Electricity Prices and Data Center Load: Where the June 2026 EIA Tables Stop

The Electric Power Monthly released on August 26, 2026 put the U.S. residential average price of electricity at 18.34 cents per kilowatthour for June 2026, against 17.47 cents in June 2025, a rise of 4.98 percent. The all-sector average over the twelve months ending June 2026 was 14.05 cents, against 13.22 cents a year earlier — a move of 0.83 cents, or 6.3 percent.

The argument around those numbers has narrowed to one question: whether new data center load moved them. That question is being asked of a table that was not built to answer it. The series can be pushed far — far enough to say what did and did not mechanically move the national average — but not that far.

The published price is a ratio, not a rate

EIA defines the figure directly, and the underlying collection is Form EIA-861M. Average revenue per kilowatthour is "calculated by dividing the total monthly revenue by the corresponding total monthly sales for each sector and geographic area." It is not a tariff, not a rate schedule, and not the price of the next kilowatthour — it is dollars collected divided by kilowatthours billed.

Three consequences follow, each detaching the number from the intuition attached to it.

Fixed charges ride in the numerator. A monthly customer charge is the same dollar amount at 1,200 kilowatthours or at 800. Illustrative arithmetic on the definition: 15 dollars fixed plus 1,000 kilowatthours at 15 cents is 165 dollars over 1,000 kilowatthours, or 16.50 cents; hold both rates and drop usage to 800 and the same customer is 135 dollars over 800, or 16.88 cents — up 2.3 percent with no rate change anywhere. The figures illustrate the arithmetic rather than any observed tariff; the direction is a property of the ratio.

The month is not a month. The EIA table notes that sales and revenue "accumulate from bills collected for periods of time (28 to 35 days) that vary dependent upon customer class and consumption occurring in and outside the calendar month." A June figure carries consumption from May.

The vintage is not final. Values for 2024 and earlier are final; values for 2025 and 2026 are preliminary estimates from a cutoff model sample. The 4.98 percent is a comparison of two preliminary numbers.

The mix shift did almost none of the work

The all-sector average is a weighted blend of four end-use sectors, so it can move when sales composition changes even if no sector's own price moves. That is worth testing rather than assuming, because the sectors are priced very differently: in June 2026, residential ran at 18.34 cents and industrial at 9.17 cents, a ratio of exactly 2.00, against 1.96 a year earlier.

Take the twelve months ending June 2026 against the twelve ending June 2025, using Tables 5.1 and 5.3. Sector shares of total sales moved: residential 37.44 percent to 37.01 percent, commercial 36.55 percent to 37.17 percent, industrial 25.83 percent to 25.64 percent, transportation flat at 0.18 percent. Splitting the change in the blended average into a price component (prior-year shares applied to each sector's own price change), a mix component (the change in shares applied to prior-year prices), and the interaction between them gives:

• Price component: +0.845 cents
• Mix component: −0.007 cents
• Interaction: −0.001 cents

The three sum to +0.837 cents, reproducing the published 0.83-cent move to within a hundredth of a cent; the residual is rounding in the published inputs. So the sector mix did not lift the national average; it shaved a fraction off, because the sector that gained share is cheaper than the one that lost it. Essentially the entire move is sectors repricing against themselves: residential up 6.6 percent, commercial 6.1 percent, industrial 6.4 percent on rolling 12-month averages.

Commercial load grew, and the series cannot name it

The volume side reads differently. Over the same twelve months, total retail sales grew 1.30 percent, from 4,029,247 to 4,081,696 thousand megawatthours. Commercial sales alone rose 3.02 percent, adding 44,528 thousand megawatthours — 84.9 percent of all the growth in U.S. retail electricity sales. Residential added 0.16 percent, industrial 0.54 percent, and transportation fell 0.99 percent.

That concentration is the strongest evidence for anyone arguing that new commercial-class load is reshaping the system. It is also where the classification starts to matter. EIA's commercial sector is defined as "service-providing facilities and equipment of businesses; Federal, State, and local governments; and other private and public organizations" — a service-sector category defined against the industrial sector, which EIA fixes as manufacturing (NAICS 31-33), agriculture, forestry, fishing and hunting (11), mining including oil and gas extraction (21), and construction (23). There is no data-center line.

The boundary is softer still. The same tables note that utilities "may classify commercial and industrial customers based on either NAICS codes or demands or usage falling within specified limits by rate schedule," and that year-to-year changes in counts, sales and revenues "may result from respondent implementation of changes in the definitions of consumers, and reclassifications." Two identical facilities on two utilities can land in two different sectors. EIA's own glossary adds that "various EIA programs differ in sectoral coverage."

The honest reading is asymmetric. The data supports "nearly all national load growth was commercial," and Table 5.4.A puts that growth on a map: commercial sales rose 4,405 thousand megawatthours between June 2025 and June 2026, and six states — Ohio (+565), Texas (+563), California (+544), Arizona (+479), Georgia (+450) and Virginia (+359) — account for about two thirds of it. What no table supports is which commercial customers inside those states the growth belongs to.

Two federal numbers for the same month, 0.95 points apart

There is a second published answer to "what happened to electricity prices," and it does not match. The CPI electricity index for U.S. city average, not seasonally adjusted, stood at 311.818 in June 2026 against 299.728 in June 2025 — a rise of 4.03 percent. EIA's residential figure for the same month rose 4.98 percent. The gap is 0.95 percentage points in the same month.

Neither is wrong; they are different constructions, differing on coverage and method. The EIA figure is revenue over kilowatthours for all fifty states and the District of Columbia, so a shift in how much households consume moves it even when tariffs hold. The CPI-U population, BLS states, "constitutes over 90 percent of the U.S. population" but covers "all urban households in core-based statistical areas (CBSAs) and in urban places of 10,000 inhabitants or more," expressly excluding rural nonmetropolitan and farm households — and its market basket "is developed from detailed expenditure information provided by families and individuals on what they actually bought." In the July 2026 release, electricity showed a 4.2 percent 12-month change against 3.4 percent for all items.

Anyone quoting a single "price of electricity" for the year is quoting one of these and not the other; the two differ by roughly a fifth of the increase itself.

Fuel is the competing explanation, and it is a large one

Attributing a price move to load growth requires ruling out fuel, which EIA lists first among the factors that influence electricity prices and singles out natural gas within: "Fuel prices, especially for natural gas and petroleum fuels … may increase during periods of high electricity demand," and "higher fuel prices, in turn, may result in higher costs to generate electricity." EIA also puts generation ahead of transmission and distribution — "the cost of generating electricity is the largest component of the price of electricity." It does not rule out easily this year. The Henry Hub natural gas spot price averaged 3.63 dollars per million Btu over the twelve months ending June 2026, against 2.97 dollars in the prior twelve — up 22.1 percent. January 2026 alone averaged 7.72 dollars, against 4.13 dollars in January 2025.

A 22 percent move in the fuel EIA names first among these factors is a candidate explanation for a 6 percent move in retail prices without any reference to new load. It also reaches bills on a different clock than the grid. EIA's own explainer is direct about this: "The cost to supply electricity changes minute by minute. However, most consumers pay rates based on the seasonal cost of electricity," and most consumers therefore "do not experience these daily price fluctuations." How far behind a retail rate sits is set jurisdiction by jurisdiction: the same page notes that some states fully regulate prices through commissions while others regulate only transmission and distribution. A January fuel spike does not have to appear in a January bill.

The states do not agree with the national number

The national figure hides most of what happened. Across the fifty states and the District of Columbia, June-on-June residential prices moved by a median of 5.15 percent; 29 of 51 came in above the national 4.98 percent, and five were negative — Arizona at −0.33 percent, Florida −1.63, West Virginia −2.34, Massachusetts −2.37 and Connecticut −10.56. At the other end, Hawaii rose 28.71 percent and Idaho 19.06 percent.

On the commercial side, the East North Central census division rose 12.25 percent year over year, against a national move from 13.54 to 14.19 cents, or 4.80 percent. An explanation that treats the country as one market discards the part of the data carrying the most information.

Three ways the next twelve months can read

Scenario A — repricing continues, mix stays quiet. Commercial sales keep outgrowing the other sectors while the mix component stays inside a hundredth of a cent, because commercial and residential prices are close enough that share shifts barely register. The national average then tracks within-sector rate changes and fuel recovery. On the last twelve months of data, this is the case the numbers most resemble.

Scenario B — the sector gap widens and mix starts to matter. If large new customers sit on distinct high-load-factor tariffs, the commercial and residential averages can diverge. Once the sectors are further apart, the same share shift produces a visibly larger mix effect. A mix component moving from −0.007 cents toward a tenth of a cent would be the first quantitative sign of it.

Scenario C — load growth slows where it was concentrated. EIA's August 11, 2026 Short-Term Energy Outlook lowered its forecast for Texas electricity load growth in 2027 to 6 percent, from 14 percent in the previous edition, after an August 3 state announcement pausing new data center development pending review. If similar pauses spread, the volume side cools while the price side keeps moving on fuel and rate cases, breaking the correlation now being read.

These are branches, not forecasts: the same monthly release is consistent with each, which is itself the finding.

What to Watch Next Week

• Henry Hub daily spot prices. EIA updates this series continuously; whether the twelve-month average holds near 3.63 dollars decides how much of the retail move is still fuel.
• The next Short-Term Energy Outlook, September 9, 2026. Watch whether the Texas revision (14 percent to 6 percent for 2027) extends to other regions or reverses.
• August CPI, September 11, 2026, 8:30 a.m. ET. The electricity index against July's 4.2 percent. A widening gap between CPI electricity and EIA residential signals usage, not tariffs.
• Electric Power Monthly with July data, September 24, 2026. July sector sales, and any revision to the preliminary 2025 and 2026 values used above.
• State commercial prices. East North Central at 12.25 percent is the outlier; a second division joining it changes the regional read.

Where This Doesn't Apply

If the interest is one household's bill, the national ratio is the wrong object. A bill is a tariff applied to one meter in one service territory. Five states saw residential averages fall while the national figure rose; a state reading sits closer to a bill.

If load growth becomes large enough, the mix logic reverses. The mix component is small because the commercial share moved 0.62 percentage points. A shift several times that size, or a wider spread between sector prices, would make mix material. The conclusion here is about the size of the observed shift, not mix effects in general.

If a state prices new large loads through a separate tariff class, the sector tables may not show it at all. Special contracts and dedicated rate schedules can sit inside the commercial or industrial totals without moving either average much, leaving this method blind to the customers under discussion.

The preliminary vintage cuts both ways. Revisions to 2025 and 2026 will change the decomposition. A revision lifting commercial sales further strengthens the volume finding; one trimming it weakens it.

Concrete Framework

A checklist for reading the next release rather than the next summary.

1. Name the series before quoting a number. EIA average revenue per kilowatthour and the CPI electricity index differed by 0.95 percentage points for June 2026. Say which is cited.
2. Run the decomposition, not the headline. Take sector sales from Table 5.1 and sector prices from Table 5.3, compute prior-year shares, and split the change into price, mix and interaction. Confirm the three sum to the published change before trusting any.
3. Check the volume side separately. A sector can supply most of the load growth while contributing almost nothing to the price change. Both were true here.
4. Subtract fuel before attributing anything to demand. Compare the 12-month Henry Hub average with the prior twelve; a 22.1 percent move is a large residual to leave unexplained.
5. Drop to the state level before drawing a national conclusion. A median of 5.15 percent with five negatives and a 28.71 percent high is not one phenomenon.
6. Mark the vintage. Anything for 2025 or 2026 is preliminary. Re-run the decomposition when values go final rather than carrying the first estimate forward.