AI Data Centers May Strain Water Supplies in US High-Stress Regions

• About two-thirds of new data centers built or planned since 2022 are in high water stress areas, per World Resources Institute data.

• Five states host 72% of facilities in the most strained locations.

• Global data center water use hits 560 billion liters annually, projected to double by 2030 amid AI expansion.

AI data centers water consumption surges in stressed U.S. regions, competing with residents for supplies. Discover impacts, stats, and political pushback in this analysis—learn how cooling tech drives 2M liters daily per site.

What is the impact of AI data centers on water consumption?

AI data centers water consumption has escalated dramatically, with facilities nationwide pulling millions of liters daily for cooling amid booming demand. A typical 100-megawatt U.S. data center uses around 2 million liters per day—matching 6,500 households—while evaporating 80% through cooling systems. Globally, these sites now account for 560 billion liters yearly, forecasted to reach 1,200 billion by 2030 as AI computing intensifies.

How do AI data centers contribute to water stress in the U.S.?

New AI-focused data centers have surged 70% in the past three years compared to prior periods, numbering over 160 sites, according to World Resources Institute and DC Byte data. Roughly two-thirds of builds or plans since 2022 target high water stress zones, where daily supplies already fall short. Five states dominate 72% of these strained-area developments, pitting corporate needs against local households.

Evaporative cooling, or swamp cooling, dominates: hot air passes through wet pads to chill server rooms packed with heat-generating AI chips. Shaolei Ren, associate professor of electrical and computer engineering at the University of California, Riverside, notes that centers evaporate about 80% of drawn water, returning just 20% to treatment—unlike homes, which lose only 10%.

An April International Energy Agency report highlights a 100MW facility’s electricity draw exceeds 75,000 homes, with water needs scaling directly to AI workloads. Municipal systems, burdened by droughts and growth, face amplified pressure from this inefficient cycle.

Frequently Asked Questions

Which U.S. states face the highest water stress from new data centers?

Five states account for 72% of data centers in the most water-stressed areas since 2022, based on Bloomberg analysis and World Resources Institute figures. These regions already compete for limited clean water, with AI builds exacerbating local shortages through high-volume evaporative cooling daily.

Why do AI data centers use so much water for cooling?

AI data centers rely on evaporative cooling to manage intense heat from advanced chips. Warm air flows through moistened pads, evaporating water to cool servers efficiently. This process consumes 2 million liters daily per 100MW site, with 80% lost to evaporation, far exceeding residential efficiency, as explained by experts like Shaolei Ren.

Key Takeaways

• Water stress concentration: Two-thirds of new data centers since 2022 are in high-stress U.S. areas, led by five states holding 72% of builds.
• Daily consumption scale: A single 100MW facility draws 2 million liters—6,500 homes’ worth—with 80% evaporated via cooling.
• Future projections: Global usage at 560 billion liters yearly could double by 2030; monitor political responses like moratorium calls.

Political resistance grows as expansion hits stressed regions

AI data center expansion sparks bipartisan backlash. Democratic Sen. Bernie Sanders of Vermont urges a nationwide pause, telling CNN: “Frankly, I think you’ve got to slow this process down. It’s not good enough for the oligarchs to tell us it’s coming, you adapt.”

Republican Gov. Ron DeSantis of Florida proposed an AI bill of rights on December 4, empowering communities to reject projects. At a Villages event, he stated: “We have a limited grid. You do not have enough grid capacity in the United States to do what they’re trying to do.” DeSantis polled residents: “Do you want a hyperscale data center? I think most people would say they don’t want it.”

Conclusion

AI data centers water consumption poses mounting challenges in U.S. high-stress regions, with new facilities consuming billions of liters amid evaporative cooling demands. Sources like the International Energy Agency and World Resources Institute underscore the scale, while experts such as Shaolei Ren highlight inefficiencies. Political figures from Bernie Sanders to Ron DeSantis signal growing resistance. As AI advances, balancing tech growth with sustainable water management remains critical—stakeholders must prioritize efficient cooling innovations.