AI could add Russia-sized emissions via fossil fuels: study
Get the Energy newsletter
Daily energy & climate — solar, EVs, oil, the policy fights and tech bets shaping the transition. Free.
- Will and Holly Alpine, former Microsoft sustainability workers, published research in npj Climate Action finding AI could increase global energy-related emissions by 1.2-4.8% as a productivity enhancer across fossil fuel extraction, refining, and electricity generation.
- The study's low-end estimate puts AI-enabled fossil fuel emissions at Mexico's annual output, while the high end equals Russia's — the world's fourth-largest emitter.
- The Alpines quit Microsoft at the start of 2024 over the company's continued work with oil and gas and now campaign publicly on what they call "enabled emissions" — pollution from AI tools boosting fossil fuel production that tech companies don't measure.
- Chevron is building a behind-the-meter gas plant in Texas to power Microsoft data centers, and Chevron New Energies president Jeff Gustavson told analysts in June the company will "use some of that compute" to power AI inside Chevron.
- The research finds that AI's emissions boost to fossil fuels outweighs its benefits to solar, wind, and other clean technologies and significantly outpaces projected data center emissions.
- Independent energy researcher Jon Koomey, not involved in the study, said machine learning can make data center cooling 30-40% more efficient but also make fossil fuel extraction "much cheaper and faster."
Why it matters: Tech companies measure their own operational emissions but ignore the far larger pollution their AI tools unlock in the fossil fuel sector. The Chevron-Microsoft gas-plant deal makes that gap concrete: Microsoft buys compute power from a plant Chevron is also using to expand its own AI capabilities.
Ask SkimNews
