AI is one of the fastest-growing sources of electricity demand, and its environmental footprint is under more scrutiny every month. You might expect that to show up as a wave of carbon credit buying. It hasn't.
Our H1 2026 Carbon Market Trends Report found that, with the single exception of Microsoft, demand for credits from the AI sector has actually declined in recent years. Our co-founder Anton set out the scale of that gap in his analysis for illuminem, looking at the offsetting activity of more than 130 companies across the AI value chain. The finding is clear: the industry's hunger for energy is not yet translating into demand for carbon credits. The more useful question now is why, and what would change it.
AI's biggest footprint is its electricity use
The reason is straightforward. AI's largest climate footprint today is the electricity its data centers consume, and the sector is addressing that directly by buying clean power rather than carbon credits.
Big Tech firms are the largest corporate buyers of renewable energy in the world. In 2024 they accounted for 43% of all clean-energy power purchase agreements signed globally, and the deals have only grown since. Meta has contracted up to 6.6 GW of nuclear power, Amazon signed a 1.92 GW nuclear agreement with Talen Energy, and Google struck the first corporate small modular reactor deal. For these companies, securing power has shifted from a sustainability goal to a business-continuity priority, and that spending flows into generation, not credits.
There is a structural reason too. Power purchase agreements tackle the emissions from electricity use, which is exactly where AI's footprint is concentrated. Carbon credits mostly address residual and hard-to-abate emissions, and increasingly removals, which come later in a company's decarbonization journey.
A firm building out data centers reaches for clean electricity first, because it is the larger and more immediate lever. Credits become relevant once the easier reductions are made and a net-zero target turns attention to what is left.
Microsoft is the one AI-linked company buying credits at scale, and the reason is its target. Its commitment to be carbon negative by 2030 requires large volumes of removals now rather than later, which is why it has accounted for the overwhelming majority of the sector's credit activity. Most other AI firms have not set comparable removal-backed targets, so they have no equivalent pull toward the market.
That also makes the sector's demand fragile. Microsoft has signaled a slowdown in new purchases, and because it has carried the sector, its pullback removes the main source of AI-linked demand rather than redistributing it.
A few things could turn AI's energy footprint into credit demand. SBTi's new Corporate Net Zero Standard, released in June 2026, mandates carbon removal for residual emissions from 2035, giving committed companies a clear reason to build removal portfolios over time. Maturing net-zero targets across the sector would do the same and the gap left by Microsoft's slowdown leaves room for another large buyer to step in. Anton's illuminem piece sets out how AI firms could take that step.
Conclusion
For now, AI's growth is showing up in power markets, not the carbon market. The energy spend is real and large. It is simply going to clean electricity rather than credits. Whether that changes depends less on how much energy AI uses and more on whether the sector sets the kind of removal commitments that pull it into the market.
AlliedOffsets' H1 2026 Carbon Market Trends Report covers AI demand alongside buyer trends, carbon dioxide removal, and policy.
You can explore these findings in more detail here.