As the tech industry has grown, so too have data centers.
Data centers are enormous buildings filled with hundreds of thousands of computers that store cloud data and power artificial intelligence. To keep up with computing demands, data centers use electricity and sometimes chilled water to keep those computers cool.
The result? A surge in energy and water use that has caught the attention of scientists and lawmakers.
Under the Biden Administration, Congress commissioned a report on data center electricity consumption. Led by Lawrence Berkeley National Laboratory, the 2024 United States Data Center Energy Usage Report forecast that by 2028, U.S. data centers could consume as much as 12% of the nation's electricity.
But the environmental footprint of AI is notoriously difficult to measure.
There are no federal nor state regulations for AI and no legal framework requiring tech companies to disclose their energy and water consumption. That's led scientists like Shaolei Ren to investigate this question independently.
Today, Ren is an associate professor of electrical and computer engineering at the University of California, Riverside. But Ren grew up in a coal mining town in northern China, where water was scarce. He learned how to make every drop count.
"We only had water access for like half an hour each day. So we just had to use water very wisely," says Ren.
Ren was among the first to study the water footprint of AI. In a 2023 pre-print of a paper, Ren's team estimated that to train the GPT-3 language model consumed hundreds of thousands of liters of fresh water. That water is evaporated and does not necessarily return to the local watershed.
Meanwhile, the biggest U.S. data center operators have pledged a green future.
Google, Microsoft and Meta have all pledged to reach at least net-zero carbon emissions by 2030. Amazon has set their net-zero deadline for 2040. All four companies have also pledged to be water positive by 2030, meaning they'd put more water back into the environment than they use. All four companies are financials supporters of NPR, and Amazon also pays to distribute some of NPR's content.
Scientists wonder how the tech industry plans to square their climate pledges alongside the increasing demand for AI computing power.
"I think before generative A.I. came along in late 2022, there was hope among these data center operators that they could go to net zero. I don't see how you can, under current infrastructure investment plans, you could possibly achieve those net zero goals," says Benjamin Lee, a professor of electrical and systems engineering at the University of Pennsylvania.
Data center construction is expected to increase.
The day after his second inauguration, President Trump announced a private joint venture to build 20 large data centers across the country. Known as Stargate, the data centers would consume 15 gigawatts of power.
Some legislators have introduced bills to regulate AI. Senator Edward Markey of Massachusetts has introduced a bipartisan bill that would set federal standards and voluntary reporting guidelines to measure the environmental footprint of AI. State lawmakers in California and Connecticut have introduced their own bills.
Meanwhile, tech companies are trying to create and train more sustainable AI models and build cleaner data centers.
Curious about tech and the environment? Email us at shortwave@npr.org — we'd love to hear from you!
Listen to Short Wave on Spotify, Apple Podcasts and Google Podcasts.
Listen to every episode of Short Wave sponsor-free and support our work at NPR by signing up for Short Wave+ at plus.npr.org/shortwave.
Today's episode was produced by Hannah Chinn and edited by Rebecca Ramirez. Tyler Jones checked the facts. Jimmy Keeley was the audio engineer. Special thanks to Brent Baughman, Johannes Doerge, at the NPR Standards team.
Copyright 2025 NPR
