NUCLEAR CAPITAL
All Episodes
Announced vs. Signed Nuclear Capacity for AI

Announced vs. Signed Nuclear Capacity for AI

0:00|0:00

Zach and Emily Martin examine the 3.8-to-1 gap between non-binding nuclear announcements and signed data center contracts as of early November 2026. They explore how firm offtake deals for existing fleet uprates differ from early-stage options in execution risk, grid reality, and capital formation.


Chapter 1

Announced Versus Signed

Zach Martin

Welcome to NUCLEAR CAPITAL on KMKT I R Hub Radio. I'm Zach Martin, and today's lens is simple. Announced versus signed. As of early November 2026, public reporting puts roughly 10.8 gigawatts of nuclear capacity under signed contracts or PPAs with AI and data center buyers, and about 40.4 gigawatts in non-binding MOUs, letters of intent, and options. That's a 3.8 to 1 ratio.

Emily Martin

And I'd start with what a signed deal does that a headline doesn't. A signed PPA is a cash flow. A lender can underwrite it, a regulator can point to it, and somebody is legally on the hook. A non-binding framework is an intention, and nobody owes anything if it fades.

Zach Martin

Think of it like training. Announcing a marathon is easy. Signing up, paying the fee, and logging the miles is the real commitment. Let's look at the miles. Amazon and Constellation, announced around September 30, a 20 year PPA at Calvert Cliffs in Maryland. Amazon gets 690 megawatts, including a 190 megawatt uprate expected between 2030 and 2032, and the companies say it enables over 3 billion dollars of infrastructure.

Emily Martin

Then Google and Constellation, October 6. A 20 year PPA for 890 megawatts of new capacity through uprates across 11 units in PJM, with more than 4.3 billion dollars of private investment by Constellation, and first uprates by 2028. Plus a separate 15 year supply agreement for 2,700 megawatts. And Meta and Vistra in January, 20 year PPAs over 2,600 megawatts at Perry, Davis-Besse, and Beaver Valley, including 433 megawatts of uprates.

Zach Martin

Notice the pattern. Existing plants, existing interconnections, investment grade buyers, long terms. That's the firm core.

Emily Martin

Now I'll push on your marathon analogy. Is the 40.4 gigawatts really just talk? Oklo and Switch have a 12 gigawatt Master Power Agreement running through 2044. That's not nothing.

Zach Martin

It's not nothing, but it's non-binding. It's an umbrella that needs individual binding PPAs as milestones are met. Twelve gigawatts is nearly the size of the entire signed pile, and it carries no contractual liability if it's abandoned. I'd call that option value, not bankable revenue.

Emily Martin

Option value matters, though. Early commitments can signal demand to suppliers and regulators. I just won't pretend it's the same asset class as Calvert Cliffs.

Zach Martin

Agreed on that. Different asset class, different risk.

Chapter 2

Execution Risk and Capital

Emily Martin

So what separates them on execution risk? The early stage announcements, including TVA explorations and advanced reactor options with Oklo and TerraPower, face first of a kind design hurdles, NRC licensing timelines, and HALEU fuel supply constraints. Those are unresolved.

Zach Martin

Versus uprates that deliver electrons inside existing grid connections, notably in PJM, without a ten year greenfield approval process. That's grid reality. Though grid reality cuts both ways. FERC rejected most of PJM's large load interconnection plan in October, so rules around large loads are still moving.

Emily Martin

And capital formation follows the paper. On October 27, the Department of Energy's Office of Energy Dominance Financing issued a conditional commitment of up to 5 billion dollars to Constellation for uprates at 8 stations, adding about 1.1 gigawatts. Conditional, mind you, contingent on final definitive agreements.

Zach Martin

Right, and that word conditional is the same lesson at federal scale. Even a big loan isn't done until the documents are signed. Debt markets reward signed contracts from creditworthy buyers. Pipelines get patience, not financing.

Emily Martin

My view is the ratio isn't a failure. It's a funnel. The question is how much converts, and when.

Zach Martin

Mine is more cautionary. Watch conversion, not announcements. A ratio of 3.8 to 1 tells you where the discipline is, and today it sits with existing fleet deals. Whether new build converts is unproven.

Emily Martin

We'll leave that one open. A reminder that the companies discussed are for informational purposes only. This is not investment advice, and not a recommendation to buy or sell any security.

Zach Martin

Thanks for listening to I R Hub. Real assets. Real ideas. A brighter tomorrow.