Electricity bills are rising across much of the country. In many regions, AI data centers are a significant — and underreported — driver of that increase.
The energy demands of AI infrastructure have crossed from "tech industry story" to "utility bill reality" — and the county-level data makes the mechanism visible in a way that national headlines rarely capture.
The scale, in numbers
The United States has over 4,500 active data centers. More than 700 are currently under construction across 38 states.
Together, they consume approximately 176 terawatt-hours of electricity per year — about 4.4% of all U.S. electricity production. For context, that's more than the entire electricity consumption of many mid-sized countries.
And that number is about to get much larger. A 2026 analysis projects that U.S. data center electricity demand will nearly double between now and 2028 — from roughly 80 gigawatts to 150 gigawatts — driven primarily by AI model training and inference, which is computationally intensive in ways that earlier generations of computing simply weren't.
Loudoun County, Virginia: the case study
The clearest way to understand what this looks like at the county level is to look at Loudoun County, Virginia — the place we've been calling the data center capital of the world, because it is.
Loudoun has 199 active data centers, with another 117 in active development. The county's data centers already consume more than twice the output of Virginia's largest nuclear power plant. Satisfying the demand from facilities currently planned or under construction in Loudoun alone would require the equivalent of several additional nuclear plants to be added to the regional grid by the end of the decade.
Here's where it becomes a utility bill story: Dominion Energy, the primary electricity provider for the region, has proposed a 14% rate increase for residential customers in 2026, citing data center growth and AI-driven demand as primary factors.
Looking further out, Dominion's own integrated resource plan projects that if data centers don't pay for their proportional share of infrastructure costs, monthly residential bills in the region will rise from an average of $159 today to between $255 and $308 by 2035 — and up to $381 by 2045.
It's not just Virginia
The same dynamic is playing out — at different speeds — across every region with significant data center concentration.
In the PJM electricity market, which covers everything from Illinois to North Carolina, data centers were responsible for an estimated $9.3 billion increase in the 2025–26 capacity market. That cost gets distributed across all ratepayers in the region. For the average residential customer in western Maryland, it translated to roughly $18 more per month. In Ohio, about $16 more per month.
In Texas, ERCOT — the state's independent grid operator — projects that North Texas wholesale electricity rates could rise by 79% in 2027, driven significantly by new data center load.
Across the country, areas with the highest concentrations of data centers have seen electricity prices jump an average of 267% over the past five years. More load on the same grid means higher capacity costs, and those costs get distributed across all ratepayers in the region.
Why this is a county-level story
National averages hide what's actually happening. Data center load is not evenly distributed — it's highly concentrated in specific counties, and the rate effects radiate outward from those concentrations through regional grid markets.
The counties that are most exposed are those that:
- Host significant data center capacity directly (the Loudoun Countys of the world)
- Share a regional grid market with high data center concentration (most of the Mid-Atlantic and parts of the Midwest and South)
- Are downstream from power plants that have increased output to serve new data center load, absorbing the associated air quality and infrastructure costs
FERC market data, EIA electricity statistics, county-level economic data from the BEA, and utility rate filings are all publicly available. What most residents in affected counties have never seen is those sources assembled into a coherent picture of where the load is concentrated, how it moves through regional grid markets, and what the projected rate trajectory looks like for their specific location.
The policy response (and what it tells you)
In November 2025, Virginia's State Corporation Commission approved a new electricity rate class specifically for large-scale AI data center customers. Starting in January 2027, affected facilities must pay for at least 85% of their contracted distribution and transmission demand, and 60% of generation demand.
That policy change — and the fact that it took regulatory action to get there — tells you something about the default: without intervention, the infrastructure costs of data center growth get distributed broadly across all ratepayers, including residential customers who have no connection to the AI industry.
Other states are watching Virginia's approach closely, and several are exploring similar frameworks. Whether those policies arrive before or after the rate increases do will say a lot about how well the regulatory environment keeps pace with the speed of AI infrastructure build-out.
What this data can tell you
If you're a resident, local official, utility regulator, or economic development professional, the county-level data on data center density, regional grid load, and projected rate trajectories is exactly the kind of information that should be in the room when decisions get made.
Data centers bring real economic benefits — jobs, property taxes, local spending — and there are legitimate reasons communities pursue them. The harder question is whether residents and local officials have access to the same information that utilities and developers already have when those negotiations happen, because that information asymmetry is often where outcomes get determined.
That's the kind of analysis our pipeline is built for. If you want to understand the energy picture for a specific county or region, reach out.
Sources: Electric Choice (2026 data center power consumption analysis); Consumer Reports AI data center investigation; PJM capacity market filings; Dominion Energy 2025 Integrated Resource Plan; ERCOT 2026 summer reliability assessment; Virginia State Corporation Commission docket; Al Jazeera energy reporting, June–July 2026.