Data-center load is turning spare generation into a timing problem

EIA's high-demand case shows why near-term data-center growth is first a dispatch, fuel, and regional price problem, even before new plants can be built.

Power & GridAnalysis
Editorial image for Data-center load is turning spare generation into a timing problem

Regional electricity load and generation

In its February 2026 Short-Term Energy Outlook, the EIA forecasts U.S. electricity load growth of 1.9% in 2026 and 2.5% in 2027, with the fastest growth in ERCOT and PJM, where annual load is forecast to grow by an average of 10% and 3% a year, respectively, between 2025 and 2027. The high demand growth scenario built on that outlook raises the 2026 and 2027 growth rates by 50% in regions with significant data-center development and by one percentage point elsewhere, then compares load, natural gas generation, coal generation, and wholesale prices across STEO regions. Electricity use by large computing facilities is rising while the generating capacity available through the near-term forecast is largely fixed by projects that are already operating or already reported.

That combination makes utilization the first adjustment. Existing power plants have to produce more electricity before newly planned generation, transmission, or interconnection capacity can change the supply stack. Large computing facilities can increase electricity demand before additional generating capacity becomes operational within the STEO forecast period.

ERCOT exposes the timing risk

ERCOT and PJM are the clearest stress tests because the EIA expects the fastest electricity load growth in those regions through 2027, while MISO, SPP, Arizona and Nevada add data-center load but see lower overall growth because household demand there is growing slowly. The planning challenge is not simply that demand rises; the highest growth can arrive in regions with different fuel mixes, transmission links, weather patterns, and operating reserves.

The EIA's higher-demand case shows a sharper wholesale price response in ERCOT: a 2027 wholesale price $37 per MWh, or 79%, above the February STEO forecast, against a $2.60 per MWh (4%) increase in PJM and $1.30 to $3.00 per MWh in California, the Southwest, New England and New York. That result follows from the region's limited connections to neighboring grids and its need to rely on more expensive generation when electricity demand is high and wind generation is seasonally lower.

Gas becomes the near-term bridge

Natural gas-fired plants, which supplied 40% of U.S. generation in 2025, are the main flexible source in the EIA case because their generating capacity already exists and can increase output; U.S. gas generation rises 7.3% (123 billion kWh) between 2025 and 2027 in the high demand case against 1.7% in the baseline, with ERCOT accounting for the largest increase, and the scenario assumes delivered gas prices about $0.50 per MMBtu above the baseline. Coal generation, 17% of the 2025 total, declines 5.0% rather than 9.3% over the two years when electricity load runs above the baseline, while wind and solar, 18% combined in 2025, continue to depend on the resource available in each hour.

Near-term data-center planning depends on available generating capacity, fuel costs, and hourly wholesale prices. The operating constraint is the dispatch position of generation during the hours when a regional grid has to serve new computing load without additional near-term capacity.

Who gains and who carries exposure

Generators with available natural gas capacity can benefit from higher utilization, and grid regions with stronger transmission connections can draw on a wider supply pool. Data-center developers gain when a utility can identify firm generation and interconnection capacity without relying on a later project that remains outside the forecast period.

The exposure sits with projects whose power case assumes that average annual supply is equivalent to dependable hourly service. Wholesale prices, fuel delivery, late-summer demand, and the pace of large-load interconnection can all change the economics of a campus even when the regional energy balance looks adequate over a full year.

Read regional load before generation

The first checkpoint is the baseline forecast for electricity load in each STEO region and the higher-demand growth rate used for regions with significant data-center development. The next checkpoint is the existing generation fleet, because the EIA analysis intentionally holds near-term capacity to plants that are operating or already reported.

The third checkpoint is dispatch. Capacity planners should identify which natural gas, coal, nuclear, wind, and solar resources change output between the baseline and higher-demand cases, then connect that mix to fuel availability and the hours that set wholesale prices. That sequence turns a broad demand headline into an operating model that can be challenged.

Compare the forecast period for electricity load with reported additions to generating capacity. If load enters service before generation or transmission additions, the project depends on higher utilization of the current fleet. If the new supply arrives first, the same demand may have a different price and reliability effect. The comparison isolates changes in load, generation, natural gas use, coal generation, and wholesale prices.

Scenario boundaries

The higher-demand case changes regional growth rates while holding future generating capacity to the February STEO forecast. It cannot prove which announced data centers will open, how quickly they will reach full load, or which new supply projects will ultimately enter service.

That limitation is useful rather than fatal. It isolates the operating question that capacity teams should test: if load arrives earlier than supply, which existing generators run more, what fuel cost follows, and where do wholesale prices reveal the tightest regional margin?

Bottom Line

The decision turns on timing.

The high demand growth scenario compares electricity load, natural gas generation, coal generation, and wholesale electricity prices. The projects most likely to hold their schedules are those that connect their load date to a specific interconnection plan, an hourly generation profile, and a credible fuel and price case for the region that must serve them.

Why it matters

Capacity plans in ERCOT and PJM now depend on when large loads arrive, which generators can run harder, and whether grid supply can respond during the same hours.

Sources

This brief was drafted with AI assistance from public sources; every number is attributed to the sources above. See the AI-assisted Disclosure and Methodology.