Lifestyle

Nature Sustainability Study: 403 US Hyperscale Data Centers Use About 82 TWh a Year, 1.8% of US Electricity

According to The Verge, a Harvard team published a study in Nature Sustainability estimating that US hyperscale data centers use about 82 TWh a year, with roughly 54% of that power possibly from natural gas and coal. The figures are estimates based on satellite imagery and grid data.

News date: About 5 min read

Share on LINE
Contents
  1. 1. What did the study find?
  2. 2. How does this scale compare with countries?
  3. 3. Why do location and power source matter so much?
  4. 4. What happens next on transparency and policy?

On November 16, 2026, The Verge reported that researchers at the Harvard T.H. Chan School of Public Health had published a new paper in the journal Nature Sustainability estimating the electricity use and carbon emissions of US hyperscale data centers. This is a published academic study, not a new product or service. For general readers, it shows that the environmental cost of AI computing is closely tied to power sources and facility location, and it may influence future energy and data-disclosure policy.

What did the study find?

According to The Verge, the study focuses on hyperscale facilities: enormous buildings covering millions of square feet that developers are racing to build to support demand such as generative AI. The paper's first author, Gianluca Guidi, and his team used satellite imagery to identify 403 US hyperscale data centers and estimated each one's electricity use from its power capacity.

They then used US Environmental Protection Agency (EPA) data to calculate how much carbon pollution each data center would produce if connected to its local grid between May 2024 and April 2025. The results suggest the 403 facilities may have used about 82 terawatt-hours that year, roughly 1.8% of US electricity consumption. (A terawatt-hour, or TWh, is a large unit of energy: 1 TWh equals 1 billion kilowatt-hours, or kWh.)

Key figures from the study (source: The Verge's report on the paper)
MetricFigure reported
Facilities included403 US hyperscale data centers
Estimation periodMay 2024 to April 2025
Estimated electricity useAbout 82 TWh
Share of US electricity useAbout 1.8%
CO2 emissions25.7 million metric tons (per the paper; comparable to Denmark's annual carbon footprint)
Share of power from natural gas and coalAbout 54%
Regional differenceWyoming's emissions per kWh are about seven times Oregon's

How does this scale compare with countries?

John Sterman, who leads the MIT System Dynamics group and directs the MIT Sloan Sustainability Initiative, said these facilities use more electricity than about 148 countries each use in a year, including Austria, Israel and Venezuela, but not the largest economies such as the US and China. He also said their total emissions exceed those of about 113 countries. Sterman was not involved in the study, so these comparisons are his statements rather than conclusions of the paper itself.

Why do location and power source matter so much?

The report says the study shows the local power mix has a huge effect: generating power in coal-heavy Wyoming emits about seven times as much carbon per kilowatt-hour as in hydropower-rich Oregon. Guidi said in an email that two data centers using exactly the same amount of electricity can have very different carbon footprints depending on location, and that these impacts are not inevitable, because there are choices about where to build and how to power them.

Fengqi You, a professor of energy systems engineering at Cornell University who was not involved in the study, pointed to the problem of "long-term carbon lock-in": power plants and supporting infrastructure built to meet data center demand may operate for decades, so today's decisions will shape emissions for many years. Sterman argued that policymakers should require data center operators to bring new clean energy online.

What happens next on transparency and policy?

Guidi and You both noted that researchers had to work hard to estimate electricity use because data center operators do not disclose it. Guidi and colleagues are using the study's results to build a website that tracks the environmental impact of individual data centers. The report also mentions a bipartisan bill advancing in Congress that could require greater nationwide disclosure of data center electricity use. The report does not describe the bill's timeline or final content.

Frequently asked questions

Where does the study's data come from?

According to The Verge, the team identified 403 US hyperscale data centers from satellite imagery, estimated electricity use from power capacity, and then used EPA data to calculate emissions if each were connected to its local grid, for May 2024 to April 2025.

Is all 82 TWh used by AI?

No. The report says hyperscale data centers run more than AI workloads and are only a subset of the thousands of data centers in the US. The 82 TWh is the estimated total electricity use of these 403 facilities.

Why can the same electricity use produce different emissions?

Because emissions depend on the grid's power mix. In the report, Wyoming (heavily coal-dependent) emits about seven times as much per kilowatt-hour as Oregon (rich in hydropower).

Are the country comparisons in the report conclusions of the paper?

No. The country comparisons (about 148 countries' electricity use, about 113 countries' emissions) come from John Sterman, who was not involved in the study. The paper itself says the emissions are comparable to Denmark's annual carbon footprint.

Browse the latest news in this topic

Found this guide helpful? Share it with someone who needs it

Share on LINE

Sources

Compiled from the references and verification dates listed above; fares, schedules, product features, and terms can change, so please confirm the latest details on the relevant official website before travelling, booking, or purchasing.

Lifestyle