Massive data centres collectively use as much electricity as an entire US state

AI-rewritten: This is a summary of an article from New Scientist, rewritten by AI (Qwen, running locally) to make it easier to read. The facts come from the original article – read it for the full story.

New Scientist • Madeleine Cuff • November 16, 2026

New figures indicate that the largest data centres in the US collectively use roughly the same amount of electricity annually as one of the country’s major states. Francesca Dominici at Harvard University notes that the rapid development of artificial intelligence systems has triggered a "fast and furious" expansion of these facilities, particularly very large "hyperscale" data centres supporting AI tools and cloud computing. However, tracking their power consumption is difficult because many operators do not publicly report such information.

Dominici and her colleagues spent two years analyzing 403 hyperscale data centres to estimate usage between May 2024 and April 2025. The team found these facilities used about 82 terawatt-hours (TWh) of electricity, equivalent to roughly 1.8 per cent of total US electricity use. This amount outstrips the annual power consumption of states like Missouri, Kentucky, and New Jersey. If the upper bound estimate of 99 TWh is accurate, hyperscale data centres would rival the annual power consumption of Indiana, Michigan, and Washington.

Researchers estimate that roughly 54 per cent of this electricity came from fossil fuel sources, resulting in annual carbon dioxide emissions of 25.7 million tonnes. While Alex de Vries-Gao at the Free University of Amsterdam says these findings are not particularly shocking compared to international estimates, he warns that the study relies on commercial data that is difficult to independently verify. Yury Dvorkin at Johns Hopkins University adds that the analysis is relatively "coarse" and ignores factors like network constraints impacting grid emissions.

Source: New Scientist • Madeleine Cuff • November 16, 2026

Read the original article at New Scientist →

Leave Comment

Your email address will not be published. Required fields are marked *