Are data centers driving up electricity prices?

Two different measurements answer differently. Backward-looking state retail rates show no statistically significant relationship with data-center concentration. In PJM's capacity market, the independent market monitor attributes $23.1 billion in cost increases across the 2025/26 through 2027/28 delivery years to data-center load. Both are on this page with sources.

When do PJM capacity costs reach household bills?

Capacity charges take effect with each delivery year, which runs June through May. When they appear on a household bill varies by state and supplier; no single region-wide date exists.

Home › Data centers › Electricity prices

Are Data Centers Raising Electricity Prices?

PJM data as of 2026-07-14 · last verified 2026-08-15 · page updated August 15, 2026

Two different measurements give two different answers, and both are real. Backward-looking state retail rates show no statistically significant relationship between how many data centers a state has and what households pay per kilowatt-hour. Forward-looking capacity markets tell a sharper story: in PJM, the largest US grid region, the independent market monitor attributes $23.1 billion of capacity-market cost increases across three delivery years (2025/26–2027/28) to data-center load. These measure different things on different clocks: what has already happened to retail rates nationwide, versus costs entering one region's bills now and over the next two years.

Measurement one: state retail rates, looking backward

Retail electricity rates are what households actually pay, and their history is public in EIA data. If data centers had already driven up retail rates, states with the most data centers should show higher or faster-rising rates. So far they do not, a finding published by the Institute for Energy Research in March 2026.

Labeled analysis · disclosure

Disclosure: this site's editor, Daniel Simmons, co-authored the IER report (with Tom Pyle). The report is a six-page descriptive analysis; its statistics are presented here as IER's claims, not as figures this site has independently reproduced, because the report does not identify its inputs (the specific EIA price series, sector and weighting, or the source of its 2025 state-level data-center counts) precisely enough to audit.

From “Have Data Centers Driven Up Electricity Prices?” (IER, March 19, 2026):

No significant correlation between a state's data-center count and its electricity price (Pearson r = −0.053, p = 0.72). The top ten data-center states averaged 14.46¢/kWh in 2025 vs 14.39¢/kWh for all other states.
Price growth in top-ten data-center states vs the rest was not significantly different over 2015–2025 (36.3% vs 28.1%, p = 0.144) or 2021–2025 (26.5% vs 21.1%, p = 0.228).
Its strongest finding runs the other way: states where electricity sales grew fastest saw the smallest price increases (2015–2025 Spearman ρ = −0.807, p < 0.0001), consistent with fixed grid costs spreading over more kilowatt-hours.

The report describes associations, not causes, and makes no claim about the future: as it puts it, “that may change in the future.” Data centers also choose states with cheap power, so causality plausibly runs in reverse.

Measurement two: PJM's capacity market, looking forward

Capacity prices are what generators are paid to be available in a future delivery year. They are set at auction years ahead, so they move before retail rates do. In PJM (13 states plus DC, 67 million people), capacity prices went from $28.92/MW-day for the 2024/25 delivery year to the price cap in each of the last three auctions:

PJM capacity auction clearing price by delivery year, $/MW-day

RTO-wide Base Residual Auction clearing prices. The 2026/27, 2027/28 and 2028/29 auctions all cleared at their FERC-approved price caps. Sources in the table below.

PJM's independent market monitor, Monitoring Analytics, models what each auction would have cost without data-center load in the forecast. Its answer: including data-center load raised costs to other PJM customers by $23,100,955,341 across the three delivery years 2025/26 through 2027/28. PJM itself wrote of the December 2025 auction that it “leaves no doubt that data centers’ demand for electricity continues to far outstrip new supply.”

Delivery year	Runs	Clearing price ($/MW-day)	At price cap	Data-center-attributed cost increase	Status
2024/2025	2024-06-01 → 2025-05-31	28.92	no	not yet published	complete
2025/2026	2025-06-01 → 2026-05-31	269.92	no	$9,332,103,858	complete
2026/2027	2026-06-01 → 2027-05-31	329.17	yes	$7,271,197,971	in effect
2027/2028	2027-06-01 → 2028-05-31	333.44	yes	$6,497,653,512	scheduled
2028/2029	2028-06-01 → 2029-05-31	325.00	yes	not yet published	scheduled

Attribution figures are Monitoring Analytics' modeled increase in capacity-market revenues from including all data-center load (existing plus forecast) in the peak-load forecast: a counterfactual model result, not a billed line item. Per-year values from IMM Analysis of the 2027/2028 BRA, Part A (January 5, 2026), Table 1. The 2028/29 row is from PJM's 2028/2029 BRA report (July 14, 2026), which also reports that auction cleared 6,831 MW short of the reliability requirement.

When do capacity costs reach household bills?

Capacity charges take effect with each delivery year, which runs June through May: the 2025/26 costs have been flowing through bills since June 2025, the 2026/27 costs since June 2026, and the 2027/28 and 2028/29 costs are scheduled. But there is no single region-wide date when a delivery year's costs show up on a given household's bill: it depends on the state and supplier — default-service procurement schedules, rate riders, and municipal and cooperative contracts all pass capacity costs through on their own timing. That lag is exactly why the two measurements on this page can disagree without contradiction: the capacity costs the market monitor attributes to data centers are only partway into the retail rates the state-level comparison measures.

How the two measurements fit together

Different geography. The retail-rate comparison is all 50 states; the capacity finding is one region (PJM), where data-center growth is most concentrated.
Different clocks. Retail rates reflect costs already passed through; capacity auctions price delivery years up to three years ahead.
Different mechanisms. A state's retail rate moves with fuel, wires, policy and demand trends together; the monitor's figure isolates one mechanism (capacity-market cost shifts) by modeling the counterfactual.

Neither measurement supports a blanket claim. The backward-looking data does not show data centers raising retail rates nationwide; the PJM analysis shows billions in capacity costs attributed to data-center load entering one region's bills now and over the next two years. For the demand side of the story, see how much electricity data centers use; for everything else on your bill, see why is my electric bill so high?

Sources

Monitoring Analytics, IMM Analysis of the 2027/2028 RPM Base Residual Auction, Part A (January 5, 2026)
PJM, 2028/2029 Base Residual Auction Report (July 14, 2026)
PJM, 2027/2028 Base Residual Auction results release (December 17, 2025)
IEEFA, PJM capacity price history summary (2024/25 and 2025/26 prices)
IER, Have Data Centers Driven Up Electricity Prices? (March 19, 2026; labeled analysis, authorship disclosed above)

The delivery-year table renders from a committed, validated dataset (pjm_capacity_timeline.csv) with per-row sources. Data as of 2026-07-14, last verified 2026-08-15.