Electricity Demand Has Grown 4-8x Faster in Data-Center-Concentrated States/Counties Than the US Average Since 2019

Summary

Electricity demand has grown far faster in the counties/states where data centers are concentrating than in the country as a whole. The clearest case is Northern Virginia (Loudoun, Fairfax, Prince William counties), the world's largest data-center market: Virginia's statewide commercial electricity sales grew 45.8% from 2019 to 2024 (EIA-861), roughly 7x the 6.6% national average over the same period. In Loudoun County specifically, data centers already consumed about 24% of all county electricity in 2023, more than residential use (18%). At the grid-operator level, PJM's Dominion Virginia zone — the transmission zone serving Data Center Alley — saw summer peak load rise 23% and winter peak load rise 45% between 2019 and 2025, and PJM projects further peak growth averaging 5.4%/year through the mid-2030s, driven overwhelmingly by data-center load. Other states with major data-center clusters (Oregon, Wyoming, Nebraska) show similarly outsized commercial-sales growth (36%-50% over 2019-2024), while a state with a smaller, single-county cluster (Missouri/Platte County) shows essentially flat statewide commercial sales, since one county's load is too small to move a state total.

Which counties are 'data-center concentrated'

Per a Climate Crisis 24/7 analysis of U.S. Department of Energy (OSTI) known data-center locations (as of September 2025, ranked per 100,000 residents), the most concentrated counties include Morrow County, OR; Crook County, OR; Loudoun County, VA (46.6 known centers per 100k, 199 total); Grant County, WA; Umatilla County, OR; Douglas County, WA; Pottawattamie County, IA; Laramie County, WY; Prince William County, VA; and Sarpy County, NE. Northern Virginia (Loudoun/Fairfax/Prince William) is the largest cluster by absolute count, with nearly 300 facilities and roughly 70% of world internet traffic passing through the region.

Quantifying the demand growth

No county-grain EIA electricity-sales dataset exists in the AskAmerica connector (energy.eia_electricity_prices and energy.eia_electricity_generation are both state-grain), so state-level commercial electricity sales are used as the best available proxy, cross-checked against grid-operator (PJM) and news-reported county-specific figures where available. Commercial-sector sales (EIA-861, million kWh) 2019 vs 2024:

The pattern is consistent: every state whose data-center cluster is large relative to the state's overall economy (VA, OR, WY, NE) shows commercial electricity growth 4x-8x the national rate. States where the cluster is one county among many larger population/industrial centers (IA, MO, NM) show growth at or below the national average — the county-level effect is real but gets diluted away at state grain.

County- and grid-level corroboration

Because no county-grain EIA sales table exists, the following county/grid-specific figures come from primary reporting and grid-operator forecasts rather than the AskAmerica connector, and are reported as such:

• Loudoun County, VA: data centers accounted for about 24% of total county electricity consumption in 2023, versus 18% for residential use (Loudoun County government / EIA commentary). Dominion Energy's demand forecast for the region rose 17% (adding ~7 GW) in a single forecast revision cycle (Dec 2024-2025).
• PJM Dominion zone (the transmission zone covering Northern Virginia's data-center cluster): summer peak load rose from roughly 19,435 MW in 2019 to 23,905 MW in 2025 (+23%); winter peak load rose from roughly 17,526 MW in 2019-20 to 25,413 MW in 2025-26 (+45%). All but 3 of the top 50 peak-hourly-load records in this zone's history occurred in 2024 or 2025. PJM's 2026 20-year forecast projects Dominion-zone peak demand growing ~5.4%/year through 2036, the largest absolute increase of any PJM zone, driven by data-center load.
• Northern Virginia region-wide: power demand is projected to nearly quadruple from about 4 GW to 15 GW by 2030 in some industry estimates, and Loudoun County alone could require 20-30 GW by 2030 under aggressive AI-driven demand scenarios (industry/utility filings), though these are forward projections, not measured outcomes.

Method, caveats, and what was and was not computed here

Computed directly from AskAmerica/EIA-861 data (energy.eia_electricity_prices): the 2019-2024 percent growth in commercial-sector electricity sales for each named state, and the national commercial-sales total for the same window (6.6% growth). These are exact, reproducible SQL aggregates.

Passed through from primary/secondary sources, not computed here: the Loudoun County 24%-of-load figure, the PJM Dominion-zone peak-load percentages (backed out from EIA's own May 2026 'Today in Energy' article, which reports the percentage changes directly rather than the underlying 2019 absolute MW, so the 2019 figures in the chart are derived by reversing the reported 23%/45% growth), the DOE/OSTI-derived county rankings, and the forward-looking GW projections.

Key limitation: no county-grain electricity-sales or generation table exists in this connector (energy.eia_electricity_generation and energy.eia_electricity_prices are both state x sector x year grain). State-level commercial sales is a reasonable proxy where the data-center cluster is large relative to the state (Virginia, Oregon, Wyoming, Nebraska) but understates or masks the true county-level effect in larger, more diversified states (Texas, Iowa, Missouri, Ohio, Washington) where a single county's hyperscale campus is a small share of a much larger statewide commercial base. The Missouri case (-1.9% statewide despite an active Platte County cluster) is the clearest illustration of this dilution.