AI productivity could raise fossil-fuel emissions despite renewables gains, study finds

A new study modeled AI’s impact on the entire power sector across 64 scenarios and found net yearly carbon pollution could rise by 0.47–1.8 gigatonnes CO2 (about 1–5% of energy-emissions) due to productivity gains in fossil fuels outpacing those in clean energy. Emissions only fall if AI-driven gains in renewables outpace fossil-fuel gains by at least fourfold; the research notes its estimates are directional. The analysis did not include datacentre energy use, but real-world deployment of AI in oil and gas has already boosted output and efficiency, according to industry, raising concerns about AI’s net climate impact.