AI Data Centers Face Water Limits
Coverage from EY, NC State, and others

AI-driven data center expansion is increasing scrutiny of water demand from facility cooling and electricity generation, particularly in water-stressed regions.
Evaporative cooling remains a major source of direct consumption, while closed-loop, liquid, air, and recycled-water systems could reduce withdrawals. Incomplete facility-level disclosure and inconsistent accounting are complicating permitting, community review, financial risk assessment, and long-term water planning.
The story now includes a much larger quantified estimate of indirect water use and indicates that water concerns are producing project-level disputes and legal challenges.
The update largely confirms the existing account while sharpening the projected 2030 growth outlook and the operational tradeoffs of lower-water cooling systems.
The story now places greater emphasis on electricity-generation water use, adding a substantially larger but less consistently measured footprint beyond facility cooling. Power producers and new quantitative estimates broaden the issue from data-center operations to the electricity system serving them.
The story has shifted from general concern about data-center water impacts to treating water as a quantified infrastructure constraint affecting project feasibility, siting, permitting, and financial risk. New accounting also highlights broader local burdens from electricity generation and wastewater-related flows.
Expanding U.S. data centers are increasing pressure on municipal water systems, aquifers, rivers, and watersheds, particularly in water-stressed regions. Evaporative cooling and electricity generation can raise water use, while closed-loop, dry, recycled-water, and rainwater systems may reduce withdrawals. Incomplete facility-level reporting is complicating permitting and local decision-making, and water availability is becoming a consideration for investors, lenders, insurers, and project developers.
