Everyone Says AI Is Draining Our Water. I Read the Actual Data.

Everyone keeps telling me AI is draining our water. So I went and read the actual data. Here’s what I found. (It’s not what you think.)

Sam Altman says the average ChatGPT query uses about one-fifteenth of a teaspoon of water. I don’t take CEOs at their word, so I checked the independent research. Data analyst Andy Masley did an exhaustive breakdown that’s been cited by Lawrence Berkeley National Laboratory, and the numbers largely hold up.

The Big Picture Numbers

In 2023, ALL U.S. data centers (not just AI, but every server running Netflix, your email, everything) used about 17 billion gallons of direct cooling water. That sounds like a lot until you realize it’s roughly 0.3% of the public water supply. And Masley’s projections show that even with aggressive AI growth, AI-specific data centers will use about 0.08% of U.S. freshwater by 2030.

Meanwhile: In Maricopa County, Arizona (one of the biggest data center regions in the country), golf courses used 29 billion gallons of water while data centers used 905 million. That’s over 30x more water for golf, while data centers generate 50x more tax revenue per gallon. Nobody protests golf courses.

And agriculture? 70% of all global freshwater goes to farming. One pound of beef takes about 1,800 gallons of water to produce (per the Water Footprint Network). A single quarter-pound burger uses more water than hundreds of ChatGPT conversations.

Where It Actually Gets Concerning

I’m not saying we shouldn’t care about AI water use. We absolutely should. But the selective outrage (AI bad, golf courses and beef fine) tells me the real objection usually isn’t about water at all.

The part that actually concerned me? It’s local, not global. A Meta data center in Newton County, Georgia uses 500,000 gallons a day, which is 10% of the entire county’s water supply. Residents’ wells went dry. The county is now projecting a water deficit by 2030. That’s significant, and the people in that community deserve real answers and real planning about how their water resources will be managed.

The Real Water Story

The global water crisis is real and serious. But AI is a footnote in it, not the headline. The primary drivers are agricultural overuse, groundwater depletion, and climate change. The UN’s own 2026 Global Water Bankruptcy report specifically recommended embedding AI into water monitoring frameworks, because the analytical power AI provides for water management far outweighs its consumption.

Read the actual reports. Then form your opinion. That’s all I’m asking.

Paige Processing