The Existing Fleet: Nuclear Power’s Near-Term Capital Story

Zach Martin

Welcome to Nuclear Capital, on KMKT I R Hub Radio. I'm Zach Martin, and this is an I R Hub exclusive. I'm joined by Emily Martin. Emily, today we're skipping the glossy renderings of future reactors. We're talking about the nuclear plants that are already humming along, and why the most bankable money in this story is landing there.

Emily Martin

Thanks, Zach. I like that framing, because the headlines tend to chase the newest technology. But the contracts being signed in the first half of October 2026 are about the operating fleet. Uprates, which means getting more output from reactors that already run. License extensions, which keep them running. And restarts of plants that went dark and are being brought back.

Zach Martin

Let's start with the headline deal. On October 6, Google and Constellation announced a 20 year power purchase agreement to add 890 megawatts of new nuclear capacity to the PJM grid. It happens through uprates at 11 Constellation units, spread across six sites in Illinois, Pennsylvania, and New Jersey. Constellation says that represents more than $4.3 billion of new investment by the company.

Emily Martin

And the word new matters there. Constellation's release says this capacity is entirely new, meaning it expands the electricity available on the PJM grid rather than relabeling power that was already flowing. The company compares 890 megawatts to a new reactor, or to the output of several small modular reactors. But nobody is pouring a foundation. It comes from modernizing turbines, steam generators, and digital control systems at plants that are already licensed and connected.

Zach Martin

Think of it like a marathon runner who already has the training base. You don't recruit a new athlete and wait years for them to develop. You improve the runner's gear, tighten the form, and the times come down. The plant is the trained athlete.

Emily Martin

That works, and there's a long track record behind it. A review in the journal Nuclear Engineering and Technology, citing the Nuclear Regulatory Commission, says more than 8 gigawatts electric of capacity were added through uprates between 1977 and 2021. That's roughly the equivalent of eight new reactors, without new construction. Another analysis puts it at 171 approved uprates as of early 2022.

Zach Martin

So this isn't science fiction. It's a playbook that fell out of the spotlight. Walk us through the categories, because I think listeners hear the word uprate and picture one thing, when it's really a ladder.

Emily Martin

It is a ladder. The lowest rung is measurement uncertainty recapture, usually under 2 percent. Regulators once required plants to hold back a margin for uncertainty in how precisely power was measured. Better instruments, like ultrasonic feedwater flow meters, let a plant reclaim part of that margin. Next are stretch power uprates, up to about 7 percent, which use margin already built into the original design. The top rung is the extended power uprate, roughly 7 to 20 percent. That's where you replace serious equipment, such as high pressure turbines, pumps, generators, and transformers.

Zach Martin

And each rung costs more and takes longer to review.

Emily Martin

Right. The NRC has set review goals of about six months for the smallest category, nine for stretch uprates, and twelve for extended ones. Those are goals, not guarantees, and engineering, procurement, and outage scheduling add time on either side. The planning material in that engineering review gives screening level cost ranges too, and I want to be clear these are rough estimates, not plant specific numbers. They run from about $800 to $1,500 per kilowatt for stretch uprates, and about $1,500 to $2,500 for extended ones, climbing much higher for very large uprates.

Zach Martin

Now compare that to building from scratch. The Vogtle units three and four in Georgia started construction in 2009 against a budget of about $14 billion, with operation planned for 2016 and 2017. They entered service in 2023 and 2024, and Georgia Power estimates total project costs above $30 billion. The estimates for first of a kind new builds that I've seen in industry planning material run several times higher per kilowatt than uprates, though those are projections, not guarantees.

Emily Martin

That's the logic of de risking. An uprate happens on a site with a trained workforce, a permitted footprint, cooling infrastructure, and a grid connection. The work gets scheduled around refueling outages that the plant already plans. A new reactor brings a long list of firsts: siting, supply chains, fuel, financing, construction labor. Every first is a place for delay to hide.

Zach Martin

Let's talk about the buyer's side, though, because Google isn't signing a 20 year contract out of charity. Why does a hyperscaler want this structure?

Emily Martin

Time. If you're building AI data centers now, a megawatt in 2028 is worth much more than a megawatt in 2035. The release says the first uprate in this agreement is expected to be delivered by 2028. And there's a second piece. Alongside the 890 megawatts, Google and Constellation signed a 15 year supply agreement covering another 2,700 megawatts from Constellation's PJM fleet. Constellation describes that as ensuring those operating assets stay economically viable without taking resources off the grid.

Zach Martin

So one contract funds new capacity, and the other gives existing plants revenue certainty. That's a very different posture from an announcement that says we plan to explore something.

Emily Martin

Which connects to where this series has been. In an earlier episode, we looked at the gap between announced and signed nuclear power for AI. Across roughly 48 corporate nuclear agreements tracked in late 2026, nonbinding memoranda and frameworks represent about 40.4 gigawatts, against about 10.8 gigawatts in signed contracts. That's roughly 3.8 to 1.

Zach Martin

And I'd offer this as interpretation, not a headline fact: when you look at where the signed side is deploying capital, a lot of it points at the existing fleet. The announcements chase the future. The signatures are showing up where the steel is already in the ground.

Emily Martin

I think that's fair, with one caution. Signed doesn't mean finished. These agreements still require regulatory approvals, engineering, outages, and equipment delivery. But a binding contract changes who carries the risk and who commits cash.

Zach Martin

Let's look at the others. Amazon and Constellation announced a 20 year agreement on September 30 supporting a 190 megawatt expansion at Calvert Cliffs in Maryland. Utility Dive reports it centers on a $3 billion investment, and the new capacity is expected between 2030 and 2032. Utility Dive also describes it as part of a 690 megawatt arrangement.

Emily Martin

Notice the timelines. Calvert Cliffs is 2030 to 2032, while Google's first uprate is expected by 2028. Not every project is the same size or speed, even inside the existing fleet. Some uprates are modest instrument and setpoint work. Others involve major equipment, and the larger ones take longer.

Zach Martin

Then there's Georgia. In September, Georgia Power and Google announced Google would support uprates at Plants Vogtle and Hatch, adding about 96 megawatts electric. That's small next to a gigawatt scale data center campus, but it's a useful template, because the structure is designed to shield other customers from the incremental costs.

Emily Martin

And Meta has been active as well. In June 2025, Meta agreed to buy the full output of Constellation's Clinton plant in Illinois for 20 years, which helps avoid premature closure and supports a 30 megawatt uprate. Meta activity involving Vistra has centered on reactors in Ohio and Pennsylvania, including Perry, Davis-Besse, and Beaver Valley, which show up on the NRC's list of potential uprates.

Zach Martin

And Vistra itself got news on October 6. The Department of Energy announced a loan of approximately $4.2 billion to Vistra to upgrade and uprate reactors in Pennsylvania and Ohio. So the money is arriving from two directions: corporate offtake and federal financing.