Texas's data center buildout, in reactors

For scale: Texas already has nuclear. It's not enough.

Texas runs four operating commercial reactors — two units at Comanche Peak (north of Fort Worth, Vistra-operated, 2.4 GW combined) and two at South Texas Project (Bay City, 2.7 GW). At fleet-average capacity factor, those four reactors together deliver about 41 TWh/year.

Set against ERCOT's ~580 TWh of projected new data center load, the entire existing Texas nuclear fleet covers roughly 7%. Filling the rest with new AP1000-class reactors would require building about 60 more — a 16× expansion of the state's commercial nuclear capacity.

The land-use side, at Texas scale

Texas has the best onshore renewable resource in the lower-48 — its solar capacity grew 92% between 2024 and 2026 per EIA. But the math on covering 580 TWh/yr with renewables firmed-up to 24/7 reliable power runs into geography fast. About 9,000 utility-scale solar farms at 100 MW apiece would occupy roughly 8,400 square miles — almost exactly the size of the entire Houston metro area (~8,269 sq mi) or the state of New Jersey (~8,729 sq mi).

The same 580 TWh from onshore Texas wind would take roughly 140,000 turbines spread across 22,000 sq mi — an area larger than West Virginia. The 64 AP1000-class reactors that would deliver the same energy occupy under 85 sq mi of total site footprint — about 1% of the solar land use, or 0.4% of the wind land use.

For natural gas (Texas's grid-staple): about 220 600-MW combined-cycle plants, emitting roughly 230 million tons of CO₂ per year — comparable to the annual emissions of every car in California.

The state knows the forecast might be high

The 78 GW figure comes from ERCOT's long-term load forecast submitted to the Public Utility Commission of Texas in 2025. ERCOT also presented a more aggressive scenario showing peak demand quadrupling by 2032 — which the PUCT promptly sent back as inflated. ERCOT's own CEO Pablo Vegas conceded the interconnection queue (over 220 GW total, ~70% data centers) "reflects higher than expected future load growth."

The likely truth sits somewhere between the queue total (clearly speculative — many projects will never break ground) and the official 78 GW projection. Either way, the direction is the same: Texas needs to build enough new firm capacity to match a state-sized data center buildout, and it needs to do it inside the same deregulated ERCOT market that's already showing capacity-price stress.

The named-project pipeline so far

A partial inventory of disclosed Texas data-center commitments since 2024 — well short of the full 78 GW projection but illustrative of the pace:

Capacity figures from company press releases and ERCOT interconnection-queue reporting; AP1000-equivalent assumes 85% data-center utilization vs 93% reactor capacity factor. Several projects (Fermi America in particular) are speculative or early-stage.

What this means for Texas's grid mix

ERCOT's near-term answer to data center demand has been new gas turbines plus continued solar buildout — EIA forecasts 9.6% generation growth in 2026 alone, with most new megawatt-hours coming from solar (which grew 92% between 2024 and 2026) and natural gas. Coal generation is also rising in some grid regions to backfill dispatchable capacity.

Senate Bill 6 (passed June 2025) gave ERCOT the legal authority to curtail data-center load during grid stress events — a recognition that adding 78 GW of firm-baseload-priced demand to a grid built around variable renewables creates reliability risk. SB-6 also redefined large-load interconnection (≥75 MW) and added transmission-cost requirements designed to make data-center developers pay for the grid they need.