Fueling the AI Nuclear Build: HALEU, TRISO, and the Enrichment Race

Zach Martin

Welcome to Nuclear Capital, right here on KMKT I R Hub Radio. I'm Zach Martin, and with me is Emily Martin. Quick note before we start. Everything we discuss today, including every company we name, is for informational purposes only. None of it is investment advice, and nothing here is a recommendation to buy or sell any security. Okay, Emily, here's the thought I want to open with. You can sign a power purchase agreement in a day. You can pour a foundation in a season. But you cannot run a reactor without fuel, and the fuel for the advanced designs we keep hearing about barely exists at commercial scale yet.

Emily Martin

That's the heart of it, Zach. Most of the conversation about nuclear and AI is about reactors. Who signed with which hyperscaler, which design, which site. But every one of those reactors is a promise that depends on a long chain of industrial steps that happen before anything gets loaded into a core. Mining, conversion, enrichment, fabrication. Today we're staying on that front end of the fuel cycle, because that's where the capital story gets very physical.

Zach Martin

So let's set the table for anyone who hasn't lived in this world. Walk us through the basic difference between the fuel that runs today's plants and the fuel the new designs need.

Emily Martin

Sure. Natural uranium is mostly uranium 238, with only a small slice of uranium 235, the isotope that sustains the chain reaction. Enrichment raises that share. Today's commercial reactors in the United States run on low enriched uranium, which is typically between about 3.5 and 5 percent uranium 235. Many advanced designs want something richer, called HALEU, high assay low enriched uranium. That's between 5 and 20 percent, and the designs we're talking about typically sit toward the upper part of that range. So it's the same element, but a much more demanding product, and today there's very little commercial supply of it.

Zach Martin

And the demand number is not small. The Department of Energy has estimated that commercial advanced reactors will need roughly 50 metric tons of HALEU a year by 2035, and about 500 metric tons a year by 2050. That's a tenfold climb in fifteen years.

Emily Martin

Right, and that's an estimate, so treat it as a planning signal, not a promise. But look at where the starting point is. The existing U.S. fleet needs on the order of 13 million separative work units of enrichment every year. That's the unit the industry uses to measure enrichment effort. And for a long time the United States leaned on Russia for enriched uranium and, effectively, for commercial HALEU imports. Congress passed the Prohibiting Russian Uranium Imports Act in May 2024. It took effect in August 2024, runs through 2040, and has temporary waivers through 2027. So the old supplier is being phased out right as new demand shows up.

Zach Martin

I think of it like building a marathon training plan. You can buy the shoes, hire the coach, announce the race date. But if the nutrition supply isn't there on the long runs, the plan doesn't hold. Reactors are the shoes. Fuel is the nutrition.

Emily Martin

I'd add one more piece, because it's the form of the fuel, not just the enrichment level. Several of these designs use TRISO, which stands for tristructural isotropic. Picture a tiny kernel of uranium, smaller than a poppy seed, wrapped in layers of ceramic and carbon coatings. Each layer is a little containment shell. Those particles then get packed into graphite, either as pebbles or as compacts. It's a very robust fuel form, and that's the appeal. But making it consistently, at volume, with documentation for every batch, is its own industrial challenge.

Zach Martin

Which brings us to a milestone worth talking about. TRISO X, the X energy subsidiary, has completed vertical construction on its TX 1 facility in Oak Ridge, Tennessee. That's the commercial scale TRISO fuel fabrication plant.

Emily Martin

Yes, the outer shell is up, and the project is moving into building completion and interior buildout, which is where the processing equipment goes in. That's an important distinction. A finished shell is progress, but the hard part of a fuel plant is inside. The furnaces, the coating systems, the inspection lines, the safeguards. The Nuclear Regulatory Commission issued a 40 year Category II Special Nuclear Material license for the facility in February 2026, according to the reporting we reviewed. And the company expanded its campus by about 70 acres, to more than 180 acres at the Horizon Center, with an 11 million dollar grant from Tennessee's Nuclear Energy Supply Chain fund.

Zach Martin

And what I find fascinating is that a state is writing a check for fuel infrastructure. That's the capital stack widening. It's not just venture money and utility balance sheets anymore. It's state programs, federal task orders, and customers paying early.

Emily Martin

That's an interpretation, but I think it's a fair one. And it connects to the other half of the X energy story. A fabrication plant is only useful if there's enriched uranium to feed it. So X energy has signed long term HALEU enrichment services agreements with two companies, Centrus Energy and General Matter. Centrus would supply from its plant in Piketon, Ohio. General Matter is working toward enrichment at a site in Paducah, Kentucky. And X energy is providing Centrus with prepayments to help fund the expansion of domestic enrichment.

Zach Martin

Let's slow down on that, because it's a quiet but big idea. Why would a reactor developer prepay an enricher?

Emily Martin

Because enrichment machines are expensive, and an enricher generally wants bankable commitments before it orders centrifuges and builds cascades. A cascade is a connected series of centrifuges that step the enrichment up. If you're a reactor company and the fuel doesn't exist, you can wait, or you can help finance the thing you need. The prepayment turns a customer into a partial funder. Now, I'd be careful not to overstate how widespread that is. The brief we're working from describes it as increasingly common, and X energy and Centrus is the clear example we have. But the logic is easy to follow. Fuel risk is moving onto the demand side.

Zach Martin

And there's a hyperscaler angle, too. The advanced developers in the public offtake pipelines with Meta, Amazon, and Google, names like Oklo, TerraPower, X energy, and Kairos Power, all depend on this front end. So when a tech company says it wants firm, carbon free power for a data center, part of what it's really signing up for is a fuel supply chain it doesn't control.

Emily Martin

Exactly. And let's not forget that TRISO X isn't alone. Standard Nuclear runs a facility in Oak Ridge called SN Zero, at about half a metric ton of TRISO a year. It has substantially completed two more identical facilities, one in Tennessee and one in Idaho, each targeting about one metric ton a year before the end of 2026 under a DOE pilot authorization. And Framatome has formed a joint venture with Standard Nuclear at its Richland, Washington site, targeting two metric tons a year. BWXT is advancing plans for a commercial TRISO facility in Gillette, Wyoming, backed by a 100 million dollar state grant.

Zach Martin

So the TRISO side is crowding up with builders. That's encouraging, but I want to point out the discipline here. Those are targets and plans. The reporting we have describes capacity being built, not capacity that's been delivering at scale for years.

Emily Martin

Well said. Let's turn to Kairos Power, because it shows a different philosophy on the same problem. Kairos uses TRISO in golf ball sized graphite pebbles inside a fluoride salt cooled reactor. And the company has commissioned an automated inspection and laser engraving system at its Manufacturing Development Campus in Albuquerque, New Mexico.

Zach Martin

Tell me what that system actually does, because it sounds like a science fair project and a factory floor had a baby.

Emily Martin

It's a very good picture, honestly. Each pebble gets measured in real time for diameter, geometry, and density. Then high precision lasers scan the outside down to the micron level, looking at the structural integrity of the surface. After that, an infrared laser etches three redundant data matrix codes, similar to QR codes, into the graphite, about half a millimeter deep, without compromising the shell. The code links to a record for that individual pebble. Its dimensions, its density, its batch, and its inspection status.

Zach Martin

So every pebble gets its own passport.

Emily Martin

Yes. And the company says that over time, those markings will help identify different pebble types, provide more accurate fuel consumption measurements, and support material accountability during operations. The near term target is 70,000 inspected pebbles for testing platforms, including its second Engineering Test Unit. One detail matters here. Those test pebbles are surrogates for the hardware tests, so this is about proving the handling and quality systems, not about loading nuclear fuel yet.

Zach Martin

That's the compounding habit idea I love. Small improvements, repeated, with data attached. If you want to run thousands of pebbles through a reactor, you'd better know the story of every single one. Quality control isn't a side note in fuel. It is the product.

Emily Martin

And it's tied to real project milestones. In April, Kairos broke ground on Hermes 2, its power producing demonstration plant in Oak Ridge. It's planned at 50 megawatts electric, under a power purchase agreement with the Tennessee Valley Authority that supplies the grid serving Google data centers in Tennessee and Alabama. Kairos expects operations by 2030. The company has also said it plans to replicate its pebble production line at Los Alamos National Laboratory's Low Enriched Fuel Fabrication Facility to make fuel for the first Hermes reactor, and it announced a collaboration with BWXT to explore commercial TRISO production.

Zach Martin

And there's another fuel side wrinkle in the Kairos story that people rarely hear about. The coolant.

Emily Martin

Right. Its molten fluoride salt coolant, which it calls Flibe, needs enriched lithium 7. Natural lithium has two isotopes, and the lithium 6 has to be removed. Kairos said that enriched lithium 7 isn't currently produced in the United States, so it's building a salt production facility in Albuquerque to close that gap. It's a good reminder that for advanced reactors, the supply chain doesn't end at uranium. Every new design brings its own list of materials that somebody has to be able to make.