Artificial intelligence could drive a net increase of 0.47 to 1.8 gigatonnes of CO₂ emissions a year, according to peer-reviewed research cited by Jefferies. The study, published in npj Climate Action, found that emissions linked to AI-enabled fossil fuel production alone could be 3.3 to 13.3 times current emissions from data centres. Jefferies' sustainability and transition strategy team said the research challenges the way AI's environmental impact is typically assessed, noting that analysis often focuses on whether AI can speed the shift to cleaner energy while accounting for the extra electricity needed to run data centres, and can overlook a third factor: AI's use by oil and gas companies to increase fossil fuel output. AI can help energy companies cut drilling costs, locate new oil and gas deposits and improve recovery rates from existing fields, the study found, and those productivity gains can make fossil fuel projects more attractive and bring additional supply online more quickly. While AI can also improve renewable power generation and make energy systems more efficient, the research found the emissions avoided through those applications may be outweighed by additional emissions from increased fossil fuel production and consumption. Jefferies said the findings are relevant to investors assessing the broader impact of the rapid expansion of AI infrastructure, as the technology's energy footprint extends beyond the electricity consumed by data centres.
Jefferies Financial Group IncJefferies is cited as the source of the research but the article reports no company-specific financial impact
Taishin Financial Holding Co Ltd
Colgate-Palmolive Company
NovaGold Resources Inc
NVIDIA Corporation