Microsoft underwrites carbon removal in six-million-credit round

Microsoft just bought six million carbon-removal credits the way a lender would, with dynamic baselines and soil-level measurement attached, a structure that turns the anchor buyer into the underwriter of a market that has run on trust. The purchase, announced this week, moves the voluntary carbon market further into the phase where diligence is priced into the asset instead of assumed away.

For years, corporate carbon commitments traded on a basic bargain—the buyer trusted a project developer's accounting, and the developer trusted that the credit would stay credible—but the weakness was that the bargain ran on reputation, and one bad project can poison the entire pool. Verification existed but cost money, and in a market where credits are cheap, the incentive to skimp was strong.

Microsoft's terms attack that weakness at the point where it lives: dynamic baselines point to a counterfactual that moves as conditions change, rather than a fixed assumption set at contract signing, and soil-level measurement grounds the carbon claim in observation of the ground itself, not in regional models. Together they turn a credit from a certificate of intent into a quantified claim about a specific piece of land, expressed in numbers that can be checked.

Additionality has always been the soft spot in carbon accounting, because a static baseline lets a developer define the counterfactual once and leave it alone while a dynamic baseline makes the counterfactual live, so the credit adjusts as conditions change. That is more rigorous and more expensive to administer, and it is the difference between an accounting assertion and a monitored position.

The lender analogy is precise: a lender does not accept a borrower's assurance about cash flow; it underwrites collateral, tests projections, and writes covenants that let it react when the deal drifts, and Microsoft is doing the functional equivalent to its suppliers. The size of the round is significant, but the structure matters more, because a purchase at this scale provides revenue certainty to the developers that win the contract and establishes the template for every negotiation that follows.

Microsoft has bought removal credits before, which showed commitment, but the latest purchase is different: dynamic baselines and soil-level measurement turn it into a specification for the market.

Once a buyer of Microsoft's scale asks for these terms, the seller's question is no longer whether to provide them; it is whether providing them is affordable. That is what underwriting does: it separates the projects that can carry the burden of proof from the projects that cannot. In the old market, a credit was a branded gesture, a way to say "we have offset our emissions" with a certificate attached; in the new market, a credit's value depends on its own verifiable properties, which is the difference between a donation and an investment.

This is the right direction, though an uncomfortable one. Project developers will have to build monitoring systems, hire analysts, and accept that a baseline can be revised downward, and that cost is also a competitive advantage: a developer that can show soil-level data will be able to sell to any buyer in the market, while a developer that cannot will be limited to the thin tail of buyers who do not care. The market's long-term value depends on this re-pricing, even though it will be painful for the projects that cannot keep up.

The effect will be felt beyond Microsoft's suppliers, because when an anchor buyer sets terms, those terms become the default in the next negotiation, and smaller corporate buyers are likely to adopt Microsoft's standard, partly because it is safer and partly because it is easy to point to. Project developers who want access to the main pool of demand will have to meet the higher bar, and that is a prediction, but it follows from the ordinary economics of negotiation rather than from optimism.

The interesting question is what happens to credits that cannot meet the bar: they will not disappear but will be sold to buyers with lower standards, at lower prices, in a segment that increasingly resembles a discount market. That segmentation is healthy, though it will not feel that way to developers on the wrong side, because the market is moving from a single price for a vague claim to a range of prices for claims of different specificity. That range is what a functioning market looks like.

Price discovery has been clumsy in carbon removal because the underlying instrument was not comparable: two credits from different projects were treated as the same product even when they represented very different levels of risk. Microsoft's terms create a way to separate them, because a credit with soil-level measurement and a dynamic baseline is not the same instrument as a credit with a static baseline and a modeled estimate, and it should not trade at the same price. That differentiation is the foundation of a real market.

There is a risk in the new terms that should be named: measurement is expensive, and if the burden falls hardest on small projects, the market could consolidate toward developers with balance sheets large enough to absorb the cost. That would be a loss, because carbon removal needs a long tail of projects, not only industrial-scale sites. The answer depends on whether measurement costs fall fast enough to keep the long tail alive, and the structure of the next few rounds of financing will reveal that.

For allocators and family offices holding carbon-removal exposure, Microsoft's terms are a useful screen: ask which portfolio companies can meet this standard today, and which can meet it in two years without diluting their margins. The companies that treat measurement as a product feature are the ones with a market after the model changes, while the companies that treat it as a compliance cost are already behind; the gap will show in their next fundraise, not in their current marketing.

Microsoft's six million credits are, on the surface, a purchase, but the terms attached to them are an underwriting standard, and underwriting is how markets grow up. The voluntary carbon market has spent years defending itself against the charge that it sells air; a credit with a dynamic baseline and soil-level measurement is a position that can be tested, not air. The next round of carbon-removal financing will price the difference.