50 GW of Data Centers: What the Figures Tell Equipment Buyers

Correction, October 18, 2026. We corrected the electricity forecast comparison and removed unsupported conversions from load, tariff and queue figures into equipment orders. The earlier headline also confused tariff counts with states. This version distinguishes dated estimates, proposed rules and project commitments.

FERC staff estimated that U.S. data centers with more than 50 GW of combined capacity were in service at the end of 2025, representing 24 percent compound annual growth since 2020. The March 2026 staff report draws on Yes Energy’s project database. This is an estimate of in-service data-center capacity, not a count of new equipment orders or a measurement of simultaneous peak demand. (FERC staff report, page 7)

Keep energy, capacity and equipment quantities separate

EIA’s October 2026 outlook forecasts total U.S. electricity consumption of 4,288 billion kWh in 2026 and 4,356 billion kWh in 2027. National annual energy is a different measure from a campus’s connection capacity or local peak load. It cannot specify the number or rating of transformers a project needs. (October STEO, Table 7a)

FERC staff also reports average data-center size increasing from 25 MW in 2020 to almost 80 MW for facilities entering service in 2025. That comparison supports asking about larger connection requirements. It does not establish a fixed multiplier for transformers, switchgear or substations. Those quantities require the actual design and interconnection study. (FERC staff report, page 8)

A tariff listing is not a committed equipment project

SEPA and the NC Clean Energy Technology Center’s DELTa database covers both proposed and approved large-load tariffs and service rules. Use it to locate the relevant utility and proceeding, then inspect the operative terms. Do not treat a proposed tariff, an approved rate structure and an executed customer agreement as equivalent commitments. (DELTa database)

Record the customer’s obligations, security, service milestones and exit terms alongside the engineering scope. A long contract term or minimum-payment provision alone does not establish an equipment award or guarantee construction.

The federal process also needs a dated reading. On June 18, 2026, FERC issued region-specific show-cause orders requiring six regional grid operators and their transmission owners to defend or revise large-load tariff provisions. That procedural action does not itself grant every customer a uniform option-to-build right or establish every project’s cost allocation. Check the applicable order, tariff and agreement before relying on either. (FERC’s June 18 announcement)

Generation queues are a different dataset

Berkeley Lab’s 2025 Queued Up edition covers transmission-connected generation and storage requests through the end of 2024. It reports about 1,400 GW of generation and 890 GW of storage seeking interconnection. It is not a data-center load queue or a distribution-equipment order book. (Berkeley Lab)

The same study found that only 13 percent of capacity requested in 2000–2019 had entered commercial operation by the end of 2024; 77 percent had withdrawn. Queue entry therefore cannot be treated as a completed project. Keep proposed capacity, executed agreements, construction and operation separate when building a forecast.

Assess replacement needs locally

DOE reports a national estimate that about 55 percent of distribution transformers are older than 33 years. Use that as context for reviewing the local asset register, condition and loading. National age estimates do not establish which units need replacement or when. (DOE distribution-transformer research)

For a useful procurement forecast, connect each project to its approved load profile, engineering requirements, responsible buyer and dated supplier commitment. Distinguish new service from replacement work and unconfirmed quantities from committed orders. That produces a project list a buyer can update when the underlying evidence changes.

Subscribe to The Feeder for utility and distribution procurement research.