Particle.news
Download on the App Store

Cornell Study Warns AI Data Centers Could Add 24–44 Million Tons of CO2 Annually by 2030

The Nature Sustainability analysis finds net‑zero pledges at risk unless siting, grid decarbonization or efficiency change.

Overview

  • The peer‑reviewed, three‑year study projects annual water withdrawals of 731–1,125 million cubic meters by decade’s end, equal to the household use of roughly 6–10 million Americans.
  • Researchers translate the projected emissions to the equivalent of adding about 5–10 million cars to U.S. roadways each year.
  • The paper maps mitigation levers, finding that smarter siting, cleaner power and operational improvements together could cut emissions by about 73% and water use by about 86%.
  • State‑level modeling highlights benefits in Midwest and wind‑belt locations with stronger renewables and lower water stress, contrasting with strained hubs such as northern Virginia and water‑scarce western sites.
  • Local pushback is already reshaping projects, with Data Center Watch tracking roughly $64 billion in developments stalled, and even high‑renewables scenarios leave residual emissions that require additional clean capacity.