farmers, while a buyer that commits to credits is committing to a registry's judgment about all three. The rice contract puts Google on the producing side of that line, and the hectare becomes the deliverable that a lender, an agronomist and a supplier can all agree on — harder to write that way and easier to check once it is signed.

Put the two deals side by side and the week's pattern shows corporate carbon procurement separating into products with distinct clocks, distinct units and distinct measures of success, and the buyers treating the contract as the thing they are actually buying. What Google purchased from Terradot was a 2030 methane claim and a 2040 removal claim; that both carry the same counterparty's name is a matter of sourcing, not a statement about the product.

GSK's eight-year offtake with Varaha takes the same logic further and then stops short of the part that matters. Eight years of contracted credit revenue gives a developer something rare: bankable cash flow, and that is what allows Varaha to bank a 50,000-hectare expansion, because the question a lender asks about a removal project is rarely whether the science works but whether anyone has promised to pay for the output, and GSK has.

The deal pairs that revenue with a machinery subsidy, and the pairing is the practical hinge: residue burning in Punjab stops when farmers have the equipment to clear fields differently and a buyer for what the changed practice produces. The subsidy supplies the first, the offtake supplies the second, and together they turn a practice into something a lender can underwrite.

Eight years is the term that does the work. The reason tenure matters is the gap between when a removal project needs capital and when it has revenue; a multi-year offtake narrows that gap by converting an uncertain future sale into a contracted one, and the 50,000-hectare target is financeable because the revenue is contractual. That is why tenure, not tonnage, is the term a copycat would be least able to change: a developer that can point to a signed buyer for eight years is a different credit from one pointing at a pipeline.

What the announcement leaves out is the price. Eight years of contracted credit revenue covers about 7% of GSK's forecast residual emissions, a share small enough to read as a pilot and large enough to matter as a template; the number that would let rivals copy the structure is the one term absent from the disclosure. Rivals can copy eight years and a machinery subsidy, but they cannot copy the economics, and the economics is the only part that decides whether they sign.

That omission is the difference between a deal and a precedent, and it is not unique to this contract: a steelmaker now operates storage it does not own under a hybrid power purchase agreement, with the terms that decide that deal's economics left out of the announcement. On the evidence of the week, structures are being published and prices withheld — a rational thing for any single buyer to do and a costly one for a market that has no way to compare one contract with the next, because published prices are what let a buyer weigh a four-year contract against an eight-year one, a hectare against a ton, and a 2030 clock against a 2040 one.

Whirlpool's carbon removal purchase is the one to study hardest, because it addresses the problem that stops most corporate buyers: they do not know how much durable removal they will need, or when. Firm offtakes give suppliers demand a lender can underwrite, while options let Whirlpool keep the second tranche to itself and decide later whether to take it at terms set now — an asymmetry that runs both ways, since the supplier gets a bankable base and the buyer gets the right rather than the obligation to buy more.

If the corporate carbon market converges on a single structure this decade, the option is the likelier candidate than the tonnage commitment, because it lets a buyer postpone the hardest question, its own abatement path, without surrendering access to supply. A firm commitment requires a company to know its residual emissions at the end of the decade; an option requires it to know only that it may want to buy, and most buyers, asked honestly, know the second and are guessing at the first.

The option is also what makes this the deal to watch for a second tranche. An option gets exercised or it lapses, and each outcome carries information: exercise says the buyer's removal demand is real at the contracted terms, lapse says it found something cheaper or decided it could wait. The coverage does not say which way Whirlpool is leaning, or whether the second tranche has a price attached; if it does, the market gets a forward price to argue with, and if it does not, the most copyable structure in carbon arrives with the same missing term as GSK's.

The smallest number in this story is the most telling. A sorbent order worth $696,000 in direct air capture, the first commercial sorbent contract in the sector, sized to a developer with seven installations and worth roughly $348,000 a year, is not a market clearing. It is a capacity test. A delivery date is only a priced asset if someone can deliver against it. If a supplier with seven installations is selling its first commercial order at that scale, the binding constraint on direct air capture is neither demand nor capital; it is the number of plants that exist to consume the material.

That puts an awkward edge on the contract-design story: a buyer can write the most careful term sheet in the removal market and still be waiting on a factory. The same instinct showed up elsewhere in the week, when Ara took a majority stake in its own engineer, Bryden Wood, on the bet that capacity beats capital. Whether a designer owned by its parent can still read the market for that parent is the open question the deal poses, and if the answer is yes, other buyers facing a buildout bottleneck have a template.

The number to watch next is a price. Whirlpool's option will be exercised or it will lapse at terms nobody has published, and the developer behind the $696,000 sorbent order can quote a larger one only as fast as it can build plants to fill it. Google's 2040 clock will run either way. The first buyer to publish a multi-year removal price hands the market a benchmark and its rivals a price to beat.