America’s AI boom forces a reckoning over who pays the bill and who benefits in the end
Data centers are multiplying across the United States at a pace few communities anticipated, and the infrastructure demands of the AI boom are now reshaping household electricity bills, grid planning, and the politics of energy. A June 2026 analysis from Lawrence Berkeley National Laboratory projects data centers could more than double their electricity use by 2030, accounting for more than 40 percent of all U.S. electricity demand growth over five years. With residential electricity prices already up more than 36 percent since 2020 and grid operators warning of looming shortfalls, lawmakers returning to their districts this summer are finding a single question waiting for them: who actually pays for America’s artificial intelligence boom?
How big is the AI data center load on the U.S. power grid?
Goldman Sachs analysts estimate that data centers now account for roughly 40 percent of electricity demand growth nationwide and expect prices to keep climbing through the end of the decade. The pattern is most visible in Virginia, home to the densest cluster of data centers in the country. Facilities there now consume roughly 40 percent of the state’s total electricity, and Dominion Energy recently proposed its first base rate increase since 1992.
Grid operators are sounding alarms. PJM Interconnection, which serves more than 65 million people across 13 states, has projected it could fall six gigawatts short of its own reliability requirements by 2027. The timing mismatch is severe: data centers can be built in 18 to 36 months, while new transmission lines routinely take seven to ten years to permit and construct.
Who is actually paying for the new power infrastructure?
In the mid-Atlantic region served by PJM, projections show the average household could be paying tens of dollars more per month by 2028, with cumulative ratepayer costs reaching well over $100 billion by the early 2030s. Families are footing a growing share of infrastructure they did not ask for and will not directly use.
A June 2026 national survey found voters oppose a data center being built in their community by more than two to one, with nearly half strongly opposed. Despite that opposition, the same survey showed nearly two-thirds of respondents would welcome a solar farm nearby, a level of support on par with a distribution center or manufacturing plant. Solar power and battery storage have supplied more than 80 percent of new grid capacity added in recent years, in part because they can be deployed on a timeline that matches the urgency communities are facing.
Costs that flow to ordinary ratepayers include:
- Higher monthly bills tied to data center growth and grid upgrades
- Tax abatements and subsidies awarded to hyperscale operators
- Extended life for aging coal and natural gas plants kept online to backfill demand
- Transmission line projects that take a decade or more to permit
Industry representatives point out that electricity prices were rising before ChatGPT existed, and aging infrastructure, gas costs, and utility profit models all play a role. That context does not erase the central complaint now showing up at town halls: a structural shift in who is drawing power off the grid, with households absorbing the bill.
What energy sources are filling the gap right now?
Because new transmission is slow, utilities are leaning on natural gas and, in several states, keeping aging coal plants running longer than planned just to keep the lights on. In Texas, wind and solar met 36 percent of demand on the ERCOT grid through the first nine months of 2025, and federal forecasters expect utility-scale solar there to surpass coal generation for the first time in 2026.
Industry data shows solar currently occupies a small fraction of total U.S. land and an even smaller share of farmland, with no state approaching significant use of its prime agricultural acreage. Still, communities are right to demand transparency and environmental review before ground is broken on any project.
What is the broader political stakes for Congress this summer?
Lawmakers heading home will find the data center question already on the agenda, from Virginia to Nevada. Surveys show the issue is motivating a meaningful share of undecided voters. The choice facing members of Congress is not between growth and stagnation but between an energy strategy that spreads the true cost of the AI boom fairly, through natural gas, nuclear, transmission upgrades, solar, and storage, or one that keeps asking households to subsidize infrastructure they never approved.
The competition angle is real: AI is not going away, and falling behind on power means falling behind on AI itself at a moment when global competitors show no sign of slowing down. Lawmakers do not need to convince Americans to like AI. They need a plan that treats families as stakeholders in its progress rather than subjects of its infrastructure or its surveillance.
FAQ
How much have U.S. residential electricity prices risen since 2020?
Residential electricity prices have climbed more than 36 percent since 2020, with analysts pointing to AI data center demand as a growing driver of that increase.
What share of U.S. electricity demand growth are data centers expected to drive?
A June 2026 Lawrence Berkeley National Laboratory analysis found data centers could more than double their electricity use by 2030, representing over 40 percent of national electricity demand growth in just five years.
How much could PJM ratepayers end up paying for data center growth?
Projections for the PJM Interconnection region, which serves more than 65 million people across 13 states, show average households could pay tens of dollars more per month by 2028, with cumulative ratepayer costs reaching well over $100 billion by the early 2030s.
This article summarizes reporting from naturalnews.com.