AI data center demand drives U.S. power plant costs up 10%

Utility capital plans hit $1.4 trillion on AI demand

The accelerating cost of building new power plants is reshaping the economics of the U.S. electric grid as demand growth from AI data centers — each of which can consume as much electricity as entire cities — collides with equipment backlogs, permitting delays and tariff-driven material costs.

Demand growth is “dominating everything in energy,” said George Bilicic, Lazard’s vice chairman of investment banking and global head of power, energy and infrastructure. “If you’re an [exploration and production] company, you’re thinking about it, a pipeline company, you’re thinking about it, utilities, power companies, certain infrastructure-services companies. It is really pushing costs up.”

Costs for large-scale solar projects increased to an average of $69 per megawatt hour, up from $58 last year, according to Lazard. Adding batteries to a solar project pushed costs to about $109 per megawatt hour, up from $91. Combined-cycle natural-gas plants, a grid workhorse, saw construction costs climb to roughly $90 per megawatt hour, the highest level in about 15 years, from $79 last year.

Capital-spending plans for 51 investor-owned utilities have reached an estimated $1.4 trillion for the next five years, up more than 20% from last year’s projections, according to PowerLines, a consumer-education group.

Much of the near-term grid additions will come from renewable energy and battery storage, which make up about 90% of projects under construction likely to start generating electricity this year, according to the Energy Information Administration.

Scott Strazik, chief executive of GE Vernova, a major gas-turbine supplier, said Wednesday that many customers are ordering equipment for delivery into the 2030s and that the company is boosting output. By the end of the year, Strazik expects to already have half of the company’s 2031 turbine output under contract. “This is just a larger market for a longer period of time,” he said.

The higher construction costs will eventually work their way into monthly utility bills as companies recover those investments over time. Electricity prices rose 4% in June from a year earlier, according to the Bureau of Labor Statistics, outpacing overall inflation.

“You’re going to see costs continue to rise. I don’t see how they can’t just as a basic supply-and-demand matter,” Bilicic said.

The cost pressures are particularly acute in the PJM Interconnection market, which serves roughly 67 million people across the mid-Atlantic and Midwest. Power prices in the region have climbed as data-center growth, concentrated in Virginia, collides with the need to add generation. PJM’s latest capacity auction fell short of the reserves it hoped to procure, and the operator plans another auction for additional capacity this fall.

Monitoring Analytics, PJM’s independent market monitor, is calling for new data-center load to be removed from the capacity market and procured separately. Of about $16.4 billion in capacity-market charges, $6.3 billion was caused by data-center demand, said Joe Bowring, president of Monitoring Analytics.

Because the auction hit a price cap, capacity prices still are not high enough to spur developers to build the additional power plants, especially natural gas, that the market needs, said Peter Gardett, chief executive of market-data platform Noreva. “The auction has pulled off making everyone unhappy,” Gardett said.

There are more than 190 cases across the country of utilities wrestling with contract provisions, special rates or ways to charge large data-center customers, according to utility filings aggregated by AI startup Halcyon.

The White House has persuaded tech companies and a group of electric utilities to pledge to avoid shifting the cost of new grid equipment needed by data centers onto regular households and businesses. It is not clear how voluntary pledges will work in practice, the report noted. Utility rates are typically overseen by state regulators, and tracking costs specifically caused by data centers will be difficult, as some system upgrades benefit all users and are traditionally applied to all customer bills.