AI’s Energy Appetite

In the year ending March 2026, Microsoft, Amazon and Google collectively emitted 119 million tonnes of carbon dioxide equivalent, almost a fifth more than the year before, a combined footprint roughly equivalent to a third of France's annual emissions.

The increase is not a one-off. United Nations data show operational emissions across the world's leading AI-focused tech companies rose an average of 150% between 2020 and 2023, and the trend has continued as AI infrastructure buildout accelerates rather than slows. Alphabet's own emissions have risen 51% since it pledged in 2020 to halve emissions and run entirely on carbon-free energy by 2030.

For companies that built entire sustainability brands on climate leadership, AI's energy appetite is no longer a footnote to their ESG story. It has become the story, and the shift is arriving faster than the targets these companies set in 2019 and 2020 were designed to absorb.

Microsoft, Amazon, Google and Meta each committed to ambitious climate targets before generative AI reshaped their businesses: Microsoft to be carbon negative by 2030, Amazon to reach net zero by 2040 through its Climate Pledge, and Alphabet to run on 24/7 carbon-free energy across all its operations by the same year. These commitments were calibrated against a data center growth trajectory that generative AI has since overtaken.

The International Energy Agency now projects global data center electricity demand could reach 945 TWh by 2030, more than Japan's entire annual electricity consumption, driven overwhelmingly by AI training and inference workloads rather than the cloud computing growth these targets originally anticipated.

This gap between commitment and consumption is now showing up directly in reported numbers. Microsoft's total emissions rose 25% year-over-year in its 2025 fiscal year, which the company attributed primarily to data center expansion, and Amazon reported a 16% rise in total emissions, with supply chain emissions, including data center construction, up 20%.

The gap between the climate commitments tech companies made in 2019 and 2020 and the emissions reality of their 2026 AI infrastructure buildout is no longer a rounding error. It is now large enough, and visible enough, that investors are treating it as a distinct risk category.

Whether the sector's response is genuine accounting reform, following Microsoft's lead in removing non-additional instruments from its claims, or a retreat toward softer disclosure, will determine whether AI's energy appetite becomes a managed transition or a credibility crisis for corporate climate commitments broadly.