New Jersey residents are staring down a 20% electricity rate hike starting June 1, and one of the biggest reasons is that tech companies need unfathomable amounts of power to help you write a slightly better cover letter. Data centers housing AI infrastructure have nearly doubled in the U.S. since 2021, and the grid can barely keep up. Guess who's paying for the gap.

Your Utility Bill Has a New Landlord Named ChatGPT

Here's the thing about AI: it is extraordinarily, almost offensively hungry for electricity. According to a study from the Electric Power Research Institute, a nonprofit, AI searches consume ten times more electricity than a standard internet search. Every time someone asks a chatbot to summarize an email they could have read themselves, a small piece of the American power grid evaporates.

A report from Schneider Electric projects that U.S. electricity demand will increase 16% by 2029, driven primarily by the proliferation of data centers. Most of those data centers pull from the national grid. That means the infrastructure costs, the rate increases, the grid strain, all of it flows downstream to regular ratepayers who never asked to subsidize Silicon Valley's infrastructure buildout.

The Numbers Are Genuinely Staggering

The number of data centers in the U.S. nearly doubled between 2021 and 2024, according to a report from Environment America. The largest concentrations sit in Virginia, California, and Texas, with Virginia in particular becoming something of a data center capital. And they're not just multiplying. They're getting bigger.

Torsten Sløk, chief economist at Apollo Global Management, estimates that data centers will need an additional 18 gigawatts of power capacity by 2030. For reference, New York City runs on about 6 gigawatts. So we're talking about building the equivalent of three New York Cities' worth of power demand from scratch, in under five years, mostly to run servers that generate heat that requires more power to cool down.

About 4.4% of all U.S. electricity went to data centers in 2023, according to a study from the Department of Energy's Lawrence Berkeley National Laboratory. That share is climbing fast. Dave Turk, former deputy secretary of the U.S. Department of Energy, told CBS MoneyWatch flat out: "AI is an increasing part of data centers and certainly responsible for increased electricity demand."

Sweetheart Deals in the Dark

Mark Wolfe, executive director of the National Energy Assistance Directors Association, put it about as plainly as anyone in this industry ever has. "As utilities race to meet skyrocketing demand from AI and cloud computing, they're building new infrastructure and raising rates, often without transparency or public input," he told CBS MoneyWatch. "That means higher electricity bills for everyday households, while tech companies benefit from sweetheart deals behind closed doors."

He's not speaking abstractly. Dominion Energy, one of Virginia's largest utilities, proposed a price hike of $8.51 per month in 2026 and in the same breath floated creating a brand new "rate class for high energy users, including data centers." Read that again. The utility is raising everyone's rates because demand is high, and separately considering a special pricing tier for the companies causing that demand. Whether that special tier will actually protect consumers or just formalize the arrangement is very much an open question.

Electricity prices have already risen 4.5% in the last year, according to Labor Department data. This summer is expected to be worse.

The Grid Itself Is Starting to Crack

Beyond the bill, there's a reliability problem forming. The North American Electric Reliability Corp warned in a recent report that AI and cryptocurrency facilities are being built faster than the power plants and transmission lines required to support them, which is resulting in what it called "lower system stability." That's a polite way of saying the lights might flicker.

PJM, the grid operator covering 13 states plus Washington D.C., flagged data center demand as a factor that could produce capacity shortages in its 2025 forecast. This is the kind of infrastructure stress that takes years to fix and seconds to cause a crisis. The companies racing to build more data centers are not the ones who will be sitting in the dark when the grid buckles.

And the Big Beautiful Bill floating through Congress right now would make all of this worse. Analysts at the Rhodium Group predict that the Republican budget package, which would strip out tax credits from the Inflation Reduction Act, could increase a typical family's energy costs by nearly $400 a year. The bill that guts clean energy incentives. Passed right as energy demand is spiking to historic levels. The timing is almost artistic.

This Isn't Just About AI

To be fair, data centers are not the only thing pushing electricity costs up. CBS News notes that natural gas prices, inflation, and the ongoing electrification of buildings and vehicles all factor in. Energy is getting more expensive for a lot of reasons at once, and it would be too simple to pin all of it on tech companies.

But data centers are the accelerant. They are the factor that is growing the fastest, receiving the least regulatory scrutiny, and generating the least public conversation relative to their impact. The AI boom is being built on physical infrastructure that requires physical power, and that power comes from somewhere. Right now, it mostly comes from your utility bill.

The Dingo Take

Picture a steel mill moving into your county, doubling the local power draw, and getting a private rate negotiation with the utility while your monthly bill quietly climbs 20%. People would lose their minds. Politicians would hold press conferences in hard hats. There would be town halls, investigations, maybe a 60 Minutes segment. Now replace the steel mill with a gleaming data center full of servers training a language model, and somehow the same story becomes a technology lifestyle piece.

The tech industry has spent years cultivating an image as a clean, frictionless, almost weightless part of the economy. Code doesn't belch smoke. Servers don't leave tire tracks. But the physical reality is that AI is an industrial operation that happens to be housed in beige buildings with good PR, and it is consuming power at a scale that is already straining the grid and landing on household bills. The fact that we are not having a serious political fight about who pays for this infrastructure is a minor miracle of branding.

Somebody is going to get stuck with this tab. Based on everything we know about how utilities, tech companies, and the current Congress operate, it is not going to be the companies reporting record profits. It is going to be the New Jersey family that just opened their June bill and has no idea why it jumped.

Sources