Microsoft has paused all new carbon removal credit purchases, with no public timeline for resumption. According to a Substack analysis by MAATTR, the company that represented roughly 90% of all carbon removal credit buying globally, and contracted more than 72 million tonnes over three years, told its CDR partners about the pause two weeks before the post was written. The official line from a spokesperson was the corporate non-answer: “We continually review and assess our carbon removal portfolio along with market conditions.”
Why this matters
If the 90% figure is accurate, then what we have called a “carbon removal market” was always one buyer with a portfolio. When that buyer pauses, the remaining 10% does not constitute a functioning market. It is a long tail of smaller commitments scattered across project developers, registries, and verifiers who organized their entire businesses around a single counterparty. The CDR field has spent three years telling itself that Microsoft was the lead, not the whole. The pause forces a more honest accounting.
The details
MAATTR’s piece adds a layer most coverage has missed. Microsoft was not only buying credits. It was building the rails. The company launched the Environmental Credit Service, an Azure-based platform designed to handle registry functions, credit issuance, provenance tracking, and verification workflows for the voluntary carbon market. Every project that ran through it was running through Azure. The Environmental Credit Service was quietly deprecated in February 2025. The purchasing pause followed roughly fourteen months later. MAATTR reads the sequence as deliberate: infrastructure play cut first, purchasing stopped second. The author draws an analogy to VHS versus Betamax. The cloud platform that locked in the world’s sustainability data would hold a durable advantage over AWS and Google. If that race is now considered over, or unwinnable, or simply not worth the cost of carrying a market on your back, the rationale for being the buyer of first and last resort weakens.
What this changes for CDR
First, it removes the comfortable fiction that the voluntary carbon removal market had achieved distributed participation. It had not. There was no second buyer with a different thesis and comparable capital waiting to step in. Project developers who modeled offtake demand against Microsoft’s published commitments are now modeling against a question mark. Second, it sharpens the distinction between durable engineered removal (DAC, mineralization, BECCS, which is bioenergy with carbon capture and storage) and nature-based credits whose accounting has always been contested. Microsoft’s portfolio spanned both. The pause does not discriminate. Third, it raises the question of what comes next for the suppliers who scaled hiring, permitting, and CapEx (capital expenditure on plant and equipment) against Microsoft offtake. Some have multi-year contracts already signed. Others were counting on the next tranche. Suspended animation, as MAATTR puts it, is not the same as collapse, but it is not growth either.
The honest read
Microsoft did more to move capital into carbon removal than any institution on Earth. The piece estimates roughly forty times more than any other buyer. That contribution is real and should not be erased by what happens next. The company funded first-of-a-kind DAC, mineralization, biochar, and forest projects when essentially no one else would write checks of that size. The harder truth is the one MAATTR lands on. A market with one buyer is not a market. It is a procurement program with a marketing budget. The CDR field benefited enormously from that program, but mistook the patron for the demand curve.
Caveats
A pause is not an exit. Microsoft has not said it is leaving the market. The company has standing contracts and could resume purchasing tomorrow. The MAATTR analysis is one writer’s reading of a sequence of events, including the Azure infrastructure thesis, which is interpretation rather than confirmed strategy. Microsoft has offered no public reasoning beyond the spokesperson statement. The 90% figure also deserves a flag. It comes from the post and tracks with what other trackers have published for 2025, but the exact share depends on how you define the market, which credits you count, and which buyers you include. Whether the true share is 80%, 85%, or 90%, the structural point holds: one buyer dominated, and that buyer paused. The other thing to watch is whether the rest of the hyperscalers, Google, Amazon, Meta, step up or step back. If they step up, the 10% becomes 30% and the market reorganizes around a more distributed buyer base. If they step back, MAATTR’s diagnosis becomes the consensus. For project developers, the operational answer is the same it always should have been. Build for multiple buyers. Price honestly. Treat any single offtaker, even one writing the biggest checks in the field, as a customer rather than a market. And for the rest of us, including those of us who write about CDR for a living, the pause is a useful reminder. CDR remains essential for hard-to-abate residual emissions. It is not, and never was, a substitute for cutting fossil fuel use at the source. The credits were always the easier conversation. The harder one is still waiting.
Source: open.substack.com