AI diffusion linked to higher local CO2 emissions

CEPR discussion paper by Bonfiglioli, Crinò, Filomena and Gancia (2025) documents that AI penetration is associated with faster CO2 emissions growth across US commuting zones between 2002 and 2022.

• Main finding and scope: The paper uses a novel dataset linking AI penetration, data center and power plant locations, and CO2 emissions for US commuting zones (2002–2022) and, using a shift–share identification strategy, finds that localities more exposed to AI experience relatively faster emissions growth; scale effects dominate and electricity generation becomes more carbon intensive.
• Mechanisms and specifics: The authors report that AI penetration raises dependence on non-renewable electricity, with proximity to data centers identified as a key driver because nearby power plants shift toward greater fossil fuel use; no monetary values, contracts, or timelines for mitigation are provided.