Why Data Centers Use So Much Water — What Everyone Gets Wrong
The short version: Daniel Stih's breakdown of how data-center water stats get calculated is a useful reality check, but measurement nuance shouldn't let real local water deals off the hook.
Our Take
Credit where it's due: Daniel Stih is asking the right question that too many headlines skip — when you see "millions of gallons," what exactly is being counted? Direct cooling water at the facility is one thing. Water used upstream to generate the electricity that runs the servers, or water baked into manufacturing the chips and steel, is another. Lumping them into one scary number can make you solve the wrong problem, or worse, argue about the wrong thing entirely.
But here's where we'd push back a little: this kind of methodological caution can just as easily become cover. Sure, the accounting is messy — but somebody in your county still signed a permit letting a hyperscale campus draw from a local aquifer or municipal system during a drought, and that number is very real to the utility customers next door. Understanding attribution doesn't make the deal disappear; it just means residents deserve to know which number applies to their tap, their bill, and their watershed, not a blended national average.
We'd rather see communities get sharper questions, not fewer of them: how much of that water is potable, what's the source, what did the county actually agree to, and what happens when the next facility gets built next door. Check our facility map to see what's been approved near you, and if your local deal looks light on water disclosure, our take-action page has a place to start.
This is GridWatch the USA’s original commentary. The video above is the work of Daniel Stih, published on YouTube — full credit to the creator.