Microsoft cuts carbon removal purchases by about 80%, report says, as AI-driven emissions rise

Microsoft is reportedly on track to sharply reduce purchases of carbon removal credits, a shift that would mark a major retreat from its approach over the past two years even as its emissions profile trends higher, according to a report cited by market news coverage.

The coverage says the company’s carbon removal purchases could fall by about 80%, an action framed as a response to rising emissions attributed to the growth of its AI workloads. The report further characterizes AI-related emissions as climbing by roughly 25%.

Carbon removal credits are used to finance activities intended to take carbon dioxide out of the atmosphere, such as certain types of engineered capture and storage or bio-based removal projects. Companies often buy these credits as part of longer-term net emissions strategies, especially when reductions in near-term operational emissions are constrained by electricity demand or supply chain realities. In Microsoft’s case, the reported change suggests the company may be revising the balance between near-term emissions growth and the pace of offset-like purchases.

The same coverage notes that the potential cut would be Microsoft’s first retreat from carbon removal purchases since 2023, implying that its prior ordering cadence had been consistent enough to be treated as a baseline. The report also indicates the company’s emissions pressure is not only a function of absolute operating growth, but specifically tied to the scaling of AI infrastructure.

Microsoft has made climate-related claims in recent years through a mix of operational targets, supplier expectations, and use of carbon credits. However, the market-news account provides limited detail on the mechanics of the alleged change, such as whether Microsoft is reducing the volume of credits purchased, switching to different types of removal projects, or altering the timing of purchases and retirements.

Sector-wide, the issue sits at the center of a broader debate about how rapidly large AI adopters can decarbonize data center power and the devices used to support model training and inference. Even when companies buy renewable electricity or invest in efficiency, AI demand can increase total energy consumption quickly. That can make it harder for firms to keep emissions within a narrow trajectory without either faster grid decarbonization or more reliance on higher-quality removal credits, both of which can be costly and subject to availability.

What remains unclear from the reported coverage is the exact documentation behind the claimed 80% cut. The post does not specify the credit categories involved, the contract counterparties, whether Microsoft has already renegotiated existing arrangements, or if the reduction is driven by revised internal accounting, changes in forecasted emissions, or market conditions affecting credit supply and pricing.

Investors and stakeholders will likely watch for Microsoft to clarify the figures in its next climate and sustainability disclosures. Key items to look for include updated forecasts for emissions tied to AI, any explanation of changes to carbon removal credit purchasing strategy, and confirmation of how any credits are accounted for over time, including retirements versus future deliveries.