American households are absorbing rising electricity costs driven in part by the rapid expansion of AI data centers, and regulators at the federal and state level are now moving to shield residential ratepayers from billions of dollars in new infrastructure spending. The Federal Energy Regulatory Commission has launched a proceeding requiring grid operators to justify how they charge large new loads for interconnection, while Virginia’s State Corporation Commission has created a dedicated rate class for data centers effective January 1, 2027, with minimum payment rules designed to stop cost shifting. The U.S. Energy Information Administration, meanwhile, has modeled scenarios showing that faster-than-expected data center demand growth could push wholesale electricity prices higher in the PJM Interconnection region by 2027.
How data center load growth is reaching household electric bills
The connection between a hyperscale server farm and a family’s monthly utility statement runs through shared transmission lines, generation capacity, and the rate cases that divide those costs among all customers. When a utility needs to build new power plants or upgrade wires to serve a single large customer, the spending often lands in a rate base that every ratepayer funds. That dynamic is accelerating. The EIA’s Short-Term Energy Outlook identifies commercial-sector demand as a key driver of rising national electricity use, with data centers accounting for a significant share of that commercial load.
The tension is sharpest in regions where data center construction has outpaced grid planning. PJM, the grid operator covering 13 states and the District of Columbia, faces an outsized concentration of new facility requests in northern Virginia and surrounding corridors. An EIA analysis found that under a faster data center demand scenario, modeled 2027 wholesale prices in PJM would rise compared with the baseline, accompanied by increased fossil-fuel generation to meet the load. Higher wholesale prices flow downstream: utilities pass fuel and purchased-power costs to customers through adjustment clauses that often take effect before a full rate case is even filed.
That pass-through happens in several steps. First, utilities pay more for power in wholesale markets when demand spikes or when new load requires running more expensive plants. Second, fuel and purchased-power riders on customer bills automatically adjust to recover those higher costs, sometimes on a monthly or quarterly basis. Only later do regulators examine whether new long-lived investments-such as gas-fired plants, transmission lines, or substations built largely for data centers-should be added to the rate base that all customers pay over decades. By the time that question reaches a hearing room, much of the near-term cost has already shown up on household statements.
The hypothesis that utilities in high-data-center-growth states will soon file rate cases explicitly quantifying data-center-driven costs, and that those filings will produce measurably larger residential rate increases than in states with little data center activity, is already being tested. Virginia’s regulatory action, discussed below, represents the first clear attempt by a state commission to isolate those costs before they spread across all customer classes. Whether other states follow within the next 18 months will determine how unevenly the AI boom’s electricity costs land on American households.
Federal and Virginia regulators act to isolate data center costs
FERC took direct aim at the cost-shifting risk when it launched targeted action to speed large-load integration while requiring grid operators to justify or reform their tariff rules for large-load interconnections. The proceeding puts the burden on regional transmission organizations to demonstrate that existing rules do not force residential and small-business customers to subsidize infrastructure built primarily for data centers. Among other issues, FERC is asking how quickly large loads should be studied, how interconnection costs should be assigned, and whether current practices unfairly socialize upgrades that have limited benefits beyond the new customer.
FERC’s order also opened a related inquiry in PJM’s market framework, which includes docket EL25-49-000 and specifically addresses co-located load arrangements in the region. In these setups, some data centers seek to connect directly to power plants or on-site generation while still relying on the broader grid as backup. Regulators worry that if those customers avoid paying for transmission while retaining access to its reliability benefits, the remaining costs will be pushed onto households and small businesses that have no comparable option to self-supply power.
Virginia moved even more concretely. The State Corporation Commission issued an order in the Dominion Energy Virginia biennial review case PUR-2025-00058 that created a new large-load rate class effective January 1, 2027. The order established minimum payment requirements for large-scale energy users, a category that captures data centers drawing hundreds of megawatts. The intent is explicit: prevent the cost of new generation and transmission capacity from being socialized across Dominion’s roughly 2.7 million residential and small-commercial accounts, especially when those assets are sized and sited primarily to serve clusters of data facilities.
Under the new structure, qualifying customers will be grouped into a separate class whose rates can be designed around their unique usage patterns-high load factors, 24/7 demand, and concentrated geographic footprints. Regulators signaled that they expect these users to shoulder a larger share of the fixed costs associated with serving them, including dedicated substations and high-voltage lines. While the details of Dominion’s eventual tariffs remain to be finalized, the framework gives the commission a clearer tool to prevent cross-subsidies as new projects seek interconnection.
These two actions, one federal and one state, represent a regulatory bet that separating data center costs from general rate bases will protect household bills. But neither proceeding has yet produced final tariff language or a public cost-allocation formula. FERC’s docket requires compliance filings from grid operators, and Virginia’s new rate class does not take effect for months. In the interim, utilities can still file for rate increases that blend data-center-driven spending with ordinary system costs, leaving residential customers to absorb a share.
Gaps in the evidence and what ratepayers should watch next
Several pieces of the puzzle are still missing. The EIA’s modeling of higher PJM wholesale prices under faster data center growth does not break out state-level or utility-specific residential bill impacts. That means no federal dataset yet shows exactly how many dollars per month a typical household in Virginia, Ohio, or Georgia will pay because of data center expansion. The Virginia commission’s order, while explicit about its intent to prevent cost shifting, likewise stops short of quantifying how much more Dominion’s residential customers would have paid without a separate large-load class.
That evidentiary gap matters because it shapes the political appetite for similar reforms elsewhere. State regulators in regions courting data center investment face competing pressures: economic development agencies tout tax revenues and construction jobs, while consumer advocates warn about long-term bill impacts. Without clear, comparable numbers showing how much data center-related infrastructure is being rolled into residential rates, each new project is debated in isolation rather than as part of a broader cost trend.
For now, ratepayers and local officials can watch several indicators. One is the language utilities use in upcoming rate cases and integrated resource plans, particularly any references to “large load additions,” “hyperscale customers,” or “compute campuses” as drivers of new capacity. Another is the emergence of special contracts or riders tailored to individual data centers, which can reveal whether those customers are being asked to guarantee revenue sufficient to cover the assets built to serve them.
Households can also track how quickly fuel and purchased-power adjustment clauses are rising relative to base rates. Because wholesale price impacts from data center growth show up first in these pass-through mechanisms, sudden increases may signal that new load is tightening supply-demand balances even before major new plants come online. Consumer advocates, for their part, are likely to scrutinize whether utilities propose non-bypassable charges that apply to all customers, including those who use less power or invest in rooftop solar, to recover data-center-related costs.
The next two years will test whether the tools FERC and Virginia have put on the table are sufficient. If grid operators respond to the federal proceeding with tariffs that clearly assign most upgrade costs to the data centers that trigger them, and if Virginia’s large-load class successfully contains Dominion’s residential increases, other states may adopt similar structures. If not, pressure will grow for more aggressive measures, such as moratoria on new interconnections in constrained areas or requirements that data centers procure dedicated clean resources rather than leaning on existing fleets.
Either way, the core policy question is no longer whether AI and cloud computing will reshape the power system-they already are-but who will pay for that transformation. The answer will show up, line by line, on the electric bills of tens of millions of American households.
More from Morning Overview
*This article was researched with the help of AI, with human editors creating the final content.
