AI Data Center Growth Drives Major Power and Water Demands

by Clarence Oxford
Los Angeles CA (SPX) Nov 11, 2025
Cornell University researchers have assessed the environmental impact of expanding artificial intelligence data centers across the United States. Their analysis finds that, if current growth continues, these facilities could emit 24 to 44 million metric tons of carbon dioxide and consume 731 to 1,125 million cubic meters of water yearly by 2030. These figures are comparable to adding up to 10 million cars to American roads and matching the annual water use of as many as 10 million US households.

The investigative team analyzed financial, manufacturing, and geolocation data to map the sector's power and water footprint. They concluded that without major sustainability efforts, the AI industry risks missing its net-zero emissions targets.

The authors recommend strategies including smarter facility siting, accelerated grid decarbonization, and increased operational efficiency. Adopting these measures could reduce associated carbon emissions by about 73 percent and water consumption by 86 percent compared to worst-case scenarios.

Facility location proved decisive. Data clusters have emerged in regions with high water stress, such as Nevada and Arizona, and in rapidly growing hubs like northern Virginia. Relocating data centers to areas with lower water risk and improving cooling technologies could halve water usage and deliver overall reductions of up to 86 percent when paired with efficiency improvements. The researchers identified Midwest and "windbelt" states - Texas, Montana, Nebraska, South Dakota - as optimal sites balancing both carbon and water impacts.

The study also notes New York's favorable profile due to its power mix of nuclear, hydroelectric, and renewables, especially when coupled with water-efficient cooling.

Grid decarbonization alone will not resolve the challenge. In a high-renewables scenario, emissions would still fall only 15 percent, leaving 11 million tons of CO2 for wind and solar deployments to offset. The researchers suggest that combining advanced cooling systems and better server utilization could drive an additional 7 percent drop in emissions and cut water use by 29 percent.

"This is the build-out moment," said Fengqi You. "The AI infrastructure choices we make this decade will decide whether AI accelerates climate progress or becomes a new environmental burden."

The peer-reviewed study appears in Nature Sustainability. The research team included collaborators from KTH Royal Institute of Technology (Sweden), Concordia University (Canada), and RFF-CMCC European Institute (Italy), supported by the National Science Foundation and the Eric and Wendy Schmidt AI in Science program.

Research Report:Environmental impact and net-zero pathways for sustainable artificial intelligence servers in the USA