Voters Back Data Center Limits as Water Use Remains Hard to Pin Down

Voters signaled this week that they want limits on AI data centers, passing 40 of 41 ballot measures that would restrict some of the more than 2,000 new facilities planned across the country.

Water and power rank at the top of ratepayers' concerns. Yet pinning down how much each data center actually draws from local resources is difficult, since operators typically report usage companywide rather than site by site.

Shaolei Ren, a data center researcher and professor at the University of California, Riverside, said giant 1-gigawatt AI data centers from companies like Meta, Google, Amazon and SpaceX can consume up to 10 million gallons of water per day. The exact figure depends on the hardware inside, where the facility is built, and how the servers and building are cooled.

Modern AI data centers are packed with high-power GPUs that run so hot they require liquid cooling; water is roughly 3,000 times more efficient than air for the job, according to Equinix sustainability head Christopher Wellise. Most large AI data centers use closed-loop liquid cooling that consumes very little water daily. The bigger factor is external cooling, where heat is rejected to the environment.

That process can be dry cooling, which uses little water but far more power, or evaporative cooling, which uses little power but can lose large amounts of water on the hottest days. Many operators use hybrids. Meta, for instance, says its Lebanon, Indiana, site will use zero water for most of the year, though its law firm put the site's allocation at up to 8 million gallons per day by 2031, a peak Meta says it expects to rarely reach.

Some companies are building water treatment and recycling plants near projects, and chipmakers are working on cooler-running processors. Wellise said each new Equinix site undergoes a water risk analysis to choose a cooling method. "Communities are beginning to ask the right questions," he said.

Source: CNBC