Research · October 29, 2026

Constellation's $5B Conditional DOE Loan: 1.1 GW More From Reactors Already Running

Constellation secured a conditional DOE loan commitment of up to $5 billion to add nearly 1.1 GW from eight existing nuclear plants — with 20-year Google and Amazon agreements, plus a Meta expansion, as the demand anchor behind the uprates.

On October 27, 2026, Constellation Energy announced a conditional loan commitment of up to $5 billion from the U.S. Department of Energy's Office of Energy Dominance Financing to support nuclear uprates and modernization across eight of its plants. An uprate raises a reactor's output through equipment and process upgrades rather than a new build, and Constellation presents it as one of the fastest and most cost-effective ways to add dependable power without building entirely new generation facilities or transmission corridors.

The eight facilities named in the announcement are Braidwood, Byron, Clinton, Dresden, LaSalle, and Quad Cities in Illinois, Calvert Cliffs in Maryland, and Limerick in Pennsylvania. Together, the planned uprates would add nearly 1.1 gigawatts of capacity. The Energy Department says that is enough reliable electricity to power more than 930,000 homes, and estimates more than 10,000 project-related jobs.

The federal commitment builds on private money already moving. Twenty-year agreements with Google and Amazon have incentivized Constellation to invest more than $7 billion to add approximately 1,080 megawatts of nuclear capacity, and an agreement announced last year with Meta supports a 30-megawatt expansion at Clinton. Constellation says the new generation flows to the regional grid and benefits all customers. In that structure, the hyperscaler contracts are the demand anchor and the federal loan is the accelerant: Constellation chairman, president, and CEO Joe Dominguez described the deal as "pairing private-sector demand and investment with smart federal financing to unlock more power from the nuclear plants we already have."

The commitment sits inside the Energy Department's Utility Power Reactor Incremental Scaling Effort (UPRISE), which aims to add 5 gigawatts from the existing nuclear fleet by 2029. With Constellation's projects included, UPRISE has identified approximately 4.7 gigawatts — 95% of that goal. The program's premise is that the fastest new megawatts come from reactors already running, not from plants still on the drawing board.

What this is — and what it is not. The commitment is conditional, not a closed or funded loan: it is subject to due diligence, negotiation of definitive financing documents, and satisfaction of applicable conditions, and it does not by itself represent a final Constellation decision to proceed with any individual uprate or at every facility under consideration. The announcement discloses no per-plant capacity figures, no interest rate or loan terms, and no completion dates for individual uprates, and the $5 billion is a commitment ceiling, not money disbursed. The $7 billion is Constellation's own investment incentivized by customer agreements, not Department of Energy funding. And uprates add output to existing reactors; they are not new reactors.

For BlitzLab readers, the through-line is the power bottleneck behind the AI buildout being attacked from the generation side. Much of the attention in AI infrastructure goes to chips, data center buildings, and cooling — but none of that runs without megawatts, and the binding constraint is how fast new power can come online. Constellation's deal shows one answer: squeeze more out of the fleet that already exists, with long-term hyperscaler contracts providing the revenue certainty and federal financing lowering the cost of moving faster. Constellation (Nasdaq: CEG) does not currently carry a BlitzLab score, and this article is not a rescore — it is a financing data point about where AI-era power comes from when new plants take years and existing reactors can give more.

Sources