DOE Commits $17.5 Billion in Conditional Loans to Build 10 AP1000 Reactors for AI Power Demand

Key Facts
  • Loan facility: $17.5 billion (conditional)
  • Reactors: Up to 10 AP1000
  • Projects: 5 special-purpose vehicles
  • Target: Construction by 2030

The U.S. Department of Energy conditionally committed $17.5 billion in low-interest loans on June 22 to rebuild the domestic supply chain for Westinghouse AP1000 nuclear reactors, a package designed to put as many as 10 large reactors under construction in the United States by 2030 to meet electricity demand from artificial-intelligence data centers.

The loans run through DOE’s newly created Office of Energy Dominance Financing and would be split across 5 special-purpose vehicles, each formed between Westinghouse and a utility or energy company. The financing covers the long-lead nuclear components that have throttled past builds (reactor pressure vessels, steam generators, and main coolant pumps), items that currently carry multi-year order queues. DOE said locking in those orders now could pull reactor construction schedules forward by up to three years.

The special-purpose vehicles must contribute $1 billion in private capital collectively to qualify, and the five component loans later draw on a separate pool of more than $50 billion in DOE construction financing. Latitude Media reported that Dominion Energy, DTE Energy, Entergy, and WEC Energy Group are among the utilities positioned to use the program, with WEC pairing with EnergySolutions to weigh a build at Wisconsin’s Kewaunee site.

What the Loan Structure Reveals

The $17.5 billion covers long-lead components, not finished reactors. Spread across up to 10 AP1000 units, that works out to roughly $1.75 billion of federal component credit per reactor, and it sits on top of the separate construction pool of more than $50 billion the five loans later draw on. The headline number is therefore a down payment: the full program, once private equity and construction debt stack on, mobilizes well past $67 billion. Ordering pressure vessels and steam generators before any project reaches financial close is what lets DOE claim up to three years of schedule compression, and it is also what puts federal money at risk first if the builds stall.

Critical Perspective

The $17.5 billion conditional loan commitment aims to restart the AP1000 nuclear supply chain for 10 reactors by 2030, targeting AI data center demand. However, this ambitious goal overlooks the historical cost overruns and construction delays inherent in large-scale nuclear projects. For instance, the Vogtle plant in Georgia faced significant issues, with a reported $9 billion overrun and years of delay. Given these precedents, the $17.5 billion loan commitment seems overly optimistic for 10 reactors to come online by 2030 without substantial additional investment. The question is: how much more will this project cost when accounting for similar delays and overruns? Will ratepayers be left with a burden of stranded assets or will the promised 10 reactors actually materialize on schedule?

Why It Matters

The AP1000s backed by these loans will not generate power until the mid-2030s, well after the data-center load they are meant to serve arrives. That gap is why hyperscalers are signing gas-turbine and battery deals now and treating nuclear as a later-decade backstop. Energy Secretary Chris Wright has said he expects Meta, Microsoft, and Google to underwrite the reactors through long-term power-purchase agreements. That makes federal credit, not merchant electricity prices, the mechanism that decides whether the AP1000 line restarts at all.

Sources

Related Coverage

On the Ground
LocationWashington, DC
StagePlanned

Related post