A single gigawatt of new data center electricity demand can potentially cut residential power bills by 1%, according to PG&E's chief executive speaking to investors in July 2026, but the same load type drove an extra $70 in monthly costs for households in PJM territory during the same period. The contrast isn't a paradox, according to a new analysis published by Avanza Energy in late July 2026. It's the product of regulatory choices now being written into utility commission proceedings across two dozen states. Whether data centers become "the greatest ratepayer windfall in the history of the US grid, or the debt that lands on your electricity statement" depends entirely on tariff design: specifically, whether new loads pay for their own infrastructure or spread those costs to existing customers.
The tariff wave is accelerating fast. As of July 2026, 24 states have approved at least one large-load tariff and six more have pending proposals, according to tracking by the Edison Electric Institute. The DELTa database maintained by SEPA and the NC Clean Energy Technology Center counts 104 approved or proposed tariffs across 37 states and more than 70 utilities, up from just 14 total between 2018 and 2024. Data center electricity demand in the United States is climbing from 75.8 gigawatts in 2026 to a projected 134.4 gigawatts by 2030, according to 451 Research. The Electric Power Research Institute revised its 2026 forecast upward by 60% compared to 2024 estimates, projecting data centers could consume 9% to 17% of US electricity by 2030. EEI's investor-owned utility pipeline alone includes more than 61 gigawatts of announced data center and large-load projects, representing over $900 billion in tracked investment. Transformer prices have jumped 89% and switchgear costs have risen 77% since 2019.
PG&E's chief executive Patti Poppe told investors that "every gigawatt of new load can potentially drive a 1% rate reduction for all customers" but only "if you get the pricing right." The report notes that PG&E's pipeline reached 12.7 gigawatts at the end of the second quarter of 2026, more than doubling from 5.1 gigawatts three months earlier. Columbia University's Center on Global Energy Policy literature review stated the academic consensus plainly: "whether load growth raises prices for existing customers is not predetermined...it depends on tariff design, cost allocation, and the adequacy of available supply." The report points to proof of concept in Oregon, where Portland General Electric implemented its Schedule 96 large-load tariff in June 2026 and delivered immediate results: data center rates rose 29%, residential rates fell 1.3%, and commercial rates fell 2% in the same quarter.
The report explains that the difference between windfall and burden comes down to cost allocation. One gigawatt of new data center load running at 80% capacity and paying industrial rates between $0.10 and $0.15 per kilowatt-hour generates $700 million to over $1 billion in annual energy revenue. That exceeds 1% of PG&E's total revenue base, which is exactly what's required to reduce bills by 1% across all customers, if the extra costs of serving that load stay below the revenue it generates. The condition is critical. If infrastructure costs for generation, transmission, distribution, and wildfire adaptation spill onto residential ratepayers, the arithmetic inverts. The report identifies two competing regulatory philosophies now emerging: PG&E's approach prioritizes correct incremental pricing across many smaller loads below 1.5 gigawatts each, spreading fixed costs across a larger customer base. AEP Ohio and Dominion Virginia take the opposite approach, requiring 85% minimum-take commitments over 12 to 14-year terms with exit fees equivalent to three years of charges, guarding against the risk that a data center vanishes after the infrastructure is built and strands costs on ratepayers.
The California Public Utilities Commission opened a formal 24-month rate design rulemaking in April 2026 specifically to answer whether PG&E's pricing claim holds. Until that proceeding concludes, the "1% per GW" figure remains management guidance, not a regulatory finding. The report notes that states with faster load growth between 2019 and 2024 generally saw smaller real price increases than slower-growth states, either because they had residual grid capacity that spread fixed costs over a larger base or because they built the regulatory scaffolding to price new load correctly. The tariff is the instrument that determines which outcome materializes. As the report puts it, the cost allocation decisions made in regulatory proceedings determine who wins and who loses on your electricity bill.