In a paradoxical twist to the digital age, America’s newest technology, artificial intelligence, is breathing life into one of its oldest and dirtiest energy sources: coal.
Across the U.S., data centers are driving electricity consumption to levels unseen in decades. As utilities scramble to meet this soaring demand, coal generation has surged roughly 20% this year, according to a report from Jefferies Research. The trend, analysts say, may persist through 2027, delaying the retirement of plants once slated for closure.
Many Forces Supporting Coal
Driving coal’s rise is a combination of factors. Natural gas prices have risen, renewables face regulatory hurdles, and AI workloads consume power in unpredictable bursts that strain the power grid. The result is a grid stretched between the constant draw of AI training clusters and the spiky, high-frequency surges of AI inference.
In response, utilities in several states have reversed plans to shutter coal-fired generators, citing the need to serve fast-growing data center loads. In Omaha, the local power company scrapped its plan to stop burning coal at its North Omaha plant, warning that new data centers could otherwise face shortages.
The shift carries major environmental implications. Coal still accounts for more than half of U.S. power-sector carbon emissions, and every extra ton burned undermines climate targets. Greenpeace calls coal “the dirtiest, most polluting way of producing energy,” a characterization many scientists echo.
A 2024 Morgan Stanley report projects that data centers could emit 2.5 billion tons of greenhouse gases worldwide by 2030—triple what would have been produced without the explosive growth of AI. The U.S. Department of Energy estimates that data centers could consume up to 17% of the nation’s electricity by the end of the decade, a remarkable rise from about 4% today.
Yet Coal is Still Fading
But coal’s apparent new popularity is not so one-sided. Energy companies are increasingly repurposing decommissioned coal sites into hubs for new-generation power.
In Pennsylvania, the shuttered Homer City coal plant, known as a source of industrial pollution, is being converted into a massive AI data center complex, powered by natural gas and set to open in 2027. Developers are reusing the site’s high-capacity grid interconnections, bypassing the years-long queue for new power hookups.
This model is spreading quickly. Xcel Energy, which operates across the Midwest and Rockies, is converting coal units in Minnesota, Texas, and Colorado to natural gas, wind, and solar, all for projects built with data center clients in mind. The U.S. Department of Energy is supporting efforts to repurpose retired coal sites through financing, planning tools, and technical guidance, and has begun exploring how such sites might be converted into data centers.
Federal Policy Moves Away from Renewables
Helping promote coal is the recent shift in federal energy policy, which has tilted away from renewables. The Trump administration has frozen approvals for some wind projects and restricted new solar development, citing land-use and cost concerns.
As Interior Secretary Doug Burgum was recently quoted in The New York Times: “What’s going to save the planet is winning the A.I. arms race. We need power to do that and we need it now,” he said. “We need to worry about the humans that are on the planet today. The real existential threat right now is not one degree of climate change.”
That framing, treating AI capacity as a national security imperative, has intensified the push to secure any available generation, even from fossil fuels.
A Temporary Shift?
Some industry experts insist that coal’s renewed relevance is only temporary. As more nuclear, solar, and battery storage projects come online in the next decade, the balance could shift decisively toward cleaner energy. For now, however, the AI revolution depends partially on an aging fleet of coal-fired turbines, an ironic reminder that the intelligence of tomorrow still runs, at least in part, on the carbon of yesterday.
