Do data centers raise residential electricity prices? What the evidence shows

A measurable correlation in Ireland

We paired CSO metered data center electricity consumption with Eurostat residential electricity prices for the same twelve half-years, from the first half of 2020 through the second half of 2025. Consumption measures electricity actually used by operating facilities, rather than announced projects or construction spending.

Data center consumption rose from 1,413 to 3,947 GWh per half-year. Household prices rose from 24.09 to 40.42 euro cents per kWh. Across the matched observations, Pearson’s correlation coefficient is r = +0.67: periods with greater data center consumption generally had higher household prices.

That result does not mean data centers caused 67% of a price increase. It measures how the two series move together. Both also trend upward over time, which can make a correlation in levels look stronger than the relationship between changes.

As a sensitivity check, we compared the change in consumption with the change in price between successive half-years. The correlation weakens to r = +0.22. One revealing counterexample: between the second half of 2023 and the first half of 2024, data center demand increased from 3,288 to 3,351 GWh while the household price fell from 37.94 to 25.90 cents.

Other forces were operating during this period. The European Commission’s energy price analysis documents the energy crisis and differences in how wholesale costs reach households. CSO background notes describe Irish electricity credits and VAT changes. Fuel costs, subsidies and taxes can move the observed household price independently of data center demand.

This is a descriptive national case study of operations, not a test of individual opening dates. It cannot determine how much of Ireland’s price change came from data centers, or whether the same relationship holds in another country.

U.S. studies connect facility growth to prices—but disagree

Research that matches facility growth to utility territories or state electricity markets goes further than our simple correlation. Its conclusions still depend on the comparison groups and assumptions used to distinguish data center growth from other changes.

These are working papers and a preprint, presented here as provisional research rather than settled peer-reviewed consensus. Their estimates should not be averaged: they use different units, periods and strategies. A state average can also conceal a price increase within a particular utility territory.

Why a large new load can raise—or reduce—costs

A household’s price response depends on what the power system must build or buy, and how those expenses enter tariffs. More electricity sales can spread existing fixed costs. If spare capacity is limited, meeting a large new load can instead require additional generation, transmission or costly purchases. The U.S. studies above illustrate these competing mechanisms.

Construction, connection and operation are different events. A facility can be announced years before consuming substantial power; utilities may begin investment sooner, and residential tariffs may change later. Plotting a construction marker beside a rising bill therefore does not isolate the facility’s contribution.

Virginia illustrates the prospective risk

Virginia’s JLARC data center review projected additional generation and transmission costs of $14–$37 per month by 2040 for a typical Dominion residential customer, in constant dollars, under its demand scenarios. The review also found that data centers were paying their allocated cost of service at the time of its 2024 analysis.

The dollar range is a modeled future cost change, not an already-observed increase on today’s bills. A projection of future system expansion and a finding about existing cost allocation answer different questions.

What to examine in your own community

Start with your utility’s service territory and rate case, rather than the nearest facility marker. Look for verified operational dates, actual demand, proposed grid investments, and rules assigning those costs to large-load customers or other ratepayers. Compare household prices with similar utilities over the same period, allowing for fuel prices, weather, taxes and policy changes. A county boundary alone rarely supplies that evidence.

Our dashboard’s nine documented campuses are illustrative examples, not a complete census suitable for a national causal analysis. Explore the dashboard, or view the source histories for Ireland and Virginia.