Policy

Trump EPA repeals power plant emission limits as AI data centers drive coal resurgence

Federal rollbacks and accelerated approvals could impose up to $21 billion in annual health costs by 2028, according to former EPA officials.

Omega Editorial· September 16, 2026· 3 min read

The artificial intelligence industry's infrastructure boom is colliding with a systematic dismantling of environmental protections, creating what former EPA officials warn could become a public health crisis costing tens of billions of dollars annually.

The Trump administration's Environmental Protection Agency announced Monday it is repealing 2024 regulations that limited greenhouse gas emissions from power plants. The agency's own analysis projects the move will increase coal production for the power sector more than tenfold.

The repeal represents just one piece of a broader regulatory rollback. According to a report published last week by the Environmental Protection Network, a nonprofit composed of former EPA staffers, federal agencies have enacted at least 30 policy changes since January 2025 that risk increasing pollution from data centers. Seventeen of these changes explicitly cite AI and data centers as justification.

The health cost calculation

Researchers from UC Riverside, Caltech, and Rochester Institute of Technology estimated in a 2026 study that data-center-related air pollution could impose between $11.7 billion and $20.9 billion in annual public health costs by 2028. Most of those costs stem from power plants supplying electricity to data centers.

That estimate predates the recent wave of federal policy changes, which the Environmental Protection Network says will likely push emissions and health costs higher through delayed power plant retirements, expanded fossil fuel generation, and increased use of backup generators.

Coal use increased 13 percent in 2025 according to Energy Information Administration data, reversing years of decline. Nearly 10 gigawatts of coal capacity that was scheduled for retirement has been delayed. Coal accounted for roughly 55 percent of total carbon emissions from the U.S. power sector in 2022 and produces sulfur dioxide, nitrogen oxides, and particulates that harm people living near coal plants.

Accelerated approvals, weakened safeguards

Beyond the power plant emission repeal, the administration has rolled back mercury and air toxics standards from 2024, delayed state deadlines for submitting plans to reduce visibility-impairing pollution by three years, and weakened oversight requirements for coal ash and coal-plant wastewater disposal.

A July 2025 executive order directed agencies to speed up federal approvals for certain data centers. The EPA has proposed narrowing its environmental reviews and moved to allow companies to begin some construction before obtaining required major air permits. Another executive order instructed agencies to identify opportunities to promote coal use specifically for AI data centers.

Why it matters

While AI safety researchers debate theoretical risks of rogue models, the environmental and health impacts of powering AI infrastructure are already measurable and growing. The tension between rapid AI development and pollution controls illustrates how competitive pressure in the technology sector can translate directly into public health consequences. For business leaders planning data center investments, the current regulatory environment may enable faster deployment but creates exposure to future liability as health costs mount and policies potentially shift.

"How these tech companies compete in these critical early years may feel like life or death to them," said Lynn Goldman, professor of environmental and occupational health at the Milken Institute School of Public Health, during a press briefing on the report. "But the AI industry's race for advantage should not become a matter of life or death for the rest of us."

These details were first reported by Latitude Media in its AI-Energy Nexus newsletter.

#data centers#environmental regulation#coal power#epa policy#ai infrastructure#public health

This is an original analysis by the Omega editorial team. Source reporting: AI Watch.

Want systems like this working for your business?

Book a Call

More in Policy

Policy· 3 min read

AI data centers could generate 13 million tons of e-waste yearly

New research counting infrastructure beyond servers reveals a vastly larger electronic waste problem than previously estimated.

Via AI Watch · Sep 16, 2026
Policy· 3 min read

U.S. Has No Law Requiring AI Companies to Report Dangerous Behavior

Despite documented cases of AI models attempting deception and system breaches, no federal statute mandates disclosure of risky incidents before concrete harm occurs.

Via AI Watch · Sep 16, 2026
Policy· 3 min read

Pentagon Accelerates AI Deployment Despite Congressional Concerns

Military leaders argue that slowing artificial intelligence adoption poses greater risks than rushing implementation, even as lawmakers seek new safeguards.

Via AI Watch · Sep 16, 2026