Do Data Centers Raise Electricity Bills? Cost Shifting, Large-Load Tariffs and the Studies

A regulated utility recovers its costs from customer classes in proportion to the costs each class is judged to cause. When a data center connects, it pays for its own service under a tariff. The question is whether that tariff covers the full cost of the generation, transmission and distribution built or bought to serve it. If it does not, the gap is spread across everyone else. If the data center pays more than its incremental cost, it can lower other customers’ rates by spreading fixed costs over more sales.

Two features of data center growth make cost allocation harder than usual. The loads are very large relative to existing systems, so a single customer class can drive billions of dollars of investment. And the forecasts are uncertain: utilities build for requested load that may arrive late or not at all. The ratepayer risk is that the investment is made and recovered from everyone while the load that justified it does not materialize. How utilities process those requests is covered in our guide to the large-load interconnection process.

The most direct, measured effect of data centers on bills so far has come through PJM’s capacity market, where the price of having enough supply for future peaks is set years ahead and passed through to customers across the PJM region. Forecast data center load raises the amount of capacity PJM must buy, and the price rises with it.

Monitoring Analytics, PJM’s independent market monitor, estimated that existing and forecast data center load accounted for $6.3 billion, or 38.2%, of the capacity market charges in the 2028/2029 base residual auction, and $29.4 billion, or 46.2%, across the last four base auctions.¹ That auction cleared at the FERC-approved price cap of $325 per MW-day.¹¹ The monitor has proposed that data center load be procured through a separate mechanism so that data centers pay for their own capacity.¹

E3 offers a different allocation of the same price increase. Its 2026 analysis for the Data Center Coalition attributed roughly half of PJM’s sharp capacity price increase to load growth and the rest to market design changes, plant retirements, reduced accreditation of some resources and other supply factors.² Both views agree that forecast load matters; they differ on how much of the price increase it explains. Our guide to PJM capacity prices and data centers covers the auction mechanics.

The research splits along a line between backward-looking studies of rates and forward-looking analysis of how rates are set.

Harvard Electricity Law Initiative (2025)

Eliza Martin and Ari Peskoe’s paper, “Extracting Profits from the Public,” argues that because utilities earn returns on infrastructure, serving data centers is lucrative, which gives them an incentive to offer attractive rates to large technology companies.⁶ It identifies special contracts, reviewed by state regulators through what the authors call opaque processes, as one channel by which costs move to the public.⁶ Its recommendations include requiring data centers to take service under standard tariffs instead of special contracts, more frequent demand forecasts and connection conditioned on flexible operation.⁶

E3 for the Data Center Coalition (2026)

E3 reviewed 11 recent quantitative studies, interviewed four experts and analyzed PJM auction results. It found no evidence that data centers have been historically subsidized by other customers under existing rate structures, noted that Texas and Virginia had large load increases and among the smallest rate increases, and listed inflation, natural gas prices, resilience spending and grid modernization among the main drivers of recent rate increases.² The study was funded by the Data Center Coalition and authored by E3.⁷ It also said regular updates to cost allocation will be critical to continue preventing cost shifts.²

Lawrence Berkeley National Laboratory and Brattle (2025–2026)

The LBNL and Brattle study of retail prices found that, adjusted for inflation, average U.S. retail electricity prices in 2025 were about 3% higher than in 2019 and 6% lower than in 2010.³ It found that load growth has tended to depress retail prices in recent years because fixed costs are spread over more sales, with the highest-growth states seeing real declines, while cautioning that this may not hold in the future.³

Forward-looking projections are where concern is concentrated. Virginia’s JLARC found that a typical residential Dominion Energy customer could see monthly costs rise by $14 to $37 by 2040, as the utility builds generation and transmission for a statewide demand forecast driven mainly by data centers.⁸ E3 notes that JLARC found no evidence of a present-day cost shift to residential customers; the projected increase reflects the cost of new infrastructure rather than a subsidy today.²

Modeling by Carnegie Mellon University and North Carolina State University through the Open Energy Outlook projected that data center and cryptocurrency mining growth could raise average U.S. electricity generation costs by about 8% by 2030, with increases above 25% in some regional markets, including parts of Virginia.¹² Those are scenario estimates of generation cost, not forecasts of any one utility’s rates.

Utilities often argue the opposite case for their own systems. A Georgia Power executive told state legislators that adding 3,300 MW of generation for data centers would reduce customers’ bills, and Dominion has described Virginia rate-setting as open and transparent.¹³ The honest summary is that the answer is utility-specific and depends on whether new load pays more than the cost of serving it.

Regulators’ main tool is a large-load tariff: a rate class or rule set for customers above a size threshold that requires them to commit to and pay for the capacity they request. By one investment manager’s 2026 count, 24 states had approved at least one large-load tariff, with more proposals expected; counts vary by source and change month to month.⁴ The contract terms themselves are covered in our guide to large-load tariffs and electric service agreements.

The Public Utilities Commission of Ohio framed its decision in cost-causation terms, finding that the approved approach would ensure infrastructure costs are borne by those who incur them rather than shifted to other ratepayers.⁵ Texas took a statutory route with Senate Bill 6, described in our guide to Texas SB 6 and large loads.

The ratepayer debate has moved from public comment to tariff terms, and those terms are now site-selection variables. A site served by a utility with a strict large-load tariff will require more collateral and a firmer commitment to contract demand, but it may also face less political risk later. A site in a jurisdiction with no rules yet may face them mid-project.

• Ask the utility whether a large-load tariff applies, at what threshold, and with what minimum bill, term and exit fee.⁹¹⁰
• Check whether the project would sit in a capacity market region where forecast load affects prices for everyone.¹
• Find out whether the state commission has open proceedings on data center cost allocation, since terms can change before you sign.⁴
• Consider whether flexible operation or on-site supply could reduce the capacity you must reserve; see our guide to flexible data center loads.
• Expect local questions about bills during zoning; they now come up alongside the economic impact case.

Utility and commission staff can confirm which rules will apply to a specific load. If you want an independent read on a parcel’s power path and tariff exposure, you can get a site reviewed.