Surging AI Power Demand is Here, Like It or Not

Artificial intelligence promises economy-wide transformation and a pace of innovation we may have never seen before. But the impact of AI on the electricity grid and the public’s concern over data center development and rising power prices are rapidly becoming significant barriers to the AI revolution.

According to the International Energy Agency, global electricity consumption from data centers is expected to more than double by 2030. Here in the United States, utilities and grid operators are forecasting the strongest period of electricity demand growth in decades. The nation’s grid watchdog now expects summer peak demand and winter peak demand to jump by over 225 gigawatts (GW) and 245 GW, respectively, over the next decade—enough electricity to power 150 million homes.

Meeting data center power demand and cushioning the impact on consumers requires more reliable and affordable power. And the need is no longer in the distance—it’s here now.

Reliability Concerns Are Delaying Coal Retirements

The challenge posed by surging demand and ballooning electricity prices – up 40% since 2020 – is already reshaping the conversation around coal.

As electricity demand accelerates and reliability concerns mount, utilities and grid operators are increasingly reconsidering coal plant closures. In fact, the amount of coal capacity closed in 2025 fell to a 15-year low according to the U.S. Energy Information Administration.

Across the country, coal plant retirements have been delayed, suspended or reversed because the math of meeting power demand growth simply doesn’t pencil out without existing coal generation.
Markets are sending strong signals. Capacity prices have surged across the country—a clear indication of capacity shortages and an unmistakable message to plant owners that closing more coal capacity would be a grave mistake.

Taken together, electricity markets, mounting demand forecasts and reliability assessments are pointing toward the same conclusion: the nation needs more coal generation, not less.

Washington Is Recognizing Coal’s Strategic Value

Federal policymakers have taken notice. The Trump administration has made strengthening America’s energy security and grid reliability a key priority. Through emergency reliability orders, support for existing coal generation and actions under the Defense Production Act (DPA), the administration has embraced the coal fleet as an essential tool for meeting the AI moment.

With power demand growing so quickly, and at such a vast scale, there are few options to meet it. Trying to match the speed of demand growth with new dispatchable capacity and enabling infrastructure, notably the development of interstate pipelines and transmission lines, is proving all but impossible.

Getting even more power out of existing coal plants is essential. Coal plant capacity factors rose from an average of 42% in 2024 to 50% in 2025. And the Trump administration knows these underutilized plants are capable of providing even more power to the grid when and where it’s needed most.

In February, the Department of Energy (DOE) announced significant investments in six Appalachian coal plants to modernize, retrofit and extend their lives. DOE’s 2027 budget request also targets repurposing $3.5 billion of Infrastructure Investment and Jobs Act funds to support reliable and affordable baseload power generation. According to the request, that includes coal power plant upgrades.

The administration invoked the DPA in April to secure coal supply chains and support affordable and reliable baseload power generation. That decision was a clear recognition of the stakes of this moment; a recognition of the reliability challenge the nation faces, of the extraordinary power demand on our doorstep and the critical role coal capacity can and should play in meeting it.

Americans are understandably wary of the surge in data center development and the impact on their energy prices. But rising power demand need not mean a power supply crunch if we use the tools already within reach.