As data centers multiply due to AI demand, volatility is socialized onto consumers while stability is privatized by corporate giants.
You may not realize it yet, but your next electricity bill may carry a hidden surcharge—the cost of the artificial-intelligence (AI) revolution.
Across the country, utilities are racing to meet the extraordinary energy appetite of the AI economy. In doing so, they are quietly transferring part of that burden to the rest of us. While the world marvels at what large language models can write or design, fewer people are asking a more basic question: who’s paying to keep their servers running?
A New Geography of Cost
According to the Energy Information Administration, between 2020 and 2024, household electricity prices in the United States rose about 25 percent—from 13.15 cents per kilowatt-hour to 16.48 cents. For most families, that increase felt like another inflation story. But peel back the averages, and a new geography of cost emerges.
In Maryland, electric bills climbed 32 percent, outpacing both inflation and wages. The culprit wasn’t household behavior; it was the surge of power demand spilling across state lines from northern Virginia’s data-center corridor — a region now consuming more electricity than entire mid-size cities.
In Texas, as a result of absorbing more than six gigawatts of new data-center load, prices rose 28 percent, but it could have been worse if it weren’t for the addition of new generation capacity and a flexible market design. And in Georgia, the completion of two new nuclear reactors at Plant Vogtle kept the household price increase near 17 percent.
The Grid Split in Two
Beneath the surface of national statistics lies a structural divide. In parts of the Mid-Atlantic grid, wholesale prices have nearly doubled since 2020. At the height of congestion in Virginia’s “Data Center Alley,” wholesale prices spiked above $500 per megawatt-hour—levels that once signaled emergency conditions. Research from Lawrence Berkeley National Laboratory found that roughly 40 percent of that congestion can be traced directly to data-center growth.
These are not abstract numbers. They are line items on kitchen tables in Baltimore and beyond—the financial echo of an energy system bending under the weight of computation.
In theory, the grid is a shared resource. In practice, the AI boom has split it in two.
The Power of Hyperscalers
Households live by the arithmetic of the monthly bill. Hyperscale companies—Amazon, Microsoft, Google, and Meta—live by the calculus of power purchase agreements. Collectively, they have locked in more than 48 gigawatts of renewable energy through decades-long contracts, shielding themselves from market swings while claiming environmental virtue.
Many now own the plants outright. When Amazon bought the 960-megawatt Cumulus Data nuclear campus in Pennsylvania, it signaled a new era: corporations no longer rent power; they own it. Others have built microgrids, on-site gas turbines, and battery farms capable of running indefinitely when public systems falter.
They even move workloads around the globe to chase cheaper, cleaner electricity—a strategy Google calls carbon-aware computing. Your refrigerator can’t shift its cycles to follow the wind in Texas. Their algorithms can.
This is the quiet asymmetry of the AI age. Households are price takers; hyperscalers are price makers. They engineer their exposure, arbitrage between markets, and build private infrastructures that turn volatility into profit.
And it is still far from clear whether this rush to build is not just another tech bubble. Even if it is, the implications for consumer electricity bills could still be impacted for years to come from stranded assets if careful system design, operation, and planning are not adhered to in the rush to build.
Who Really Pays for Growth
If this were merely a story of corporate ingenuity, it might be admirable. But the financial architecture of the grid ensures that costs don’t vanish—they migrate.
Utilities recover infrastructure spending through rate cases approved by regulators. When a new substation or transmission line is built to serve a data-center cluster, that cost often enters the general rate base. The household that uses no more power than last year ends up paying for the expansion of a digital empire it will never visit.
In some states, lawmakers are beginning to push back. New Jersey is considering rules that would require data centers to source their electricity from new projects rather than from capacity already serving homes and schools. Oregon and Washington are exploring similar measures. The principle is simple: if AI creates new demand, it should also bring new supply.
That logic can be extended. Regulators could impose impact fees so that large energy consumers pay proportionally for the grid stress they cause. They could require on-site generation standards, ensuring that new data campuses meet part of their demand with renewables or modular natural gas units. And they could design bill-stabilization shields to keep wholesale price shocks from cascading into residential rates.
In short, those who profit from the grid’s strain should help fund its strength.
Efficiency’s Mirage
Tech companies insist that efficiency will save us—faster chips, smarter cooling, streamlined algorithms. These innovations matter. But efficiency without governance is a mirage. Every leap in computing power seems to invite an even larger leap in consumption. The energy saved by one generation of processors is devoured by the next generation of models.
By 2030, data centers—many devoted to AI—could consume nearly nine percent of all US electricity, more than the entire steel industry. That trajectory doesn’t just challenge our generation capacity; it tests our social contract.
What, then, would fairness look like?
It begins with acknowledging that the grid is not infinite, and that private innovation built atop public infrastructure carries public obligations. The companies shaping the digital future should not be permitted to privatize stability while socializing volatility.
Who Pays to Power the Future
Policymakers now face a choice as stark as any in a generation. They can design mechanisms that protect households and enlist hyperscalers as partners in grid resilience, or they can allow this imbalance to harden until public trust fractures.
The question is no longer whether AI will change the world. It already has. The question is who will pay to power it.
About the Authors: Morgan Bazilian and Brandon N. Owens
Morgan D. Bazilian is the Director of the Payne Institute and Professor at the Colorado School of Mines, with over 30 years of experience in global energy security and investment. A former World Bank lead energy specialist and senior diplomat at the UN, he has held roles in the Irish government.
Brandon N. Owens is a clean energy innovation executive and global thought leader at the nexus of energy, artificial intelligence, and institutional governance. His career spans work at the National Renewable Energy Laboratory, General Electric, and S&P. He is the author of The Wind Power Story (2019) and the forthcoming book Cognitive Infrastructure (2025). He can be reached at [email protected]
Image: Daniele Mezzadri/shutterstock
