Right Number, Wrong Blame
A post went viral this week. We checked it against the government's own numbers.
The post.
Your electricity, it said, now costs nineteen point six cents a kilowatt-hour. Forty percent more than in 2019. Some states up nearly seventy, Maine up ninety. And Big Tech's AI gold rush is why.
The post went up on X on 20 October 2026 from a private-sized account, so it is not named here. It put four figures on the table: "hitting 19.6¢ per kWh in August 2026", "40% higher than in 2019", "nearly 70%" for California and New York and "90%" for Maine, and it laid the rise at the door of "Big Tech's AI gold rush". Two agencies publish the price of residential electricity in the United States, by different methods. We checked each figure against both, and the blame against the years.
Two agencies, two methods.
We checked it against the government's own numbers. There are two.
The Bureau of Labor Statistics tracks what city households pay. Its figure for August: nineteen point six cents. Exactly.
The Energy Information Administration counts every residential customer in the country. Its August figure: eighteen point one six.
Different methods. Both real. And forty percent since 2019? On both, it holds, before inflation.
The Bureau of Labor Statistics' series is a consumer price: "average prices are estimates of the average price paid by the consumer for a good or service", drawn from the CPI's sample of urban households, and its name says "U.S. city average". Its August 2026 figure is $0.196 per kilowatt-hour, and so is September's; August 2019 was $0.139, which is 41.0 per cent more, our arithmetic. The Energy Information Administration's figure is revenue divided by sales: "calculated by dividing the total monthly revenue by the corresponding total monthly sales for each sector and geographic area", for every residential customer in the fifty states and the District of Columbia, from the utilities' own monthly reports.
Its August 2026 figure is 18.16 cents, a preliminary estimate in its own words, against a 2019 annual average of 13.01: 39.6 per cent more, our arithmetic, a month set against a year. Both comparisons are in the money of their day, not adjusted for inflation.
The number is right.
So the headline number is right.
The post's 19.6 cents is the Bureau of Labor Statistics' own figure for August 2026, to the digit. Its forty per cent since 2019 holds on the Bureau's series (41.0 per cent, August against August) and on the Energy Information Administration's (39.6 per cent, August 2026 against the 2019 average), before inflation. The two agencies measure different things, a sample of city households and the whole residential market, and arrive at the same rise.
The states.
The states are another matter.
Maine: up seventy-nine percent, not ninety. California: up fifty. New York: fifty-six. Not nearly seventy.
Big rises. Just not the ones in the post.
The Energy Information Administration's Table 5.6.A gives each state's residential price for August 2026, preliminary, and August 2025, final: Maine 32.11 cents per kilowatt-hour, California 29.75, New York 28.65, the U.S. total 18.16. Their August 2019 figures come from the pipeline's own record of the agency's API: Maine 17.92, California 19.78, New York 18.38, the U.S. total 13.30. August 2026 against August 2019, our arithmetic: Maine up 79 per cent, California up 50, New York up 56, the country up 37. The post said 90 per cent for Maine and nearly 70 for California and New York. The rises are large, and nominal; the post's figures for them are not the agency's.
Now the blame.
On the energy agency's yearly numbers, prices climbed from thirteen cents in 2019 to sixteen by 2023.
That's most of the rise, and it came before the data-centre build-out of the last three years.
Since 2023, the climb has been slower. Sixteen to eighteen.
The Energy Information Administration's Table 5.3 gives the residential price as a yearly average: 13.01 cents in 2019, 13.15 in 2020, 13.66 in 2021, 15.04 in 2022, 16.00 in 2023, 16.48 in 2024, 17.23 in 2025, and 18.19 for the first eight months of 2026, preliminary. From 2019 to 2023 the price rose 2.99 cents; from 2023 to the 2026 figure, 2.19 cents more. So 2.99 of the 5.18 cents of rise since 2019, about 58 per cent, had come by 2023, our arithmetic on the agency's figures. The table states no cause for any of it, and this piece names none: it says when the rise came, not why. "Values for 2025 and prior years are final."
To be fair.
To be fair, data centres do show up in power prices. In some markets the price of reserving power for them has jumped, and we've covered it.
But the national bill was rising before the boom.
Where data centres do reach the bill is the price of reserving power: PJM's capacity auction, which this site followed in Who Pays for the Power?. That is a market-by-market story. The national residential price in the agency's yearly series was already up 2.99 cents between 2019 and 2023. Neither agency's table says what moved it, and this piece does not either. Prices here are nominal, and the 2026 figures are the agency's preliminary estimates.
So the viral post got the number right.
It got the states wrong. And it put the blame on the wrong years.
Every figure from the Bureau of Labor Statistics and the Energy Information Administration.
Sources
- Bureau of Labor Statistics, Average Price Data, series APU000072610, "Electricity per KWH in U.S. city average", read through the BLS public API v2 on 27 October 2026: 93 monthly observations from January 2019 to September 2026, in dollars per kilowatt-hour; August 2026 0.196, September 2026 0.196, August 2019 0.139. The viral post's 19.6 cents is this series' August 2026 figure.
- Bureau of Labor Statistics, "CPI: Average Price Data" (factsheet): "Specifically, average prices are estimates of the average price paid by the consumer for a good or service."; "Average prices are estimated from a subset of CPI data"; "the CPI sample is optimized for the calculation of price indexes, not for the calculation of average prices."; "Electricity prices are reported in kilowatt hours (kwh)."
- Energy Information Administration, Electric Power Monthly, Table 5.3, "Average Price of Electricity to Ultimate Customers: Total by End-Use Sector, 2016 - August 2026 (Cents per Kilowatthour)", read 27 October 2026. Residential, annual: 2019 13.01; 2020 13.15; 2021 13.66; 2022 15.04; 2023 16.00; 2024 16.48; 2025 17.23; 2026 year to date (January to August) 18.19. Residential, August 2026: 18.16; August 2025: 17.57. "Values for 2025 and prior years are final."; "Values for 2026 are preliminary estimates based on a cutoff model sample."; source Form EIA-861M.
- Energy Information Administration, Electric Power Monthly, Table 5.6.A, "Average Price of Electricity to Ultimate Customers by End-Use Sector, by State, August 2026 and 2025 (Cents per Kilowatthour)", residential, read 27 October 2026: Maine 32.11 (August 2025: 28.14); California 29.75 (30.92); New York 28.65 (26.64); U.S. total 18.16 (17.57). "Values for 2025 are final. Values for 2026 are preliminary estimates based on a cutoff model sample."
- Energy Information Administration, API v2, electricity/retail-sales, residential average price by state and for the U.S. total, monthly, as the pipeline's own record of 19 October 2026 holds it (data through July 2026): August 2019, Maine 17.92, California 19.78, New York 18.38, U.S. total 13.30 cents per kilowatt-hour; the U.S. residential series January 2019 to July 2026 drawn in the first chart.
- Energy Information Administration, Glossary, "Average revenue per kilowatthour": "The average revenue per kilowatthour of electricity sold by sector (residential, commercial, industrial, or other) and geographic area (State, Census division, and national) is calculated by dividing the total monthly revenue by the corresponding total monthly sales for each sector and geographic area."