Utilities are grappling with how much AI data center power demand is real
While the stock market predicts large amounts of money will be spent on infrastructure to support the construction of a massive data center, electric utilities across the US are struggling to calculate how much demand will actually be realized from the AI boom.
“There’s a question of whether all the predictions are real,” Willie Phillips, who chairs the Federal Energy Regulatory Commission from 2023 to April 2025, told CNBC. he said. “There are some districts that predicted big increases, and they recalibrated those.”
AI companies are rolling out ambitious plans to build server farms that, in some cases, will consume as much electricity as entire cities. But the tech industry is selling the same large projects to multiple utilities as it searches for the fastest access to power.
“We’re starting to see demand for similar projects with exactly the same footprint in different parts of the country,” GridUnity CEO Brian Fitzsimons told CNBC. GridUnity uses software to give utilities and transmission operators a clearer picture of where energy projects demand connections across the U.S. power grid.
This data center exchange makes it difficult for utilities to determine how much energy generation they will need to ensure the reliability of the electric grid. Meanwhile, electricity prices for consumers are also rising because electricity supply is already struggling to keep up with demand.
FERC Chairman David Rosner Warned in September He said a difference of a few percentage points in electricity load forecasts “can have an impact of billions of dollars in investments and customer bills.”
“Simply put, we cannot efficiently plan the generation and transmission of electricity needed to serve new customers if we do not predict as accurately as possible how much energy they will need,” Rosner said. he said.
Constellation Energy CEO Joe Dominguez warned about the problem on the nuclear power operator’s May earnings call: “I have to tell you guys, I think the load is overrated. We need to pump the brakes here.”
AI bubble fears
But the stock market isn’t actually putting on the brakes. The steady pace of big data center announcements is fueling one of the biggest rallies in energy company stocks in two decades.
The utilities sector has gained nearly 21% this year, following a more than 19% rise in 2024. Companies providing electricity to the USA gained approximately 500 billion dollars in value in this two-year period. The last time utilities increased by more than 40% in a row was 2003 and 2004.
OpenAI CEO Sam Altman warned in August that the stock market was facing an AI bubble and warned investors that they were “overexcited.”
Although the exact magnitude of future demand is unclear, experts generally agree that the United States is facing a historic increase in electricity consumption after a long period of steady growth. Rob Gramlich, president of Grid Strategies, said current data centers point to the future.
“We can see it. Data centers are available.” he said. “They operate using a lot of electricity every day. A 50-megawatt data center used to be pretty big. Now it’s very common to have data centers that are 20 times that size, which is one gigawatt.”
Grid Strategies, an energy industry consulting firm, estimates 120 gigawatts of additional electricity demand by 2030. This includes 60 gigawatts from data centers, according to estimates from utilities. To put that into perspective, 60 gigawatts is roughly equivalent to 2024. Italy’s highest hourly power demandThe eighth largest economy in the world.
“This is not a bubble,” Fitzsimons said. “It will completely transform our country. It will continue to grow. We need a 50-year energy policy.”
But utilities need to get solid financial commitments from data centers, Gramlich said.
“This will help us rationalize all these demands and better address the overall forecast,” he said. “But the industry needs to plan based on the best information we have right now.”
Uncertainty about demand forecasts has raised concerns that utilities could spend billions of dollars on infrastructure that ultimately isn’t needed. Utilities spent $178 billion on grid improvements last year and are projecting $1.1 trillion in capital investments by 2029. Edison Electrical Institute.
Fitzsimons said the risk of utilities being overbuilt was lower than it was two decades ago because of market constraints.
“They are in a very different environment where we have major supply chain issues,” Fitzsimons said. “We’ve been trying to beat inflation for a long time. They can’t afford to overgrow. It depends on better planning.”
Infrastructure restrictions
Altman one month after bubble warning reached an agreement with Nvidia That OpenAI will build 10-gigawatt data centers with the chipmaker’s graphics processing units. The plan requires as much electricity as New York City on energy-intensive summer days, raising questions about whether they can actually provide the electricity needed.
Gramlich said it might be possible to secure that much electricity, but the AI industry faces limitations as its plans grow. He said companies are competing for scarce infrastructure, which is driving up prices of basic electrical equipment such as transformers, switches and breakers.
“We don’t have the electrical infrastructure to meet truly aggressive targets,” the analyst said. “We do not have sufficient generation or transmission infrastructure to meet even modest midpoint targets.”
The question, he said, is how quickly the next generation can be built. Natural gas turbines became largely extinct towards the end of this decade. The tech industry is investing in advanced nuclear energy, but these technologies are not expected to reach commercial scale until the 2030s at the earliest.
Meanwhile, renewable energy can be deployed fastest, especially solar power and battery storage. According to August data from energy consultancy firm Enverus, more than 90% of energy projects currently awaiting grid connection consist of solar, battery storage or wind.
“Over the last 10 years our interconnection leads have been largely filled with renewable energy,” Fitzsimons said. “Renewable energy is the fastest way to create new capacity. There’s no doubt about that because of the supply chain issues with natural gas turbines.”
But President Donald Trump has promoted coal, natural gas and nuclear power while opposing solar and wind, raising uncertainty about whether enough new ones can be built. Gramlich said utilities will shut down data centers if they don’t have enough power.
“If they don’t have the power to truly serve a customer, they won’t compromise on reliability,” Gramlich said. “That’s their main job.”
Some AI companies believe they have found a solution — generating power in an off-grid data center, referred to in industry parlance as “behind the meter.”
“We should be investing in almost every way we can produce energy,” Nvidia CEO Jensen Huang said in an Oct. 8 interview with CNBC. “The power that the data center generates itself can move much faster than putting it on the grid, and we should do that,” he said.
