Coal Use Rises as AI Data Center Demand Strains U.S. Power Grid

Coal Makes a Comeback as AI Reshapes the Grid

A new federal report is putting hard numbers behind a trend utilities and regulators have been warning about for months: the explosive growth of artificial intelligence is straining the U.S. electricity system enough to revive coal power. According to the Energy Information Administration, coal-fired generation jumped 13% last year, helping drive a roughly 4% increase in power-sector carbon emissions — the first such rise after years of steady decline 1. That reversal underscores how quickly surging data center electricity demand is reordering the country's energy priorities, pushing grid operators back toward the very fuel sources the sector had been phasing out.

Demand Growth Outpacing Infrastructure

The strain is not evenly distributed, but it is widespread. In Texas, one of the country's most-watched power markets, grid operators are grappling with rising demand tied directly to new AI data center construction, reigniting debate over how the state should balance growth with reliability 6. New Jersey lawmakers are asking similar questions, examining whether the state's grid — and ratepayers — can absorb the costs of hosting large-scale AI facilities 4. At the municipal level, the town of Gilbert, Arizona has moved to regulate data centers, joining other Valley communities responding to AI infrastructure expansion that is arriving faster than local planning processes can accommodate 9.

The Investment Boom Behind the Strain

The scale of the buildout helps explain the pressure. Hyperscaler capital spending on data centers is climbing at nearly double the pace of the housing boom's peak, with some estimates suggesting such capex could reach 3% of U.S. GDP by the 2027–2029 window — a trajectory that could boost economic growth but also carries the risk of a sharp pullback if AI demand cools 5. That investment wave is also reshaping corporate technology strategy more broadly, as rising energy costs and AI workloads push businesses to abandon in-house server infrastructure in favor of outsourced, larger-scale facilities better equipped to handle power-hungry workloads 8.

Grid Reliability Problems Emerge

The demand surge isn't just a planning headache — it's beginning to cause operational failures. Reporting has found that the volatile, spiky power draw from AI training runs is damaging the very data centers it powers, with equipment failures delaying projects and prompting at least one rare grid-level regulatory alert 3. That volatility highlights a mismatch between how AI workloads consume electricity and how traditional grid infrastructure was designed to deliver it.

Storage and Batteries as a Possible Fix

Industry and startups are racing to close that gap. The energy storage sector is expanding rapidly as a response to AI-driven demand, with companies working to supply the buffering capacity grids need to handle unpredictable loads 2. Among the more notable efforts, European startup Ore Energy is developing long-duration battery technology aimed at delivering reliable, renewable baseload power — a potential answer to the reliability problems now surfacing across AI-heavy grids 7. Whether such technologies can scale quickly enough to offset the return to coal remains an open question, and one likely to shape both energy policy and AI infrastructure siting decisions in the years ahead.