Data Centers Aren't Raising Your Power Bill. Bad Policies Are.

Large, steady customers can make electricity cheaper—as long as regulators let supply keep pace.

Courtesy Joe Raedle/Getty Images

As data centers have sprung up across the country to power the AI boom, many Americans have begun to worry about what they will do to electricity bills. A recent Harvard poll found that two-thirds of Americans believe that building a new data center nearby would raise local electricity prices. The worry is already fueling a backlash: lawmakers in more than 30 states have introduced over 300 data-center bills this year, including outright construction moratoriums, many prompted by concerns over electricity prices.

The concern is understandable: the industry’s rapid growth has coincided with a sharp increase in residential electricity prices in many parts of the country. Data centers are notoriously power hungry, consuming nearly 5 percent of all U.S. electricity in 2024. More demand typically means higher prices.

But a series of new studies suggests that AI’s energy demands need not raise household electricity bills—in fact, so far, they haven’t. If anything, data centers have placed modest downward pressure on retail rates. Large new loads often lower rates for everyone by spreading the electrical grid’s enormous fixed costs across a broader customer base. As data centers continue to expand, the real question is whether regulators will allow enough new generation and grid investment to meet rising demand while ensuring that data centers—not existing ratepayers—pay for the infrastructure their growth requires.

“Have Data Centers Raised Your Electric Bill?” That’s the question a new working paper by Asa Watten and Geoffrey Blanford of the Electric Power Research Institute (EPRI) and John Bistline of Watershed asks. Their answer, surprisingly, is no: data centers pushed residential rates modestly down between 2015 and 2024. After accounting for the possibility that developers simply chose states where electricity was already likely to remain cheap, the authors estimate that doubling a state’s data-center capacity caused residential rates to fall about 3.5 percent. The average American lived in a state where data-center capacity grew 160 percent between 2019 and 2024, leaving rates roughly 6 percent lower than they otherwise would have been.

A June paper from Columbia University’s Center on Global Energy Policy surveys the literature and finds that recent price increases were driven not by demand from new users such as data centers but by a host of other factors. These include grid hardening and expansion, disaster recovery, volatile fuel prices, and regulatory mandates. States with greater load growth generally saw smaller price increases, or even price declines. A recent analysis from Lawrence Berkeley National Laboratory reached similar conclusions.

That appears to be true even in the largest data-center markets. The consultancy E3 examined Virginia, which hosts more data centers than anywhere else in the country, and found no evidence that the industry shifted costs onto households. A separate analysis of hyperscale data centers in California, Oregon, Virginia, and Mississippi found either no effect on retail rates or the potential to generate “surplus value.” A typical 100-megawatt data center, they estimate, can produce about $3.4 million a year more in revenue than it costs to serve. That surplus puts downward pressure on rates for other customers.

How can more demand lead to lower prices? Retail electricity does not function like an ordinary competitive market. Utilities are regulated monopolies whose rates are set to recover average costs. Most of those costs are fixed. Poles, wires, substations, and power plants must be paid for regardless of how much electricity customers use. When a large customer like a data center plugs in, those costs are spread across more kilowatt-hours, reducing the average cost of supplying each one.

Data centers are unusually large and steady customers. That allows a utility to sell more power through infrastructure it has already built, meaning it collects more revenue while lowering the average cost per kilowatt-hour. This is how the American grid worked for most of the twentieth century, when demand grew year after year while real electricity prices fell.

It’s also why utilities often court large industrial customers. Pacific Gas and Electric, for example, estimates that every gigawatt of new data-center load in its territory could cut household bills 1 to 2 percent by making fuller use of a grid that currently operates at about 45 percent utilization.

So if data centers haven’t been raising residential bills, what has? Prices haven’t increased everywhere. According to Berkeley Lab data, inflation-adjusted residential retail electricity prices fell in 23 states from 2019 to 2025, including in North Dakota, Nebraska, and New Mexico, which experienced some of the fastest electricity-demand growth.

The largest price increases were concentrated on the West Coast and in the Northeast, where state policies have made electricity more expensive to produce and deliver. Renewable mandates require utilities to procure growing shares of wind and solar. Net-metering programs shift grid costs from rooftop-solar owners onto other customers. And carbon-pricing programs add further costs to fossil-fuel generation. Together, these policies have added substantially to rates in many of the states where electricity prices have risen sharply.

Other policies compound the increases. In California, wildfire-related expenses now account for a substantial share of utility costs—a burden that reflects the state’s longstanding failure to manage vegetation and reduce extreme fire risks. And in New England, high prices are in part the result of policy decisions not to expand the region’s natural-gas pipeline capacity, which leaves its utilities paying a premium for the fuel they depend on.

The sharpest increases are found in states that have pursued the country’s most aggressive climate policies, not those with the most data centers. California, with some of the nation’s fastest-rising electricity rates, has seen relatively modest data-center growth. Virginia, where data centers consume more than a fifth of the state’s electricity, has experienced price increases near the national average.

Still, that hasn’t stopped some politicians from fearmongering. New York lawmakers last month passed a first-in-the-nation moratorium on large data-center construction, blaming the facilities for rising rates—in a state where the climate law accounts for 5 to 10 percent of customers’ monthly bills and where regulators have blocked new gas plants since 2021. Senator Elizabeth Warren claimed last month that electricity bills near large data centers “have gone up by as much as 267 percent” over five years. (PolitiFact looked into it: the figure refers to wholesale prices in a few locations, not what anyone actually pays at home.)

The EPRI researchers are careful to note that, while data centers thus far haven’t driven price increases, supply constraints could reverse the effect in the years ahead. The past decade’s data-center growth occurred while the grid still had slack; the coming AI buildout will require new grid capacity at a time when key equipment like transformers and gas turbines face multiyear lead times and backlogs. If regulators prevent supply from responding to the new demand, prices will rise—but that’s an argument for making it easier to build, not for banning the data centers who want to pay for it.

It is also an argument for policies that make large loads carry their own weight. This is already happening. At least 38 special tariffs for large customers have been adopted since 2018. This month, Oregon became one of the first states to put data centers in their own rate class, approving a 29.7 percent rate increase for Portland General Electric’s largest users to cover the costs their demand creates. Residential rates there will now fall 1.3 percent.

Data centers do not inherently cause higher electricity prices. The evidence so far suggests they have lowered, not raised, household bills. Whether the AI buildout benefits or burdens ratepayers will depend on whether regulators allow electricity supply to grow with demand—and ensure that data centers, rather than existing customers, pay for the new infrastructure they require.