Sierra Club Analysis Says AI Data Center Power Buildout Could Set Back U.S. Emissions Cuts

The Sierra Club, a U.S. environmental group, released an internal analysis warning that the boom in fossil fuel power plants built to supply AI data centers could offset 15 years of greenhouse gas reductions in the U.S. power sector.

The group estimates that, including planned AI projects, U.S. power sector carbon dioxide emissions could reach 2.1 billion metric tons by 2030, up 618 million metric tons from 2025, an increase of about 42%. Jeremy Fisher, senior advisor for climate and energy at the Sierra Club, said that if such a projection had been made a year and a half ago, the estimated scale and impact might have been only one-quarter to one-half of current forecasts.

Much of the new generation is expected to come from plants not directly connected to the public grid, as tech companies build their own power sources to avoid long interconnection queues. Fisher said these emissions could exceed 460 million metric tons, roughly equivalent to Mexico's annual emissions.

Helen Kou, an analyst at BloombergNEF, said emissions are likely to be higher than this year's levels if new facilities rely heavily on natural gas. The Sierra Club's analysis combines public filings and other forecasts of data center construction, including a tally of planned power plants and emissions through 2030 compiled by an environmental group's project.

New generating units connected only to the public grid would add 154 million metric tons of CO2 by 2030, a 10% increase above the baseline scenario, with additional emissions from on-site plants. Another analysis by Earthjustice and the Better Data Center Project suggests that up to 80% of the energy from such facilities could come from natural gas.