A household in Virginia opened a January power bill for $281. The month before, it had been about $100. That jump became the picture attached to a much bigger fight over who pays for the AI buildout — and the mechanism is worth understanding, because it explains both why bills are climbing and why the pain isn't spread evenly.
The size of the thing
One hyperscale data center can pull around 100 megawatts. The International Energy Agency puts that at roughly what 100,000 households use. Lawrence Berkeley National Laboratory expects data centers to account for 12% of all US electricity by 2028.
Demand like that doesn't slip in unnoticed. It lands in particular places, on particular grids, faster than anyone can build the generation and transmission to serve it.
How it gets to your bill
Two routes, and neither one puts a data center's power on your statement as a line item.
First, wholesale prices. When demand climbs against tight supply, the market price of electricity rises for everyone pulling from that grid. Analysts point to California, the Midwest and the mid-Atlantic as the most exposed. One 2026 study modeled demand-weighted wholesale prices rising 6% to 29% nationally on average by 2030 — up to 57% in the worst-hit regions.
Second, and this is the bigger one: cost allocation. Hooking up a huge new customer means new substations, transmission lines, upgrades. Utilities recover that through rates, and in most states the recovery gets spread across all customer classes rather than billed to the customer whose arrival caused the spending. Utilities nationwide asked for a record $31 billion in rate increases in 2025 — more than double the year before.
That spreading isn't a law of nature. It's a regulatory choice, made in rate cases, and it can be made another way.
Don't pin the whole increase on AI
Your bill is going up for several reasons at once. Transmission systems are being rebuilt. Wildfire hardening is expensive out West. Gas prices move. Extreme weather drives both demand and repair bills. Data centers are one ingredient, and treating them as the whole recipe makes it harder to judge what your utility is actually charging you for.
The honest version: AI demand is a serious and growing factor in some regions and nearly invisible in others.
What you can actually do about it
Read the rate case instead of the headline. When a utility files for an increase, the filing spells out what the money is for. Public utility commissions run comment periods, and residential comments get counted. That's the one moment where the allocation question — should big loads pay for the infrastructure they need — actually gets decided.
Find out whether your state has large-load tariffs. Several have moved toward special rate classes for data centers, with minimum-take commitments so the costs don't land on households when a project underdelivers. Whether your state has one is the best single predictor of whether this buildout shows up on your bill.
Look inside your own increase. Bills separate supply from delivery. If delivery charges are rising faster than supply, the money's going into infrastructure — and that's where the data center effect lives.
Then check whether a time-of-use rate helps. Where the crunch happens at specific hours, moving laundry, dishwashing and EV charging off the evening peak is the one lever a household genuinely controls.
The deal on the table
Communities hosting these campuses get tax revenue, construction jobs, and in some counties a serious chunk of the local economy. What they're asked to take in return is higher regional power costs and a very large industrial neighbor.
That might be a fair trade. It isn't a free one, and the bill gets split among a lot of people who were never asked.
Image: Connor Scott McManus, via Pexels
