Artificial intelligence is often presented as a digital revolution, but its physical footprint is becoming increasingly visible in electricity bills, transmission projects, and local planning debates. As large technology firms expand data centers to support AI computing, Virginia has become a central test case for a wider national question: when new grid infrastructure is built primarily for exceptionally large customers, should ordinary households help finance it?
A Digital Industry With Physical Demands
Data centers are essential to cloud computing, online services, and AI systems, yet they require substantial and continuous electricity supplies. In high growth states, utilities must add generation capacity, substations, transmission lines, and distribution equipment to accommodate this demand. Those investments can strengthen the grid overall, but they can also be prompted by a small number of very large facilities.
Virginia is especially significant because it hosts one of the world’s largest concentrations of data centers. The rapid expansion of these facilities has intensified scrutiny of how Dominion Energy allocates the costs of maintaining and expanding the electricity system. The issue is not simply whether data centers consume more power. It is whether the pricing rules used by regulated utilities accurately distinguish between investments that benefit all customers and investments required mainly to connect and serve particular large users.
This distinction matters because electricity systems have traditionally spread many infrastructure costs across broad customer groups. That approach can be reasonable when upgrades provide widely shared reliability benefits. However, it becomes more contentious when a dedicated line, substation, or capacity expansion is closely tied to one customer’s projected demand.
Virginia Begins to Redraw the Rules
Virginia regulators have already started changing the framework. The State Corporation Commission approved a separate Dominion rate category, known as GS 5, for customers with demand of at least 25 megawatts and a load factor of at least 75 percent. Beginning in 2027, qualifying customers will face contracts that require them to pay for at least 85 percent of their contracted transmission and distribution demand, as well as at least 60 percent of their contracted generation demand.
The policy reflects an effort to make large customers bear a greater share of the infrastructure costs associated with their electricity needs. It also seeks to reduce the risk that a utility builds expensive facilities for anticipated demand that later fails to materialize. Requiring collateral payments and long term commitments gives utilities greater assurance that major customers will contribute even if their business plans change.
Yet the reform does not mean that every electricity cost connected to data center growth can be assigned neatly to one industry. Grid upgrades often have multiple purposes, including improving reliability, serving future residential development, and integrating new energy resources. Regulators therefore face a practical challenge: allocating costs too broadly may burden households, while allocating them too narrowly could discourage investment or overlook shared system benefits.
The Household Ratepayer Question
The debate has become more immediate as Virginia households face rising electricity charges. The commission’s earlier base rate decision increased the typical monthly residential bill by about $11.24 in 2026, with a further $2.36 increase in 2027. Those increases were linked to broader pressures, including the rising cost of grid equipment and utility operations, rather than being identified as a data center specific charge.
In August, however, the commission directed Dominion to develop tariffs that would assign the full cost of transmission infrastructure built exclusively for data centers and other large load customers directly to those users. Dominion had initially sought to recover approximately $998.5 million in transmission costs through a surcharge that could have added $2.90 to a benchmark residential bill. The regulatory direction suggests that Virginia is moving toward a more targeted principle: if infrastructure serves a specific large facility, the facility should be the primary payer.
Supporters of this approach argue that it better aligns costs with the customers creating them and can protect household affordability. Large technology companies, meanwhile, may argue that predictable electricity pricing is necessary for investment, job creation, tax revenue, and continued digital innovation. From that perspective, sharply higher charges could affect where future projects are located and how quickly new capacity is developed.
A National Policy Test
Virginia’s experience is likely to influence debates in Texas, Arizona, and other states attracting AI related investment. Texas officials have also considered whether data centers should fully fund the transmission and distribution upgrades needed for their operations, rather than shifting those costs to residential customers. Arizona has faced similar concerns as utilities seek investment for rapid load growth and customers question how new capacity will affect rates.
The broader implication is that AI policy is no longer limited to software regulation, competition, or labor markets. It now includes utility governance. State regulators must decide how to preserve reliable service, attract investment, and ensure that the financial consequences of rapid industrial growth are distributed transparently.
A Final Note
Households need not automatically absorb the infrastructure costs created by large technology firms, but neither can policymakers assume every grid investment serves only one customer. Virginia’s emerging model points toward a balanced standard: assign dedicated costs directly to the users that cause them, while sharing only those investments that demonstrably benefit the wider electricity system.
