June 28, 2026

The same power lines that feed a data center feed the houses around it, which is why the AI boom is showing up on household electricity bills. In regions thick with data centers, power prices have jumped 267 percent in five years, and a single obscure grid auction recently added billions in costs that all ratepayers absorb. The question is no longer whether AI uses enormous power; it is who pays for it, and right now the default answer is you.

The Meter Is Running: How the AI Boom Ended Up on Your Power Bill

A Virginia man opened a $281 electricity bill in January, nearly triple what he paid the month before, and the cause was not his thermostat. It was the data centers humming a few miles down the road.

By The Daily Reflection · June 28, 2026 · 8 min read

Start with that bill. One month it read around $100. The next it read $281. The homeowner had not bought a hot tub or left the air conditioning running with the windows open. He lives in northern Virginia, the densest concentration of data centers on earth, and the cost of feeding those buildings had quietly arrived in his mailbox.

He is not alone, and this is not a quirk of one utility. In the regions where data centers cluster, electricity prices have climbed 267 percent over the past five years. Nationally, power prices rose 6.9 percent in 2025, more than double the 2.9 percent headline inflation rate, and data centers accounted for roughly 40 percent of the growth in electricity demand driving those increases. Nearly four in five Americans now say they are worried that the data-center construction boom will push their energy bills higher.

Here is the part that should bother you: almost nobody agreed to this, and almost nobody can see how it happened.

The Bill That Has Nothing to Do With Your Thermostat

For most of the past two decades, American electricity demand was essentially flat. Efficiency gains offset growth, utilities planned around a stable baseline, and the grid stayed boring. Then generative AI arrived, and with it an appetite for computing power that has no modern precedent.

A single large data center can draw as much electricity as a midsize city, and it draws it around the clock. Independent analysts and the Department of Energy project that data centers will consume somewhere between 6.7 and 12 percent of all US electricity by 2028, up from about 4.4 percent in 2023. The Federal Reserve Bank of Dallas estimates that as data-center demand doubles over the next five years, wholesale power prices could climb by as much as 50 percent.

That demand does not stay politely inside the tech company's fence line. The grid is a shared system. When a new buyer shows up needing enormous, constant power, everyone connected to the same wires helps pay to build the generation and transmission required to serve it. Households are projected to see electricity prices rise another 6 percent through 2027, and that is before the largest wave of announced data centers even comes online.

You did not add this load. You are helping pay for it anyway.

The Auction You Have Never Heard Of

Here is the piece of this story most coverage skips, because it lives inside a market almost no ordinary ratepayer has any reason to know exists.

Much of the eastern United States gets its power through a grid operator called PJM, which runs an annual event called a capacity auction. In plain terms, the auction pays power plants in advance to promise they will be available when demand peaks, and the price it sets flows straight onto everyone's bills. For years that price was modest. In the 2024 to 2025 auction it sat at $28.92 per megawatt-day.

Then it detonated. The next auction cleared at $329.17 per megawatt-day, more than ten times higher, slamming into the price ceiling regulators had set. The December 2025 auction hit the cap again, $333.44 per megawatt-day, for the 2027 to 2028 delivery year. Two years running, the market essentially maxed out.

Why? PJM's own market monitor put numbers on it. Data centers caused 63 percent of the price increase in one auction, adding $9.3 billion in capacity costs that all ratepayers must absorb. In the December auction, data-center load accounted for $6.5 billion, about 40 percent, of $16.4 billion in total capacity costs.

Read that again: a single annual auction most Americans have never heard of just repriced their electricity, and a large share of the increase traces straight to data centers.

Translated into a monthly bill, the capacity spike alone is expected to add about $18 a month for the average household in western Maryland and about $16 a month in Ohio. That is the cost of the AI boom, itemized, except it is not itemized. It is folded into the same line as the power that runs your refrigerator.

"Don't Worry, We Pledged"

The politics of this have not caught up to the math, and what political response exists is mostly theater.

Earlier this year, President Trump summoned executives from the leading AI companies to the White House to affirm what was branded a "Ratepayer Protection Pledge," a commitment that the costs of new data centers would not be passed on to ordinary consumers. It made for a clean headline. The problem is that the White House built no mechanism to enforce it: no rule, no penalty, no agency tasked with checking whether the promise survives contact with reality.

A pledge without enforcement is a feeling, not a policy. And it sits awkwardly next to the capacity-auction numbers, which show the costs are already being socialized across the grid right now, pledge or no pledge. You cannot credibly promise that something will not happen once it has already started happening.

This is the recurring shape of AI-era policy: a confident announcement, a reassuring noun like "protection," and underneath it nothing that would actually bind the companies whose demand is moving the price.

Who Pays, and Who Gets to Decide

Strip away the press conference and the real fight is happening in places that never trend: state utility commissions and the dry mechanics of how a "rate class" gets drawn.

The central question is simple to state and brutal to resolve. When a data center needs a new substation, a new transmission line, or an entire new power plant, who pays for it? If the cost goes into the general rate base, every household splits the tab. If it goes into a dedicated data-center rate class, the companies that created the demand carry their own weight.

A handful of states are starting to force the question. Oregon became the first to create a dedicated data-center rate class, walling those costs off from ordinary customers. Virginia, home to the densest data-center corridor in the country, has advanced a bill, SB 253, that would shift distribution and capacity costs away from households and onto the data centers themselves. At least six states have floated construction moratoriums, and seven have moved to repeal or restrict the tax incentives, worth billions, that lured the data centers in the first place.

That last detail is the quiet absurdity of the current arrangement. In many places, taxpayers first subsidized the data centers to come, through tax breaks, and now subsidize them again, through higher power bills. The benefit was privatized twice and the cost was socialized twice.

When the people who create the demand do not pay for the supply, the difference does not vanish. It lands on everyone else.

The Bargain Nobody Offered You

It would be unfair to pretend the other side has no case, so here it is, put as strongly as it deserves.

The AI buildout is not frivolous. If even some of the productivity promises hold, the economic return could be enormous, and a country that wants to lead in AI needs the physical capacity to run it. Data centers bring construction jobs, a tax base, and, in some communities, real investment. The grid upgrades that all this demand requires could, handled well, modernize aging infrastructure that needed replacing anyway. There is a version of this story where the buildout is a generational investment rather than a hidden tax.

But notice who sits inside each half of that bargain. The upside flows to shareholders, to a handful of host counties, and to the firms selling the models. The cost flows to every household wired to the same grid, including the ones that will never run a single AI query and the ones for whom an extra $16 or $18 a month is the difference between covering groceries and not. Goldman Sachs has warned that middle-class and lower-income Americans are absorbing the data-center boom through higher electricity bills and even higher food costs. A flat surcharge is regressive by nature: trivial at the top, painful at the bottom.

The deeper problem is not that AI uses power. Everything uses power. The problem is that a massive reallocation of who pays for the grid is being decided through capacity auctions, utility filings, and a White House pledge with no teeth, none of which the people footing the bill were ever asked to weigh in on.

The Bottom Line

The AI revolution has a physical body, and that body runs hot. The intelligence may live in the cloud, but the cloud lives in enormous buildings that pull power from the same wires as your house, and the cost of feeding them is already moving onto your bill through machinery you were never meant to notice.

The pledge from the White House is theater. The real decisions are being made in capacity auctions priced in megawatt-days and in state rate cases argued by utility lawyers, and most of them currently default to the same answer: spread the cost across everyone. A few states are trying to change that default by making data centers pay for the grid they require. Whether that approach spreads will decide who funds AI's infrastructure: the companies profiting from it, or the public.

The question was never whether artificial intelligence would change your life. It already has, in a way you can measure. Open the envelope. The meter has been running the whole time, and for now, the bill is addressed to you.

The Daily Reflection cuts through the noise to find the stories that actually matter. Follow for thoughtful takes on politics, technology, and whatever's shaping our world.

Search This Blog

The Daily Reflection