The relationship between data centers and the electrical grid is not simply a matter of one large customer drawing more power. It is a relationship that restructures how infrastructure is built, who pays for it, and how reliably power is delivered to everyone else on the same system. For residential customers in regions where data center development is concentrated, those restructuring effects are beginning to appear in monthly electricity bills — and are likely to intensify.
Understanding how this works requires understanding how the grid actually functions, how data centers connect to it, and how the costs of serving very large industrial customers flow through utility rate structures to ordinary households.
How the Grid Works and What Data Centers Require
The electrical grid is a system of generating stations, high-voltage transmission lines, lower-voltage distribution lines, and substations that step voltage up and down between these components. When a new large load — a factory, a data center, a large commercial building — connects to the grid, it must do so through an existing substation or through a new one built for the purpose. If the load is large enough, it may require new transmission lines as well.
A large data center operates at a scale that is qualitatively different from most other commercial electricity consumers. A large office building might draw a few megawatts. A large manufacturing plant might draw tens of megawatts. A hyperscaler data center campus can require 500 megawatts to 1 gigawatt or more of capacity — and unlike a factory that may operate at full load only during certain hours, a data center draws power continuously, at or near its contracted capacity, every hour of every day. This combination of scale and constancy makes data centers among the most demanding types of new grid interconnections.
Meeting that demand requires, in most cases, significant infrastructure investment: upgrades to existing transmission lines or construction of new ones, new substations, and potentially new or expanded generating capacity. These investments are capital projects that take years to build and cost tens to hundreds of millions of dollars.
Who Pays for the Infrastructure
When a utility builds new transmission infrastructure to serve a large customer, the cost of that infrastructure is recovered through rates — charges that are spread across the utility’s entire customer base, including residential ratepayers. The specific mechanism varies by state and by regulatory framework, but the fundamental dynamic is common: capital investments made to serve a particular customer or class of customers become part of the rate base that all customers pay.
This cost-spreading reflects the traditional logic of utility regulation: the grid is a shared public asset, and the costs of expanding it benefit the system as a whole by maintaining reliability and enabling economic growth. But the logic becomes strained when the infrastructure investment is driven by the specific demands of a small number of very large customers — particularly when those customers negotiate favorable rate structures, tax abatements, and other benefits that reduce their own contribution to the system costs they are generating.
In Northern Virginia, where more than 200 data centers host a disproportionate share of the world’s internet infrastructure, this dynamic has played out in visible and documented ways. Dominion Energy — the electric utility serving most of Virginia — has requested and received a series of rate increases in recent years, and the connection between those increases and data center growth has been a central issue in regulatory proceedings before the Virginia State Corporation Commission (SCC).
The Virginia Rate Cases
In November 2025, Virginia’s State Corporation Commission approved new Dominion Energy rates that added approximately $16 per month to the typical residential electricity bill in 2026. The SCC simultaneously created a new rate class — designated GS-5 — specifically for large customers using 25 megawatts or more, including AI data centers. Under this rate class, effective January 2027, large customers must pay a minimum of 85 percent of contracted distribution and transmission demand, and 60 percent of generation demand. The minimum bill structure is designed to ensure that data center operators pay for the grid capacity reserved for them even if they do not use all of it.
According to the American Action Forum’s analysis, the GS-5 rate class and the associated minimum bill requirements were specifically structured to protect residential ratepayers from bearing the full cost of grid upgrades driven by data center demand. The analysis found that the typical residential customer would nonetheless see monthly bill increases of $11.24 in 2026 and $2.36 in 2027 — increases attributed in significant part to the capital investment required to serve the data center load.
The Piedmont Environmental Council, a Virginia nonprofit, found in its analysis that the SCC’s rate structure would still place 61 percent of the costs of grid upgrades onto individual ratepayers after the 14-year minimum contract term ends — meaning that even the new protections are partial and time-limited.
Dominion Energy itself has disputed the characterization that data centers are raising residential bills. The company’s vice president of customer relations told WHRO that recent bill increases were due to weather and inflation, and that “data centers are paying their fair share.” Energy researchers interviewed by the same outlet disagreed. The dispute over whether data centers are paying their fair share of infrastructure costs — or whether residential customers are subsidizing the infrastructure that serves them — is not primarily a technical debate. It is a question about how the costs and benefits of data center development are distributed within a community.
The Sierra Club’s Virginia chapter estimated that nearly 1,300 data centers would be operating in Virginia within a few years, a figure that would make the grid investment requirements and corresponding rate pressures substantially larger than what has been experienced to date.
The PJM Interconnection Queue
Northern Virginia is part of the PJM Interconnection grid — the regional transmission organization that manages the electrical grid serving 65 million people across 13 states and the District of Columbia, from the mid-Atlantic through parts of the Midwest. PJM manages the queue of projects seeking to connect to the grid, and the state of that queue illustrates the scale of the problem nationally.
Grid interconnection queues have stretched to five to seven year average wait times in major markets, driven in significant part by the volume of data center interconnection requests. The backlog reflects not just high demand but infrastructure constraints: the transmission lines, substations, and transformers needed to support new interconnections take years to design, permit, and build. Transformer shortages have created multi-year equipment backlogs that further slow the process.
FERC has been actively engaged in reforming the interconnection process to address these backlogs, with proposals related to co-located data centers — facilities that seek to connect directly to power plants rather than through the grid — and to the rules governing very large loads that bring their own generation. These proceedings are technical and ongoing, but their outcomes will shape how the costs of data center expansion are allocated between operators and ratepayers for the foreseeable future.
Reliability and Peak Demand
Beyond the question of who pays for infrastructure, there is a question of how data center growth affects grid reliability for everyone else. Data centers are among the most reliable power consumers — they are designed to draw power continuously and to have backup generation for when the grid fails them. But when many large data centers are concentrated in a single region, and when that region’s transmission infrastructure is already strained, the aggregate demand creates reliability concerns during peak periods.
PJM has been grappling with capacity adequacy — the question of whether the grid has enough firm generating capacity to meet peak demand — in part because of the rapid growth of data center load. The regional grid operator has reported a projected shortfall in firm capacity, and has proposed that large loads that cannot secure sufficient new generation will be subject to curtailment protocols — meaning that during grid emergencies, data centers may be required to switch to on-site backup generators to relieve pressure on the system. The backup generators that data centers use are typically diesel-powered, introducing air quality implications for neighboring communities during periods when they are activated.
The Decarbonization Complication
The relationship between data center growth and electricity decarbonization is more complicated than it is typically presented.
Major data center operators — Amazon, Microsoft, Google, Meta — have made prominent commitments to power their operations with renewable energy. They accomplish this primarily through power purchase agreements (PPAs): contracts with renewable energy developers that allow them to claim credit for wind or solar generation that is added to the grid, even when their facilities are physically drawing power from whatever is available on the grid at the time. These commitments are real in the sense that they drive investment in new renewable generation. They are misleading in the sense that they do not ensure that the specific electrons powering a data center are from renewable sources at any given moment.
The practical grid-level dynamic is that data center demand growth is outpacing renewable energy additions in some regions, meaning that new data center load — particularly load that comes online quickly in response to AI demand — is frequently served by whatever generating capacity is available on the grid, including natural gas and coal. The new renewable generation that operators are contracting for takes years to build; the data centers are being built faster.
Lawrence Berkeley National Laboratory’s 2024 report projected that data center electricity consumption could reach between 325 and 580 terawatt-hours annually by 2028. That range is wide because it depends heavily on how AI demand develops, but even the lower bound represents a significant expansion of a demand category that is growing faster than the renewable energy supply that operators are claiming to offset it.
This gap has implications for communities beyond electricity bills. New fossil generation built to serve data center demand creates air quality impacts. Decisions about which generating facilities to retire versus keep running depend on load projections that data centers are pushing upward. These are choices with real consequences for public health and climate that tend not to appear in the local permitting process for a data center.
Participating in Utility Commission Proceedings
State public utility commission proceedings — rate cases, integrated resource plans, and interconnection-related filings — are the primary venue in which data center impacts on electricity rates and grid planning are formally addressed. These proceedings are public, but they are also technical, lengthy, and require sustained engagement to participate meaningfully.
Most state utility commissions allow any person to intervene in a rate case and file comments or testimony. Environmental and consumer advocacy organizations in many states actively participate in these proceedings and have developed expertise in the data center-related rate issues. For residents who want to engage, the most practical approach is to connect with those organizations rather than to navigate the proceedings independently.
State utility commission websites publish pending cases and filing deadlines. Most require that intervenors file a notice of intervention within a specified period after a case is opened — missing that window forecloses formal party status. Even without formal intervention, written comments from members of the public are typically accepted in rate proceedings and included in the record.
The question of what residential customers pay for the grid infrastructure that serves data centers is ultimately a question that regulators decide. Communities that are not represented in those proceedings cede that decision entirely to the utilities and the large industrial customers who are already well-organized to participate.
Inside Climate News: Virginia Regulators Approve New Dominion Rates | WHRO: Dominion Energy and Data Center Rate Debate | American Action Forum: Virginia Data Center Electricity Rate Class | Lawrence Berkeley National Laboratory: Data Center Energy Report | Utility Dive: FERC Data Center Interconnection Reform
This article was researched and drafted with AI assistance under human review. See our full AI and editorial practices.