Electricity consumption of hyperscale data centres rivals that of some US states

Electricity consumption of hyperscale data centres rivals that of some US states

Hyperscale data centres in the United States consume between 68 and 99 TWh of electricity over a 12-month period, approaching the amount that some US states, such as Indiana, Michigan, and Tennessee, use in a year, a new Harvard study has found.

The study, Assessing the Carbon Emissions and Energy Consumption of US Hyperscale Data Centers, was led by Harvard T.H. Chan School of Public Health and the Harvard Data Science Initiative and published in the Nature Sustainability journal.

Electricity consumption

It found that hyperscale data centres in the US account for approximately 1.8% of total national electricity consumption, while said centres were also associated with approximately 25.7 million metric tonnes of CO2 between May 2024 and April 2025.

This is around two to three times higher than a previous estimate, taken in 2018.

“What we’re seeing is already striking, but it’s only a snapshot of a rapidly expanding industry,” commented lead author Gianluca Guidi, postdoctoral fellow in the Department of Biostatistics. “The data centre boom hasn’t slowed down since our study period ended in April 2025. New facilities have been built, existing campuses have expanded, and demand for AI computing continues to grow.

“That means the electricity use and carbon footprint we’re reporting could already be substantially larger now in 2026.”

Energy mix

In conducting the study, the researchers used satellite imagery to identify 403 hyperscale data centres across the US, estimating not just how much electricity each facility uses, but what fuels generated this electricity. They then used Environmental Protection Agency data to evaluate the carbon emissions associate with each data centre, based on the energy mix in its regional grid.

More than half (54%) of the electricity used by the data centres studied came from fossil fuels, with natural gas accounting for 38%, and coal for 15%. In addition, around a quarter of the electricity was sourced from grid regions with higher-than-average emissions.

The study also noted high concentrations of electricity consumption in some areas, with just four states, Virginia, Oregon, Ohio, and Iowa, accounting for more than half of the total electricity use.

The researchers also pointed out a number of limitations to the study, noting that there wasn’t sufficient data to study all hyperscale data centres in the US. In addition, the data did not allow them to distinguish between electricity used for general-purpose computing and AI-specific operations.

‘Up-to-date picture’

“Our goal is to give the public, policymakers, investors, and developers a much more concrete, up-to-date picture than was available before,” added corresponding author Francesca Dominici, Clarence James Gamble Professor of Biostatistics, Population, and Data Science and director of the Harvard Data Science Initiative.

“Decisions about where data centres go and how they’re powered are critical environmental and public health decisions, and they should be made with good data.” Read more here.