Natural Gas Demand Surge: How Data Centers Are Reshaping the US Power Grid

The Problem: Record demand without increased fuel consumption

The U.S. Energy Information Administration forecasts natural gas consumption in the electric power sector will hold steady at 43.7 billion cubic feet per day this summer, matching 2025 levels and sitting just 4% above the five-year average (2021–2025). On the surface, that looks like stability. It isn't.

The EIA projects a new record will hit in summer 2027, when natural gas consumption for power generation climbs to 46.1 Bcf/d, a 6% increase (2.4 Bcf/d) over 2026 and 3% above the previous record set in 2024. The difference? Data centers and large manufacturing facilities are coming online at a scale that hasn't been seen before.

The Drivers: ERCOT and PJM

The West South Central region (primarily ERCOT in Texas) and the Mid-Atlantic (primarily PJM in Virginia) are where the action is. The EIA expects commercial and industrial electricity demand in the West South Central region to jump 20% from summer 2025 to summer 2027. That figure alone should make anyone designing power infrastructure take notice.

ERCOT will likely meet rising demand with more natural gas and solar generation, with a 22% increase in natural gas-fired generation expected between 2025 and 2027. I expect natural gas peakers and intermediate-load units will see higher capacity factors in that market as data center loads cluster near metropolitan areas with existing gas pipeline infrastructure.

PJM tells a slightly different story. The region has been steadily increasing natural gas consumption for electricity for at least a decade, driven by both rising power demand and coal-to-gas fuel switching. The EIA forecasts natural gas consumption for electricity generation in PJM will increase 6% (9 billion kilowatt-hours) in summer 2027 compared to summer 2025. Solar generation in PJM is expected to grow 32% (4 BkWh) over the same period, but natural gas remains the dominant flexible resource.

The Grid Operation Reality

The broader U.S. generation mix continues shifting away from coal, but that trajectory isn't clean or linear. During the first four months of 2026, electricity, natural gas, and coal prices have created favorable economics for coal generation in the MISO region. The "dark spread" (fuel cost to wholesale electricity price for coal) in MISO has outpaced the "spark spread" for natural gas generators.

That gap isn't academic. During Winter Storm Fern in January, the difference between the dark and spark spreads reached $/MWh, a figure that tells you coal units can run profitably even when gas prices spike. Grid operators in MISO territory still need those coal facilities available for winter peaks—something engineers designing generation portfolios should factor in rather than assume coal is on its way out.

What This Means for Power Sector Engineers

Two conclusions stand out. First, natural gas generation capacity additions remain essential despite renewable growth. The 22% ERCOT increase and 6% PJM increase reflect situations where intermittent resources can't meet continuous data center loads without gas backup. Second, coal retirement timelines depend heavily on regional economics. MISO face value suggests 2026-2028 timelines are optimistic for baseload coal in that territory.

My take: anyone specifying generation equipment for facilities in ERCOT or PJM should plan for gas-fired capacity with high dispatchability assumptions, not treat renewables as firm capacity replacements.

#