Squeezing the Veterans: The Rise of Nuclear Plant Uprates

Zach Martin

Welcome to Nuclear Capital on KMKT I R Hub Radio. I'm Zach Martin, and I'm here with Emily Martin. Today we're talking about a quieter side of the nuclear story. It isn't the shiny new reactor. It's the reactor already running, and what happens when the biggest technology buyers on earth decide they want more power out of it.

Emily Martin

And it's a good moment for it, Zach. Two deals in about two weeks, both with Constellation Energy, one from Amazon and one from Google. Both are about squeezing more output from plants that already exist. Before we go further, the standard note. Everything we discuss on this show about companies and projects is for informational and analytical purposes only. It's not investment advice, and it's not a recommendation to buy or sell anything.

Zach Martin

Right. Think of it like a coach looking at a roster. Everybody wants to talk about the rookie phenom, the small modular reactor that hasn't been built yet. But the championship teams usually win by getting ten more percent out of the veterans. That's the idea here. So let's start with the Amazon deal, because it came first. Late September 2026, a twenty year agreement with Constellation for roughly 190 megawatts of incremental uprate capacity at Calvert Cliffs in Maryland.

Emily Martin

And that 190 megawatts sits inside a wider 690 megawatt agreement, tied to more than 3 billion dollars of capital for plant improvements and license extensions. The delivery window is 2030 to 2032. There's a nice historical wrinkle too. According to the Power Engineering background piece, the first power uprate the NRC ever approved was in September 1977, for Calvert Cliffs Unit 1. It was 5.5 percent, and Unit 2 got the same size increase less than two months later. So this is a plant with an uprate pedigree.

Zach Martin

That's a great detail. Now the Google deal, and I want to be clear that we covered the announcement mechanics in an earlier episode, the one about uprates, not unicorns. So I'll be brief. October 6, 2026. A twenty year power purchase agreement supporting 890 megawatts of new capacity through uprates at 11 existing units across 6 sites in Illinois, Pennsylvania, and New Jersey. Constellation is investing over 4.3 billion dollars. First uprate by 2028, the full amount before the end of 2032. And separately, a fifteen year agreement for 2,700 megawatts of existing generation.

Emily Martin

What I want to do today is zoom out from those headlines, because I think the interesting thing is the shape of the two deals together. Put them side by side and you see a template. A hyperscaler signs a long contract, twenty years in both cases. That contract gives the plant owner a revenue stream it can underwrite. The owner commits its own capital to upgrade the plant. And the new megawatts show up on the grid, not hidden behind a data center fence.

Zach Martin

Let me test that, because it's my job to be a little cautious. Is it a template, or is it two deals with the same counterparty? Constellation is on both sides of this. Fair question.

Emily Martin

It's fair, and the honest answer is that we're calling it a template as an interpretation, not a fact. But there's evidence it's spreading beyond one counterparty. In the same stretch of news, Google and Georgia Power announced a partnership to uprate Vogtle and Hatch and add about 96 megawatts electric. And on October 6, the same day as the Google and Constellation news, the DOE announced a roughly 4.2 billion dollar loan to Vistra for planned uprates in Ohio and Pennsylvania. Different owners, different structures, same direction.

Zach Martin

So the pattern is buyer with a balance sheet, owner with an engineering plan, and sometimes Washington with a loan. Now Emily, I'd like to get this one right for listeners. Why uprates? Why not just wait for new reactors?

Emily Martin

Because of time and risk. A new reactor is a first of a kind project for most of the designs people talk about, and nobody can promise you a clean schedule. An uprate happens inside a plant that already has a license, a workforce, a grid connection, and decades of operating history. The NEI Future of Nuclear Power survey, the 2025 update, put it bluntly. It said getting more output from the existing fleet is the most rapid and cost effective method to add large scale nuclear capacity. Now, that's an industry trade group, so it has a point of view. But the NRC's own backgrounder says approved uprates have already added capacity equal to about eight new reactors.

Zach Martin

Eight reactors worth, hiding in the margins of plants we already had. That's the compounding habit idea I love. Small improvements, over and over, add up to something big. But the word uprate covers a lot of ground, so let's get into the mechanics. Because a two percent tweak and a twenty percent rebuild are very different animals.

Emily Martin

The NRC sorts them into three buckets. The smallest is the measurement uncertainty recapture, or MUR, under two percent. When most plants were built, the safety analysis had to assume the reactor might be running at one point oh two times its licensed power, because the flow meters weren't perfectly accurate. Modern ultrasonic feedwater flow meters measure much more precisely. So the margin you were holding for instrument error can be given back as real output. No big hardware change. You're essentially removing a safety cushion that you've now proven you don't need.

Zach Martin

So it's like finding out your bathroom scale was off by a few pounds and recalibrating. Next bucket?

Emily Martin

Stretch power uprates, roughly two to seven percent. These stay within the plant's original design margin. You're changing instrument setpoints, procedures, maybe some valves and pump parts, but no major plant rebuild. And then the heavy one is the extended power uprate, or EPU. These have been approved for increases as high as twenty percent. They touch the non nuclear side of the plant, the balance of plant. High pressure turbines, condensate pumps and motors, main generators, transformers, moisture separator reheaters. More heat makes more steam, more steam has to go somewhere, and every piece of equipment in that path has to handle it.

Zach Martin

And that's where the money goes. When you hear 4.3 billion dollars across 11 units, you're mostly hearing about turbines, generators, and transformers, not the reactor core itself. I want to be careful, though. We don't have the breakdown of which unit gets which type of uprate, and the NRC's public list of expected applications masks a lot of Constellation's entries as proprietary. So we shouldn't guess at how that 890 megawatts splits up.

Emily Martin

Agreed. What the NRC list does show is the size of the wave. As of its September 2026 update, it expects 31 uprate applications across 2026 through 2032, about 2,400 megawatts electric in total. The busiest year is 2027, with 17 expected, and 2028 has 6, mostly extended power uprates. You can see Duke at McGuire and Brunswick, Southern at Hatch and Vogtle, PSEG at Salem, and later Vistra at Perry, Beaver Valley, and Davis Besse.

Zach Martin

And NEI's survey says more than seventy three percent of surveyed sites have some level of interest or planning for uprates, with more than 5 gigawatts electric of combined capacity, and over 22 billion dollars of planned capital for these kinds of plant changes over the next decade. Those are survey numbers, to be clear, and NEI says they aren't firm commitments. But what the Google and Amazon deals do is turn some of that interest into signed paper.