American data centers consumed 183 terawatts-hours of electricity in 2024, representing more than 4% of the country’s total electrical consumption. This demand could more than double by 2030 to reach 426 TWh. This energy explosion is radically transforming the American electrical economy and reveals a troubling asymmetry: tech giants capture the tax profits from digital infrastructure while ordinary consumers finance their electrical supply.

The artificial intelligence revolution is generating unprecedented electrical demand that is upending the economic balance of the American energy system. 78% of Americans fear that new data centers will increase their electricity bills, a concern grounded in an implacable accounting reality: technology companies privatize their gains while mutualizing their costs.

The Essentials

  • American data centers consumed 183 TWh in 2024 and could reach 426 TWh by 2030, a 133% increase
  • Electricity prices have risen up to 267% over 5 years in areas with high concentrations of data centers
  • In Virginia, AI data center demand caused PJM capacity market prices to spike 833% and could increase the state’s energy demand by 183% by 2040
  • Microsoft is financing the reopening of Three Mile Island with a federal loan of $1.6 billion to exclusively power its data centers

The Explosive Bill: 267% Increase in High-Density Areas

Monthly electricity costs have increased up to 267% compared to five years ago in areas near significant data center activity. This tariff spike is not a regional anecdote but a systemic transformation of the American electrical market.

A Carnegie Mellon University study estimates that data centers and cryptocurrency mining could lead to an 8% increase in the average American electricity bill by 2030, potentially exceeding 25% in markets with high demand in central and northern Virginia. This conservative projection masks a reality already tangible: the typical American household was billed $142 per month for electricity in 2024, a 25% increase from the $114 monthly average in 2014.

Virginia perfectly illustrates this shift. In 2023, data centers consumed approximately 26% of the state’s total electrical supply. In certain areas like Frankfurt, they represent 42% of local demand, and nearly 80% in Dublin. This geographic concentration creates unprecedented tensions on local electrical grids and forces utilities to invest massively in new infrastructure.

Demand from artificial intelligence data centers in Virginia contributed to an 833% increase in PJM capacity market prices for 2025-2026 compared to the previous year. This tariff explosion reveals the scale of the infrastructure challenge: the state’s energy demand could increase by 183% by 2040.

Microsoft Privatizes Three Mile Island But Socializes Network Costs

The agreement between Microsoft and Constellation Energy for the reopening of Three Mile Island crystallizes this logic of economic asymmetry. The owner of the closed nuclear facility is seeking a federal loan guarantee of $1.6 billion to finance its restart plan and sell electricity to Microsoft to power data centers.

Microsoft plans to absorb 100% of the output from a restarted Three Mile Island nuclear plant. The 20-year agreement will power its data centers in Pennsylvania as well as Chicago, Virginia, and Ohio. This contractual duration, significantly longer than Microsoft’s traditional solar and wind agreements, reveals the priority placed on securing power supplies for AI.

Yet this energy privatization comes with a socialization of infrastructure costs. The CEO of Constellation confirmed that federal money is essential for this project to move forward, particularly through the Inflation Reduction Act tax credits reserved for existing nuclear plants. Constellation plans to spend approximately $1.6 billion to restart Three Mile Island Unit 1, which will be eligible for federal 45Y clean energy tax credits.

This strategy reveals a fundamental contradiction: Microsoft secures decarbonized and reliable electrical supply for its lucrative AI activities, but the transmission and distribution infrastructure that delivers this electricity to its data centers remains financed by all taxpayers through regulated rates.

Virginia Imposes New Rate Structure to Protect Consumers

Facing this growing asymmetry, Virginia is experimenting with an unprecedented regulatory approach. Virginia’s State Commission approved a new electricity rate for large customers, notably artificial intelligence data centers. Beginning January 2027, affected customers must pay at least 85% of contracted demand for distribution and transmission and 60% of production demand.

This measure explicitly aims to protect other consumers. Previously, a data center could sign an agreement for 100 megawatts of electricity but use only 20 megawatts during construction. Dominion would build network improvements for the full contract amount, and the charges were passed on to all consumers. The data center only paid for the 20 megawatts used. The change means that data centers will bear a larger share of costs to serve them, regardless of their actual consumption.

The effectiveness of this approach is already evident in other jurisdictions. In Ohio, data center requests dropped from 30 gigawatts to 13 gigawatts after local regulators adopted rate provisions similar to Virginia’s.

Growing Concern Among American Consumers

This transformation of the electrical market is generating growing concern among American consumers. More Americans say that data centers have a negative rather than positive effect on the environment, domestic energy costs, and the quality of life of nearby residents. 39% believe data centers are mostly bad for the environment compared to 4% who view them favorably, 38% versus 6% on domestic energy costs.

This negative perception transcends traditional partisan divides. This situation creates interesting realignments and potential for bipartisan coalitions because it is not simply a left-right issue. Liberals worry for environmental reasons and distrust AI companies, but many conservatives are also unhappy with data centers. Being critical of data centers becomes a winning electoral issue for candidates.

The scale of the budgetary challenge explains this political convergence. Electricity costs have outpaced inflation, rising 42% since 2019, while the overall consumer price index increased only 29%. 68% of Americans report that their household finances are strained to some extent by domestic energy costs, and nearly a quarter (23%) say their finances are very strained.

The Structural Asymmetry of the Economic Model

This situation reveals a structural asymmetry in the economic model of digital infrastructure. Utilities build infrastructure, and we all pay for it because that’s how the utility economic model has always worked. New data centers can drive up residential electricity prices for two main reasons: they require the development of new generation capacity, and they require the construction of improvements for electricity transmission and distribution. The cost of this new infrastructure is passed on to consumers through rate changes and shifted from data center operators through special contracts or incentive packages.

Technology giants have recently attempted to address these concerns through voluntary commitments. Microsoft has outlined a general plan promising to cover its own electricity costs, reduce water consumption, create local jobs, and avoid seeking tax relief. Anthropic pledged to cover electricity price increases related to its data center development. Leaders of these companies went to the White House to sign a non-binding “taxpayer protection pledge.”

But these promises don’t address the structural problem. The commitment “does nothing to help consumers” because neither the president nor technology companies control who pays for electrical network extensions. These billion-dollar companies obtain personalized deals at reduced rates in secret, then bill ordinary customers for the difference and infrastructure through higher bills. Utilities earn guaranteed returns on new facilities and lines, while cloud and AI giants secure cheaper electricity, long-term capacity reservations, and priority network access.

This silent transformation of the American energy system illustrates an economic logic now familiar: privatization of profits and socialization of costs. Data centers represent critical infrastructure for the digital economy, but their financing model reproduces the asymmetries that already characterize other strategic sectors. Between the energy appetite of artificial intelligence and the budgetary constraints of American households, the economic balance of the electrical system is still seeking its new stability.

Sources

  1. Pew Research Center - US data centers’ energy use amid the artificial intelligence boom

  2. American Action Forum - Virginia’s New Data Center Electricity Rate Class

  3. Inside Climate News - Virginia Regulators Approve New Dominion Rates, Assign More Costs to Data Centers

  4. NBC News - Data centers and utility costs: The sleeper issue that could play a huge role in upcoming elections

  5. Consumer Reports - AI Data Centers: Big Tech’s Impact on Electric Bills, Water, and More

  6. Pew Research Center - US views of how data centers affect the environment, energy costs, jobs and more

  7. The Washington Post - Three Mile Island seeks taxpayer subsidies to reopen for Microsoft AI deal

  8. MIT Technology Review - Why Microsoft made a deal to help restart Three Mile Island

  9. CNBC - Trump administration backs Three Mile Island nuclear restart with $1 billion loan to Constellation

  10. Utility Dive - Constellation plans 2028 restart of Three Mile Island unit 1, spurred by Microsoft PPA