The Great Uprate Rush: Powering AI with Existing Nuclear Fleets

Zach Martin

Data center electricity use jumped 17% in 2025 and is on track to double by 2030. Everyone says nuclear is the answer for AI. But the capital that's actually moving right now is going into the reactors we already have. This is NUCLEAR CAPITAL on I R Hub Radio, and today we're calling it the Great Uprate Rush. I'm Zach Martin.

Emily Martin

And I'm Emily Martin. One quick note up front. Companies we mention, including Constellation Energy, Vistra and Cameco, are named for informational purposes only. Nothing here is investment advice, and nothing is a recommendation to buy or sell.

Zach Martin

Good ground rules. So what's an uprate? Think of it like tuning an engine you already own. Operators modernize turbines, steam generators, instrumentation and digital controls, and the plant squeezes more electricity out of the same site.

Emily Martin

And the headline deal is Google and Constellation, announced October 6, 2026. It's a 20-year power purchase agreement supporting 890 megawatts of new capacity through uprates across 11 Constellation units at six sites in Illinois, Pennsylvania and New Jersey. Constellation says it represents more than $4.3 billion of investment. The first uprate is expected by 2028.

Zach Martin

Plus a separate 15-year agreement for 2,700 megawatts of existing PJM nuclear output. So one piece is new electrons, and the other keeps current plants economically secure.

Emily Martin

Right, and that same month, on October 27, the Department of Energy announced a conditional loan commitment of up to $5 billion to Constellation. It would finance uprates at eight facilities, including Braidwood, Byron, Clinton, Dresden, LaSalle, Quad Cities, Calvert Cliffs and Limerick. That unlocks nearly 1.1 gigawatts. Byron alone could gain about 150 megawatts. It's conditional, so it's not a done deal. Vistra also reportedly received a conditional commitment of about $4.2 billion earlier in October.

Zach Martin

And Amazon is in the mix. On September 30, Constellation and Amazon announced a 20-year agreement tied to a $3 billion, 190 megawatt uprate at Calvert Cliffs in Maryland. That takes the total commitment there to 690 megawatts, with delivery between 2030 and 2032.

Emily Martin

Notice that timeline. Even an uprate can take years. But compare it with building from scratch. Greenfield nuclear faces seven to ten years of licensing and construction. An uprate leans on an existing license, existing water rights, a switchyard and transmission connections that are already there. It doesn't have to stand in a brand new interconnection queue.

Zach Martin

That matters on the grid side too. Back in late 2024, FERC raised hurdles for co-located, behind the meter arrangements like the Amazon and Talen Susquehanna campus. Uprates put net new power directly on the regional grid, which fits PJM's Bring Your Own Power idea and the White House ratepayer protection pledge. That's our interpretation of why the structure's attractive, but Google and Constellation framed it in similar terms.

Emily Martin

Now the long game. Hyperscalers have contracted or agreed to roughly 14 gigawatts of nuclear capacity. That's a mix of existing output, uprates, restarts and new builds. Yet almost no brand new advanced capacity arrives before 2027 or 2028.

Zach Martin

Meta's portfolio shows the spread. It has 20-year agreements with Vistra for 2,176 megawatts from Perry and Davis-Besse, plus 433 megawatts of planned uprates including Beaver Valley. Then the long-dated bets. TerraPower's Natrium, up to 2.8 gigawatts, earliest targeted 2032 to 2035. And Oklo, up to 1.2 gigawatts in Ohio, targeted from 2030.

Emily Martin

And on October 19, AEP, X-energy and Amazon agreed to evaluate a 960 megawatt project, twelve 80 megawatt Xe-100 reactors at the retired Breed coal site in Sullivan County, Indiana. Amazon would fund development and plans a long-term offtake contract to protect utility ratepayers. But it's feasibility and pre-development, with delivery far into the 2030s. Execution risk lives there.

Zach Martin

So the fleet's strengths are speed and known operations. Its challenges are real too. Aging equipment, outage scheduling, and the fuel. Uprates can call for slightly higher enrichment or altered core reloads, which raises the value of stable long-term fuel supply. Cameco, across mining and conversion, is a name people watch on that theme, informationally only.

Emily Martin

The capital structure is the part I find striking. Big Tech acts almost like a landlord and underwriter. Multi-decade contracts, milestone-based development funding and prepayments reduce the risk of utility spending. The tenant helps pay for the building.

Zach Martin

For infrastructure capital, the takeaway is discipline. Small improvements to assets you already control can compound faster than a moonshot. New reactors still matter later in the decade. But the bankable story today is the existing fleet.

Emily Martin

Again, this is informational, not investment advice, and the companies we discussed are not recommendations. Thanks for listening to NUCLEAR CAPITAL on I R Hub Radio.

Zach Martin

Real assets. Real ideas. A brighter tomorrow.