How Big Tech’s AI Dreams are Driving Your Electricity Bills Through the Roof
EU tech policy fellow George Colville argues that the AI industry's growing energy demands are unfairly driving up electricity prices for ordinary Americans, with tech giants leveraging their power to shift the cost burden onto taxpayers and households instead of bearing it themselves.
Americans are learning the AI boom comes with many costs. Not only do the world’s largest corporations appropriate people’s and businesses’ data without payment, but they also drive up the prices of everything from semiconductors to prime industrial real estate, then force taxpayers to subsidize their sprees. Now another bill has come due – for electricity to power AI.
This summer has seen energy bills soar by up to 35% in parts of the U.S. From Ohio, to Pennsylvania, to New Jersey, where Google, Amazon, Microsoft, and others are building data centers at a record rate, average monthly bills have already jumped by up to $27. And this is just the beginning.
The AI-driven data center buildout is projected to push U.S. energy demand growth from a low and steady 0.41% annual rate between 2000 and 2023 to 2.4% per year from 2025. Over a decade, that represents a 26.7% total increase compared with the 4.2% expected under pre-AI conditions. Accordingly, energy utilities and grid operators are investing in new infrastructure – much of it based on fossil fuels – that can only be paid for by higher energy prices. One recent study found that residential ratepayers served by PJM, the energy utility for the mid-Atlantic and parts of the Midwest, could see their bills increase by between 30% and 60% by 2030.
To put the data center boom into perspective, consider Mark Zuckerberg’s latest project: a Manhattan-sized complex in Louisiana that will require 5GW of energy. That equates to the electricity consumption of 3.75 million homes, or roughly the residential power demand of the entire Chicago metro area. It is also more than any single power plant in the U.S. can produce, with the exception of the Grand Coulee Dam.
Energy regulators have long recognized that different customers impose different cost burdens on energy infrastructure. In what is known as the cost causation principle, users have historically been broken into different classes, each of which is charged a price per unit broadly representing the cost it imposes on the grid.
Most utilities, for instance, require industrial scale users to pay for the initial cost of connecting their facility to the grid. They then, however, charge these industrial facilities less per unit of energy than households pay, because it costs less to serve a single industrial plant with a peak demand of 1 MW than it does to serve 2,000 homes with a collective peak demand of 1 MW.
This history and its underlying principles imply the costs of building the energy infrastructure required by data centers should be borne by those causing them. This means the new plants and grid infrastructure catering to AI energy demand should be paid for by those making this investment necessary: the AI industry.
Increasingly, however, the corporations dominating data center development have used their concentrated financial and political power to push these costs onto the bills of individual households, using them and other industries to subsidize their enormous energy appetite.
One way they are doing so is by misleadingly claiming that data centers will deliver major employment and economic benefits to local communities, which makes local authorities fear losing out to neighboring towns and states. The corporations that dominate AI are also adept at stoking race-to-the-bottom thinking among the utilities and grid operators themselves, by making the case the AI boom represents a once-a-lifetime opportunity to gain new energy-intensive customers.
The ultimate result is to incentize the utilities to exploit their own monopoly power over their captive residential and commercial customers, raising prices to subsidize lower-price offers to potential data center investors deciding between competing energy providers. As a recent report demonstrates, utility companies are offering data center operators favorable – even below-cost – rates, either as a group or as tailored special contracts. In parallel, some utility regulators are being undermined by state legislatures. In Mississippi, for instance, legislators recently stripped their regulator of any authority to review contracts between utilities and data centers.
Making AI companies pay for the energy infrastructure they require is possible. Options include directing public utility commissions to create a new rate classification for data centers, or requiring them to commit to fairly-priced long-term contracts and minimum contract payments. Ohio’s Public Utilities Commission is seeking to do this but faces concerted lobbying from Big Tech.
Energy markets and their regulators have a long history of fairly distributing the cost of infrastructure spending, making those responsible for higher costs pay their fair share. Tech giants are using their concentrated economic and political power to upend this tradition. Most recently, they have wooed President Trump into promising to help them tap more easily into local and regional energy grids, in ways that would put them into competition with even more American households.
If they have their way, many citizens will be left unable to pay their bills and ordinary businesses may struggle to stay afloat.
This article was featured in The Corner Newsletter: September 9th, 2025. Subscribe to The Corner below.