How AI Data Centers Find Their Way Onto Your Electric Bill
PHILADELPHIA — The artificial intelligence boom runs on electricity, enormous amounts of it, and the facilities that supply that power are increasingly showing up in household electric bills. Understanding how a distant server farm ends up on your monthly statement requires a short tour of how electricity is bought, moved, and paid for.
The scale of the new demand is hard to overstate. A single AI-focused hyperscale data center can draw 100 megawatts or more, consuming as much electricity in a year as roughly 100,000 households, according to industry and research accounts. U.S. data centers accounted for more than 4 percent of the nation’s electricity use in 2024, and some projections put that share as high as 9 to 17 percent by 2030. Per square foot, a data center can use 10 to 50 times the energy of a typical office building, with cooling the servers accounting for a meaningful share of the load.
When demand rises this fast, the grid has to keep up. Utilities and grid operators must ensure enough power is available at all times, which means building new power plants, upgrading transmission lines, and paying to keep reserve capacity ready for peak moments. A data center can be built in a couple of years, but power plants and transmission lines take far longer to plan, permit, and construct, and the U.S. Department of Energy estimates data center electricity use could double or triple between 2023 and 2028.
Here is where the household bill enters the picture. In most cases, the costs of new grid infrastructure are spread across all customers through electric rates. Rate structures have generally not been redesigned to assign data center costs only to the companies creating them. As Penn State researchers put it, even if you never use AI directly, you may still pay higher bills because you are helping fund the expanded infrastructure needed to support that growth.
The costs reach households through several channels. In regions where data centers cluster, concentrated demand pushes up wholesale electricity prices. In the PJM grid region, which spans 13 states and hosts the country’s largest concentration of data centers, capacity market prices rose 174 percent for the 2025 to 2026 delivery year, and Baltimore residents saw average monthly bills jump by more than $17 after a 2025 auction, according to published analyses. Separately, the cost of new substations, feeders, and high-voltage lines, routinely running into the billions of dollars, is divided among all customers under current tariff rules. Customers across seven PJM states are covering $4.4 billion in transmission upgrades tied to data centers built between 2022 and 2024, the Union of Concerned Scientists has estimated.
Philadelphia sits inside the PJM region, so these mechanisms apply to local households, though the exact impact on any given bill varies by utility and rate plan. Even where data centers are only one factor among many behind rising rates, the added pressure is real and growing.
Policy efforts to shift the burden have struggled. A federal measure known as the Ratepayer Protection Act, which addressed who should pay for grid costs tied to large energy users, failed in the Senate on Sept. 30 by a vote of 57 to 43. Some utilities have asked federal regulators for rule changes that would let large customers cover their own infrastructure costs. State utility commissions, which set the rate structures that determine who pays, remain the key decision makers, and their proceedings are worth watching.
None of this means data centers are the only reason bills are rising. Residential electricity prices have climbed for many reasons, and tech companies disclose limited detail about their energy use, which makes it hard to isolate one driver. But the mechanism is straightforward. Big new demand needs big new infrastructure, and until rates are redesigned, the cost of both lands on everyone connected to the same grid.