The rapid expansion of artificial intelligence is creating an unprecedented thirst for water to cool warehouse-sized data centers, threatening to overwhelm local utilities.
A recent study conducted by UC Riverside and Caltech projects that within the next four years, peak daily water requirements for these facilities could surge by 697 million to 1.45 billion gallons. To support this massive draw, researchers estimate that between $10 billion and $58 billion in new water infrastructure will be required nationwide.
For the construction and utility trades, this data signals a looming boom in heavy civil projects. The core problem identified in the study is extreme peak demand. While a data center’s annual water averages might look manageable, their evaporative cooling systems can pull 6 to 10 times — and sometimes up to 30 times — their normal volume during the hottest summer days. To prevent dry taps, municipalities will be forced to construct high-capacity water treatment plants, pump stations, large-scale transmission pipelines and reservoirs specifically designed to handle these rare but intense summer spikes. This places a heavy burden on an aging national water grid that already requires trillions in basic upkeep.
The rush to secure resources is already underway. In February 2026 alone, major tech firms locked in rights for millions of gallons daily across Indiana, Louisiana and Virginia, driving nearly $1 billion in infrastructure costs. To prevent local ratepayers from footing the bill, the researchers urge developers to co-fund utility upgrades and transparently report peak usage. They also recommend outfitting data centers with hybrid systems that can switch between water-based cooling and electricity-heavy dry cooling, depending on whether the local water supply or the power grid is experiencing more stress.
Read more about it at the University of California.
