The Trump administration and utility industry are sending a clear message about the politically volatile rise in electricity prices: don’t blame data centers.
“No one’s electric bill has gone up because of a data center,” Interior Secretary Doug Burgum told Fox News last month.
“Demand growth is actually the way to drive down the price,” Energy Secretary Chris Wright said in January.
The reality, however, is more complicated. The direction of power prices rests on whether regulators and utilities can coordinate to insulate Americans from spiraling infrastructure costs.
Federal energy data shows that utility rates didn’t rise substantially in states where demand grew the fastest over the past five years — a point Wright and Burgum return to as political tensions around inflation focus in part on how fast energy costs could go up in the coming years.
But analysts told POLITICO’s E&E News that the direction electricity costs moved in the past may say nothing about the future. The Trump administration has fostered an artificial intelligence boom, and that’s led to projections of record electricity demand through the end of the decade.
“The dynamics that have contributed to the crisis in electricity rates and that tension today are not driven by data centers, but going forward data centers are absolutely going to amplify it,” said Shanthi Muthiah, who leads the energy advisory practice at the consulting firm ICF.
“The reality is there’s going to have to be investments in the grid to accommodate increased load, so the question is whether those increased costs translate into higher rates,” Muthiah added.
Trump has staked political capital on the debate, bringing tech executives to the White House in early March for a splashy signing of the “Ratepayer Protection Pledge,” although experts say it’s unclear if the plan will work. The companies said they would not foist the cost of their infrastructure on Americans, and Trump even said the pledge would help bring down utility costs that have become a sore point for the administration.
“Unfortunately,” Trump noted, “it will take a little time to get there.”
Here are three questions answered about data centers and electricity prices:
How much have data centers raised electricity prices so far?
It’s a mixed bag. Data centers can contribute to both lower and higher prices, but they have not been the only factor in many locations.
States that saw the most demand growth between 2019 and 2024 — including from data centers, but also new manufacturing, industry or population increases — saw residential electricity prices go down, according to an analysis of data from the U.S. Energy Information Administration. States like North Dakota and New Mexico saw double digit increases in load, while rates went down more than other states.
States with little change in electricity demand, like Hawaii, sometimes saw prices rise.
The reason is that spending to fix or improve transmission and distribution lines has been a chief driver of rising electricity prices. When the costs of fixing the grid are spread out among more customers because of higher demand, it can help push prices down.
“Electricity prices have been going up the last five years, even before data centers really entered the scene, largely due to an aging grid that costs a lot of money to replace the repair and expand, and other factors like extreme weather events,” said Charles Hua, executive director of the nonprofit PowerLines, which advocates for lowering consumer bills.
Even Virginia, the world’s data center capital, saw prices stay relatively flat in that time despite surging demand from AI. A 2024 report by state legislative analysts concluded that current electricity rates “appropriately allocate costs to the customers responsible for incurring them, including data centers.”
However, the report cautioned that increased energy demand “will likely increase system costs for all customers, including non-data center customers,” because of the new infrastructure required and the need for utilities to import power at potentially higher prices.
The Electric Power Research Institute estimates on the high end that data centers could consume as much as 17 percent of the country’s generation by 2030, which is forcing utilities to build new generation, transmission lines and distribution equipment. A recent report by the Union of Concerned Scientists estimated that the total cost of electricity linked only to data centers could range from $886 billion to $978 billion by 2050.
“Even if a small fraction of that cost gets passed on, that’s a pretty significant increase for customers,” said Steve Clemmer, director of energy research for UCS and author of the report.
What do state examples tell us?
There’s not a cookie-cutter formula for when data centers raise or lower prices, but one factor is whether the grid in a given region has spare capacity.
That was the case with North Dakota, which also benefited from an abundant electricity supply of gas, coal and growing low-price wind power, according to Lawrence Berkeley National Laboratory. As large loads entered the state and absorbed more wind power, prices dropped even with the load increase.
It was a different story with the nation’s largest grid operator, PJM Interconnection, partly because the grid was already congested as data centers came online. PJM covers states across the mid-Atlantic and Midwest.
Hua said that vertically integrated electricity markets like those in the Southeast and West, where utilities are allowed to own generation, transmission and distribution, may have more potential for data centers to lower prices, although it’s not a given. That’s not the case in PJM.
A recent study commissioned by the Edison Electric Institute, which represents investor-owned utilities, concluded that PJM is the only area where rates have increased because of data centers. Going forward, the report found, states and utilities that charge data centers responsibly could see overall prices go down.
“You can identify the needs to serve a new customer and go out of your way to hold existing customers harmless,” said Matthew DeCourcey, vice president at Charles River Associates and a co-author of the report. “A big customer absorbing the shared costs means bills might go down.”
Patti Poppe, the CEO of Pacific Gas and Electric, has dubbed the phenomenon “rate-reducing load.” Poppe’s utility, the largest in California, has seen prices increase rapidly over the past few years, in part because of wildfire management and rising insurance costs linked to the company’s role in deadly fires in the state.
On the company’s earnings call in February, however, Poppe said that any increases could be balanced out as data centers help cover infrastructure costs. For each gigawatt of large load the utility adds, she said, customers could see savings of at least 1 percent.
Richard McCann, an economist with the consulting firm M.Cubed who has worked with large utilities, said in an interview that PG&E’s cost increases have had more to do with shifting fuel prices rather than wildfire costs, which largely have not hit customers’ bills. There is potential for data centers to blunt the impact, but that depends on things going right, he said.
“The big thing right now is uncertainty,” McCann said. “Yes, rates should go down. The problem is there are more data center proposals out there than will get built, so it’s hard to project what could materialize.”
Clemmer of the Union of Concerned Scientists added that a significant question is what kind of power is built and used. The Trump administration has been working to keep coal-fired power plants on the grid and to build more natural gas generation. As a result, utility customers might pay more because of rising natural gas prices or to cover the cost of keeping an expensive coal plant running.
“The reality is some of the actions of the administration work against these goals,” Clemmer said, referring to the policy goal of keeping utility rates down.
So now what?
Although data centers haven’t always raised electricity prices, analysts caution that the pattern in the future may be different, as projects become larger and gigawatt-scale complexes dominate the landscape.
“What’s happened already is not very reflective of a grid with a significant data center load,” said ICF’s Muthiah. “Now every year going forward, there will be a step change.”
Ryan Hledik, an analyst at Brattle Group, said the “headroom” of excess capacity that existed on electric grids through 2024 is getting filled up with all the new power demand, including from massive data center complexes. That headroom previously allowed demand to grow without always putting pressure on pricing.
One issue to watch is the growing use of large load tariffs, the contracts and rates between utilities and large electricity customers like data centers, he said. So far there are more than sixty proposed or existing such tariffs, and they can help shield residential customers from the costs of new infrastructure required by AI.
“If the price and the terms in [those tariffs] are such that they’re fully recovering the new costs associated with serving a data center,” Hledik said, “then you mitigate the risk of other customers having to foot the bill.”
Also, researchers are increasingly looking at ways to get more power out of existing lines and wires through grid-enhancing technologies. “There are software or hardware solutions that can squeeze more juice out of the current system,” Hua said. “I like to describe them as ibuprofen for the grid.”
One option is to have data centers agree to reduce their electricity use from the grid at peak demand times, perhaps by using their own on-site generation during certain periods. “If data centers can provide that flexibility,” Hledik said, “then you can potentially connect those data centers to the grid without needing to invest in as much new infrastructure.”
Some major tech companies also are exploring the possibility of paying other customers to reduce their demand during peak hours, he said.
In a Newsweek editorial in January, former Energy Secretary Jennifer Granholm, who served under former President Joe Biden, and former Arizona Gov. Doug Ducey (R) made a similar point, calling on states to incentivize technologies such as advanced grid conductors.
According to Hua, the role of public utility commissions will be critical in determining what happens with prices, since they play a role in making sure that utilities don’t overbuild for the AI boom. Historically, utilities have often overestimated demand, so accurate forecasting will be important, he said.
If many data centers don’t actually get constructed, then there could be stranded assets and “existing residential consumers may be left footing the bill,” he added.
