Ren, Shaolei, and Amy Luers. “The Real Story on Ai’s Water Use–and How to Tackle It.” IEEE Spectrum, 22 Sept. 2025. spectrum.ieee.org/ai-water-usage.
This article was extremely beneficial in explaining the water usage part of data centers. AI data centers run incredibly hot, and keeping them cool takes SO MUCH water! It is way more than most people realize. A lot of facilities use evaporative cooling systems that pull from the same water supplies as local homes and businesses, and U.S. data centers consumed an estimated 17.5 billion gallons of water in 2023 alone, with that number potentially doubling or quadrupling by 2028. The electricity powering these data centers comes mostly from fossil fuel plants that need cooling water too, and that indirect water use makes up 80 percent or more of the total footprint. This article talks about the significance of even asking Chat GPT a single question. One study found that a single short text response consumed about 16.9 milliliters of water. The article does point to some solutions like better cooling technology and switching to renewables, but the authors are honest that there’s no simple fix. Reducing water on one end often just increases energy use on the other.
Honestly, I picked this one because there were so many different views in the articles we’ve read for our discussions. We’ve heard plenty about AI using a lot of energy and water, but it felt like there were some articles saying that we are not affecting the environment significantly and some that were saying yes, of course, we are killing the environment. The fact that about two-thirds of data centers built since 2022 are sitting in high water-stress areas was something I noted in the article. These aren’t just abstract environmental stats, they’re affecting real communities competing for the same water. I thought it was worth bringing to class because it is what we have been discussing of late and it ties in perfectly to my current research on how AI affects the environment.
Leave a Reply