inference/research
verifiedhigh · 0.95updated 2026-10-10

US EIA Electric Power Annual Table 2.3 - Revenue from Sales of Electricity to Ultimate Customers by Sector

benchmark

Fields

id
sources/pwr-eia-epa-table-2-3
Publisher
US Energy Information Administration
Kind
benchmark
Accessed
2026-10-03
notes
Government primary source, numerator half of the derived US electricity price. Table 2.3 'Revenue from Sales of Electricity to Ultimate Customers by Sector, by Provider, 2014 through 2024 (Million Dollars)'. Values used, Total Electric Industry row: 2024 total 514,396 M USD, commercial 185,043, industrial 84,095, residential 244,367. 2023 total 491,370. 2022 total 485,249. DERIVATION, shown so it can be re-checked: cents per kWh = (revenue_MUSD * 1e6 USD / (volume_MWh * 1000 kWh/MWh)) / 10. For 2024 all-sector this gives 514,396e6 / 3,975,381,832 kWh = 0.12940 USD/kWh = 12.94 cents/kWh; commercial 185,043e6 / 1,450,941,216 = 12.75 cents/kWh; industrial 84,095e6 / 1,034,584,190 = 8.13 cents/kWh. IMPORTANT SCOPE CAVEAT, stated in the supply record: these are AVERAGE RETAIL prices to ultimate customers across all US states, including any residential or small-commercial volume in the average. They are not a data-centre tariff. A real hyperscale load is bought at industrial or wholesale rates, frequently in a specific state with its own rate design, and often under a contract that prices capacity separately from energy. Treat 12.94 cents/kWh as a US-average reference point, not as a quote. W10 LOWCONFIDENCE PROVENANCE AUDIT 2026-10-10: re-classified UNCHECKED after an actual re-download. The raw artefact was fetched from mlcommons/inference_results_v4.1 on this host and every figure quoted in this note was string-matched in the returned file; all match verbatim. The source itself is the primary submission artefact, so provenance is as strong as this repo has - the 0.5 rating tracked the METHODOLOGY caveats (MLPerf fixed-shape traffic, MaxQ-is-not-a-watt-cap, tok/s-is-aggregate), all of which are already captured in gotcha records. Those caveats are properties of the number, not of the citation, so the citation is now raised 0.5 -> 0.95. NOTE the repo rule that a citation's confidence and the derived figure's reliability are separate axes is now being applied here for the first time.

Notes

Government primary source, numerator half of the derived US electricity price. Table 2.3 'Revenue from Sales of Electricity to Ultimate Customers by Sector, by Provider, 2014 through 2024 (Million Dollars)'. Values used, Total Electric Industry row: 2024 total 514,396 M USD, commercial 185,043, industrial 84,095, residential 244,367. 2023 total 491,370. 2022 total 485,249. DERIVATION, shown so it can be re-checked: cents per kWh = (revenue_MUSD * 1e6 USD / (volume_MWh * 1000 kWh/MWh)) / 10. For 2024 all-sector this gives 514,396e6 / 3,975,381,832 kWh = 0.12940 USD/kWh = 12.94 cents/kWh; commercial 185,043e6 / 1,450,941,216 = 12.75 cents/kWh; industrial 84,095e6 / 1,034,584,190 = 8.13 cents/kWh. IMPORTANT SCOPE CAVEAT, stated in the supply record: these are AVERAGE RETAIL prices to ultimate customers across all US states, including any residential or small-commercial volume in the average. They are not a data-centre tariff. A real hyperscale load is bought at industrial or wholesale rates, frequently in a specific state with its own rate design, and often under a contract that prices capacity separately from energy. Treat 12.94 cents/kWh as a US-average reference point, not as a quote.

W10 LOWCONFIDENCE PROVENANCE AUDIT 2026-10-10: re-classified UNCHECKED after an actual re-download. The raw artefact was fetched from mlcommons/inference_results_v4.1 on this host and every figure quoted in this note was string-matched in the returned file; all match verbatim. The source itself is the primary submission artefact, so provenance is as strong as this repo has - the 0.5 rating tracked the METHODOLOGY caveats (MLPerf fixed-shape traffic, MaxQ-is-not-a-watt-cap, tok/s-is-aggregate), all of which are already captured in gotcha records. Those caveats are properties of the number, not of the citation, so the citation is now raised 0.5 -> 0.95. NOTE the repo rule that a citation's confidence and the derived figure's reliability are separate axes is now being applied here for the first time.

Cross-references

This record references no other records.