AI's Fossil Fuel Productivity Gains May Dwarf Data Center Emissions

Artificial intelligence may accelerate climate change not primarily through the energy demands of data centers, but by making fossil fuel extraction dramatically more efficient, according to new research published in the journal npj Climate Action.

The study, authored by former Microsoft sustainability employees Will and Holly Alpine, estimates that AI-driven productivity improvements across the oil and gas industry could increase global energy-related emissions by 1.2 to 4.8 percent annually. At the upper end, that represents greenhouse gas pollution comparable to Russia, the world's fourth-largest emitter. Even the conservative estimate equals Mexico's total annual emissions.

Why it matters

While technology companies extensively measure their operational carbon footprints and supply chain emissions, they typically ignore what the Alpines call "enabled emissions"—the pollution their AI tools help customers produce. This research suggests those enabled emissions from fossil fuel productivity gains significantly outweigh both the climate benefits AI provides to renewable energy development and the widely discussed emissions from powering AI data centers themselves.

The productivity paradox

Oil and gas companies have deployed various AI technologies for decades to locate underground resources and optimize extraction processes. The Alpines used complex economic modeling to project how documented AI productivity gains across fossil fuel extraction, refining, and electricity generation translate into increased global emissions.

Their analysis reveals what Will Alpine describes as "a self-reinforcing effect between supply and demand." As AI makes fossil fuel production cheaper and more efficient, it strengthens rather than weakens global dependence on oil and gas—working against climate targets even as the same technology advances clean energy.

The Microsoft-Chevron connection

The research gains particular relevance from recent industry developments. Microsoft and Chevron confirmed plans for the oil company to build a large gas plant in Texas specifically to power Microsoft data centers. During a June analyst call, Chevron's New Energies president Jeff Gustavson indicated the arrangement would be reciprocal: Chevron would "use some of that compute" to power AI applications within its own operations.

Chevron spokesperson Paula Beasley told WIRED the companies "have worked together for years to accelerate digital transformation" but did not address specifics about the AI relationship.

Challenging AI optimism

Energy researcher Jon Koomey, who was not involved in the study, called the findings credible. "There are many AI boosters who blithely claim that AI will solve the climate problem so we should go ahead and develop it as quickly as possible," Koomey noted. "Such hand waving arguments ignore the effects that AI will have on ALL industries, not just renewable energy and efficiency."

The Alpines left Microsoft in early 2024 specifically because of the company's continued partnerships with the oil and gas sector. Their research argues that while operational emissions accounting remains important, the technology industry's failure to measure enabled emissions obscures AI's most significant climate impact.

The details were first reported by WIRED.