Fossil fuels are still the biggest source of electricity for hyperscale data centers in the US, a new paper published in the journal Nature Sustainability shows. They likely burn through more electricity collectively than many countries do, resulting in a serious amount of greenhouse gas emissions.
The study homes in on hyperscale facilities — the massive structures spanning millions of square feet that developers are racing to build in large part to support generative AI. Researchers say the findings show that there’s a lot of work left to do to cut down pollution stemming from AI’s reliance on fossil fuels.
“We tend to discuss AI’s environmental footprint as though it were some fixed property of the technology, when in reality the same computing infrastructure can have dramatically different consequences depending on the electricity system supporting it,” Gianluca Guidi, lead author of the new study and a postdoctoral fellow in the Department of Biostatistics at Harvard T.H. Chan School of Public Health, says in an email to The Verge. “I actually find that encouraging, because it means these impacts aren’t inevitable. We have choices about where we build and how we power this infrastructure.”
“The same computing infrastructure can have dramatically different consequences depending on the electricity system supporting it.”
Guidi and his team used satellite imagery to identify 403 hyperscale data centers in the US, estimating each facility’s power consumption based on its electrical capacity. Using data from the Environmental Protection Agency, they approximated how much carbon pollution each data center would have produced if connected to the local power grid between May 2024 and April 2025.
Altogether, the 403 hyperscale facilities likely consumed around 82 terawatt-hours of electricity that year, about 1.8 percent of all US electricity consumption. That’s even more than the amount of electricity that about 148 nations — a majority of the countries in the world including Austria, Israel, and Venezuela but excluding the largest economies including the US and China — individually burn through each year, according to John Sterman, director of the MIT System Dynamics group and MIT Sloan Sustainability Initiative.
Electricity demand from these data centers produced 25.7 million metric tons of CO2 emissions, comparable to the annual carbon footprint of Denmark, according to the research paper. Sterman says that exceeds the total carbon emissions of about 113 nations. That’s because about 54 percent of the electricity the hyperscale facilities consumed likely came from gas and coal.
Sterman, who was not involved in the research with Guidi, disclosed that he used the help of Claude to compare the paper’s findings to electricity use and emissions of countries. “I think AI can be extraordinarily helpful,” Sterman says. “But we’ve got to figure out a way to limit the harms … Especially because those impacts are irreversible. Once you allow these data centers to emit the CO2, you can never undo that.”
Sterman says policymakers should require data center operators to bring new clean energy online. The study doesn’t account for local air pollution that can result from new gas infrastructure that’s being built to meet data center demand, or gas turbines that some AI companies have installed at their facilities to power themselves without connecting to the grid. And while hyperscale data centers encompass more than just AI-related computing, they also make up just a fraction of the thousands of data centers in the US — with more cropping up each year.
The study showed how big a difference the local electricity mix makes when it comes to the environmental impact of AI and data centers. Generating electricity in heavily coal-reliant Wyoming led to roughly seven times more carbon emissions per kilowatt-hour than in hydropower-rich Oregon. “Two data centers consuming exactly the same amount of electricity can have dramatically different carbon footprints simply because of where they’re located,” Guidi writes.
Fengqi You, a professor in energy systems engineering at Cornell University who was also not involved in the study, says the results highlight the problem of “long-term carbon lock-in.” “Power plants and supporting infrastructure built to meet rapidly growing data center demand may operate for decades. Decisions made today could therefore shape emissions well into the future,” You says in an email to The Verge.
Both You and Guidi also pointed out how problematic it is that researchers have to go through such great lengths to estimate energy consumption because data center operators won’t disclose that information. To that end, Guidi and his colleagues are using their research to build a website to track the environmental impact of individual data centers. A bipartisan bill making its way through Congress could also mandate greater disclosures nationally on data centers’ electricity use.
“Communities and policymakers need reliable, independent data on what these facilities mean for our air, health, and the climate. This study is an important step.” Gina McCarthy, former White House national climate advisor and EPA administrator, said in a statement shared by the research authors.
Follow topics and authors from this story to see more like this in your personalized homepage feed and to receive email updates.
- Justine Calma
Source: The Verge · Justine Calma
At a glance
- Topic
- Climate
- Country
- 🇺🇸 US
- Publisher language
- EN
- Added to Allnira