Data Centers Pushed Electricity Prices Down From 2015–2024, But the Next Chapter Is Murkier

For years, the political story around data centers has been simple: they guzzle power, they crowd local grids, and ordinary ratepayers foot the bill. A working paper now complicates that narrative sharply — finding that every 10% increase in data center capacity caused average residential electricity prices to fall by roughly 0.4%, meaning the data center boom of 2019 to 2024 likely cut the average American household's electricity rate by about 6%.

The finding comes from a working paper posted to arXiv that uses an instrumental variables method to isolate causation rather than correlation. The researchers used the total length of the 1947 Eisenhower Interstate Highway Plan within each state as their instrument — arguing it is unrelated to electricity prices but predicts where data centers eventually clustered — allowing them to filter out the confounding factors that plague simpler comparisons.

We estimate that data centers caused average retail electricity rates to fall modestly in the United States from 2015 to 2024 using an instrumental variables approach. Despite prevailing sentiment, the finding is consistent with economic reasoning: existing large power system fixed costs, economies of scale in transmission and distribution, and declining unit costs for generation imply that durable demand growth lowers average prices.— arXiv study authors

The mechanism is counterintuitive but grounded in how electricity is actually priced. Unlike most commodity markets, retail electricity rates are set partly on average costs, not just marginal costs. Because the grid carries enormous fixed costs — transmission lines, distribution infrastructure, generation capacity — adding a large, steady new customer like a data center spreads those fixed costs across more kilowatt-hours, pulling the per-unit price down for everyone else. The study finds evidence of these economies of scale across transmission, distribution, and generation, and across both residential and commercial customer classes.

That historical picture, however, is already straining against present-day reality. Data centers consumed about 4.5% of US electricity in 2024, more than double their share five years earlier, according to the Electric Power Research Institute. The same arXiv paper projects that share could reach 9–17% by 2030 — a pace of growth that the authors themselves warn could reverse the price-lowering effect if supply cannot keep up.

There are early signs that reversal is underway in some regions. The grid operator PJM, which covers much of the mid-Atlantic and Midwest, held a capacity auction on July 14, 2026 for the 2028–2029 delivery year that cleared at $325 per megawatt-day and still fell 6,831 MW short of its reliability requirement, according to costofowning.com. PJM stated that 'demand for electricity continues to grow faster than electricity supply' and pointed to continued large data center load additions. Those capacity costs flow into household bills in the region, though the share any given customer pays depends on their utility.

Nationally, residential electricity averaged 18.31 cents per kilowatt-hour in July 2026, up 4.9% from a year earlier, according to the EIA. The Bureau of Labor Statistics put its electricity price index up 3.8% for the 12 months ending August 2026 — both measures running ahead of the 3.4% rise in the overall consumer price index over the same period, according to costofowning.com.

The arXiv study notes that recent price increases have not been uniform. States where load grew tended to see below-average price increases, while the sharpest rises were concentrated in the Northeast and West Coast. California's average prices rose nearly 40% from 2021 to 2024, largely due to wildfire costs rather than data center load. Virginia, the national leader in data center density — where they consume over 20% of the state's electricity — showed rate increases close to the national average.

The question of who pays for the new grid infrastructure data centers require is increasingly a legislative one. The House passed the Ratepayer Protection Act 417 to 3 on September 16, 2026, according to costofowning.com. The bill would ask states to consider requiring loads above 100 megawatts — the scale of a large data center — to pay the full incremental cost of the infrastructure they need, rather than spreading those costs across all ratepayers. The Senate has not acted on the bill.

The political dimension has surfaced in unexpected places. Ohio gubernatorial candidate Vivek Ramaswamy has called for residents living near data centers to receive free electricity, according to RealClearMarkets, which frames the proposal as inadvertently pointing toward a longer-term economic logic: that data centers, as massive and consistent electricity consumers, could eventually drive prices toward zero through the same cost-spreading dynamic the arXiv study documents — though that outcome remains speculative and the paper itself cautions that future supply constraints could reverse the effect.

Data centers will eventually drive down electricity prices, perhaps to zero. If not, close to zero.— RealClearMarkets columnist

Columbia University's Center on Global Energy Policy, cited by costofowning.com, reviewed the research and found the effect is 'complex and context-dependent': data centers can spread fixed costs and lower rates, or raise local prices by as much as 20% where transmission is tight. The determining factor, in most analyses, is not the data centers themselves but the regulatory rules governing who pays for the grid upgrades they require.

EIA forecasts commercial electricity sales — driven by hyperscalers, bitcoin miners, and cloud computing — growing 2.2% in 2026 and 5.3% in 2027, when commercial use is expected to surpass residential use for the first time, according to costofowning.com. In the two regions with the heaviest data center development, ERCOT in Texas and PJM, forecast annual load growth averaged 10% and 3% respectively between 2025 and 2027. For a household using 900 kilowatt-hours a month, costofowning.com estimates the 2026 price rise adds roughly $93 a year.

Why it matters — Whether data centers ultimately lower or raise your electricity bill depends on grid capacity, state regulation, and who is required to pay for new infrastructure — decisions being made right now in statehouses and the US Senate.

Sources differ on Whether data centers raise or lower household electricity bills: Historical data (2015–2024) shows data centers modestly lowered average retail rates by spreading fixed grid costs. (arxiv.org) vs In regions with tight transmission, data centers can raise local prices by up to 20%; current national prices are rising faster than inflation. (costofowning.com) vs Long-term proliferation will drive prices toward zero through economies of scale. (realclearmarkets.com)

Reported by cepr.org, arxiv.org, costofowning.com, realclearmarkets.com