Where is the U.S. going to get the power needed to meet surging demand? It’s a question that has become a day-one issue for the Trump administration and the AI economic engine.

Just how much power are we talking and how fast? A recent forecast sees U.S. electricity demand rising 128 gigawatts (GW) over just the next five years—that’s like adding 80 million homes. And that forecast may already be too low. Just this week, PJM, the largest U.S. grid serving 65 million Americans, revised upward by 38% its electricity demand forecast for this coming decade. PJM now expects peak summer power demand to surge to 210 GW in 2035, a jump of 58 GW in a decade. Just a year earlier PJM saw peak summer demand reaching 177 GW in 2034. Load growth from planned data centers is changing the game.

While tech companies are pointing to the restart of a few mothballed nuclear power plants and the potential of small modular reactors to meet their massive, baseload power needs, those headlines don’t match reality. The power demand now on the horizon is coming at a speed and scale that requires generation immediately.

Some observers predict a natural gas generation boom, but utility executives are tempering expectations. NextEra CEO John Ketchum recently warned that new gas generation wouldn’t be available in large amounts until 2030, and gas plants will only be a viable option in certain parts of the country and at a much higher price point than just a few years ago.

The intermittency of renewable energy also makes it a poor fit for the around-the-clock power needs of data centers, and that’s not to overlook the towering hurdles now facing industrial scale renewable projects and their enabling infrastructure.

The options to find power – much less reliable, immediately available generation capacity – are few and far between. The answer is the underutilized coal fleet.

Ramp up Coal Utilization

Due to prior years of flat power demand and competition from other heavily subsidized power sources, the coal fleet’s annual capacity factor – a measure of how often it’s providing power to the grid – is now just 42%.

This is not a measure of how of the fleet can perform. In critical, high-demand weeks and months – or when natural gas prices spike – the fleet regularly ramps up generation, often reaching capacity factors above 60%. As recently as 2021, there were several months when the fleet had a capacity factor above 65%. Today’s reduced capacity factor is rather a reflection of a power system that has invested countless billions into weather dependent, variable sources that have eroded the economics of the very dispatchable capacity that remains irreplaceable to keeping the lights on when consumers need reliable power most.

Quite simply, the coal fleet could be running far more often. In fact, it was designed to. Increased utilization rates for the fleet would mean far more efficient use of the plants and decreased cost per gigawatt hour of production for consumers.

“Nothing Can Destroy Coal”

As new industrial projects, whether massive data centers or major new manufacturing hubs, scramble for power, the baseload capacity to meet their needs is hiding in plain sight. Utilities have already begun postponing or altogether cancelling coal plant retirements to maintain reliability and meet soaring demand. While there is no question the nation needs more generating capacity, the bridge to get us to our energy future is in place—we now need to choose to take it. President Trump is fully on board.

Announcing a plan to meet AI-driven power demand, the President said the tech industry can, “fuel it with anything they want, and they may have coal as a backup — good, clean coal.” He added, referencing coal’s fuel security, “nothing can destroy coal,” and “we have more coal than anybody.” No argument here.