Everyone is buying reactor stocks. The reactor cannot run without the fuel. The bottleneck model scored the entire nuclear supply chain. The fuel chain won.
Every retail investor is buying reactor stocks. NuScale. Oklo. Nano Nuclear. X-Energy. The thesis is simple: AI needs power, power needs nuclear. The narrative is correct. But the trade is pointed at the wrong part of the supply chain.
Advanced reactors need fuel. The fuel needs enrichment. That enrichment capacity does not exist at commercial scale in the United States. Not my opinion. The Department of Energy said it explicitly.
The real constraint in the nuclear buildout is not the reactor design. It is not the NRC licensing timeline. It is the fuel chain sitting three steps upstream of every reactor on every drawing board.
I scored twenty-three publicly listed companies across the nuclear and uranium supply chain on five variables: dependency, irreplaceability, concentration risk, qualification difficulty, and time-to-scale. The fuel chain won by a wide margin.
| # | Nuclear supply chain node | Why it ranks this high | Score | |
|---|---|---|---|---|
| 1 | HALEU Enrichment | No commercial-scale U.S. supplier. One DOE-backed operator. No practical substitute for advanced reactor fuel. | 100.0 | |
| 2 | TRISO Fuel Manufacturing | One U.S. company makes irradiation-tested TRISO at production scale. Every high-temperature reactor design needs it. | 100.0 | |
| 3 | Domestic Uranium Processing | White Mesa Mill is the only fully licensed and operating conventional uranium mill in the U.S. Has processed roughly two-thirds of domestic uranium since 2017. | 85.0 | |
| 4 | Uranium Supply at Scale | Kazakhstan is 45% of global supply. Only 8% of U.S. reactor purchases were domestic-origin in 2024. That concentration is a real vulnerability. | 52.7 | |
| 5 | Advanced Reactor Platform | Licensing complexity and fuel integration push the score up. But commercial deployment is early 2030s at best. Everything above this row has to work first. | 51.0 |
HALEU is uranium enriched to between 5% and 20% U-235. Every advanced reactor on the drawing board needs it. The DOE has said explicitly it is not available from domestic suppliers at commercial scale.
Until recently, the only commercial source was Russia through Rosatom. The U.S. response was the HALEU Availability Program and a DOE contract with Centrus at its Piketon, Ohio facility, supporting 900 kg per year. A single commercial advanced reactor will need significantly more than that annually at full operation.
People talk about this as a technology risk. It is not. The enrichment cascade works. The problem is scale. Building enrichment capacity takes years, requires NRC licensing, and the equipment is not something you can order off a shelf. Centrus has the only operating licensed HALEU facility in the country.
TRISO is not a variation on existing nuclear fuel. Each particle is a uranium kernel wrapped in ceramic coating layers that trap fission products under extreme heat. It is what X-Energy's Xe-100 runs on, the reactor Amazon has backed, currently under construction permit review for the Dow Seadrift facility in Texas.
BWX Technologies is the only U.S. company that manufactures irradiation-tested TRISO fuel using production-scale equipment. Already demonstrated, including delivery for Project Pele, the military's mobile microreactor program. Getting qualified to make this stuff is not quick; testing and regulatory approval alone take years.
BWXT also supplies the existing fleet and defense nuclear programs, so it has real revenue today, something the pure-play reactor names cannot say. X-Energy's TRISO-X subsidiary holds a 40-year NRC Part 70 HALEU fuel fabrication license, which makes it the second meaningful player. But BWXT built the track record first.
Before uranium can be enriched it has to be processed. Energy Fuels' White Mesa Mill in Utah is the only fully licensed and operating conventional uranium mill in the United States. It has processed roughly two-thirds of all domestically produced uranium since 2017.
White Mesa is also building out rare earth processing alongside uranium, which gives it a role in the critical materials supply chain well beyond just uranium. One licensed operating facility. Nothing else in the country does what it does at commercial scale.
U.S. uranium production was around 1.7 million pounds in 2023. At the 1980 peak it was 43 million. The DOE uranium reserve, Section 232 investigations, and critical minerals policy are all pushing in the same direction now. White Mesa is sitting right in the middle of that reversal.
Kazakhstan supplies 45% of global uranium. U.S. reactor operators bought 55.9 million pounds in 2024. Of that, 4.3 million pounds were domestic origin, roughly 8%. The other 92% came from abroad, with meaningful exposure to countries that are either geopolitically complicated or outright adversarial.
Cameco is the Western world's most credible large-scale uranium supplier. It produced 21 million pounds in 2025, owns 49% of Westinghouse, and operates at a scale that junior developers simply cannot touch. Its time-to-scale score of 5 is the highest in the whole analysis because it is already producing and selling at commercial volume.
The domestic ISR producers are a different story but an important one. They are rebuilding U.S. primary production from scratch. UEC started Burke Hollow in 2026. Ur-Energy turned on Shirley Basin in April 2026. enCore is running three licensed ISR plants in Texas. Individually none of them is irreplaceable. Together they matter a lot for where this country is trying to go on energy security.
X-Energy, Oklo, NuScale, Nano Nuclear. None of this is wrong as a long-term thesis. Advanced reactors will get built and AI data centers genuinely need power. The problem is timing. The first commercial advanced reactor in the U.S. is an early 2030s event at the optimistic end. The fuel chain problem is happening right now.
NuScale has the most solid regulatory foundation, NRC completed its US460 standard design review in May 2025. Oklo has the most ambitious technology, a fast reactor that could theoretically run on spent fuel. X-Energy has the most complete package: a reactor design, a licensed fuel fabrication facility, a construction permit application in process, and Amazon backing it.
The scores tell the honest story. Oklo is 37.4. NuScale is 31.8. Nano Nuclear is 4.8. A name that scores 4.8 can still go parabolic on hype and momentum. I have seen it happen. But that is a momentum trade, not a bottleneck trade. Worth being clear with yourself about which one you are sitting in.
Same sector. Different position in the supply chain. That gap is the trade.
| Topic | What the market sees | What the bottleneck model sees |
|---|---|---|
| AI power demand | Buy reactor names. Nuclear is the clean power solution for data centers. | Reactors need fuel that does not commercially exist yet. The real constraint is three steps upstream of the reactor. |
| SMR licensing | NRC milestones are the key catalyst. Own the licensed designs. | A licensed reactor design without a fuel supply is a blueprint, not a business. |
| Uranium price | Buy uranium miners. Spot price movement drives the trade. | Domestic processing and enrichment are the scarcer assets. Mining is less differentiated than it looks. |
| NNE and hype names | High retail momentum. Parabolic moves on any headline. | 4.8 on the bottleneck model. No commercial revenue. No near-term constraint position. |
| CEG / VST PPAs | AI power trade. Microsoft and Meta signing long-term nuclear deals. | Correct beneficiary position. These names benefit while advanced reactors are still being built. |
| BWXT / LEU / UUUU | Less exciting than reactor names. Slower moving. Institutional. | The three highest-scoring bottleneck positions in the entire analysis. Real revenue, real constraints. |
Every ticker mapped to where it sits in the supply chain. Not a buy list. Just a clear picture of who owns what part of the constraint.
| Ticker | Score | Bottleneck role | Type |
|---|---|---|---|
| LEU | 100.0 | Only U.S. HALEU producer; DOE contract operator | Owns bottleneck |
| BWXT | 100.0 | Only U.S. production-scale TRISO manufacturer | Owns bottleneck |
| UUUU | 85.0 | White Mesa, the only U.S. conventional uranium mill operating today | Owns bottleneck |
| CCJ | 52.7 | Western uranium supply anchor; Westinghouse stake | Scale supplier |
| XE | 51.0 | Xe-100 + TRISO-X fuel license; Amazon-backed | Platform disruptor |
| ASPI | 47.8 | Alternative enrichment tech; potential HALEU disruptor | Bottleneck disruptor |
| OKLO | 37.4 | Fast reactor; fuel recycling; DOE design approval | Reactor disruptor |
| LTBR | 32.4 | Metallic fuel for existing reactors; bypasses HALEU need | Demand-side disruptor |
| SMR | 31.8 | NRC-certified design; strongest regulatory proof in SMR | Reactor disruptor |
| IMSR | 31.8 | Molten salt reactor; CNSC Phase 2 cleared | Reactor disruptor |
| GEV | 19.7 | BWRX-300 SMR + grid turbines + nuclear services | Nuclear/grid hybrid |
| UEC | 17.2 | Two operating U.S. ISR uranium platforms | Domestic uranium |
| URG | 17.2 | Lost Creek + Shirley Basin; operating April 2026 | Domestic uranium |
| EU | 17.2 | Three licensed ISR plants in Texas; two operating | Domestic uranium |
| NXE | 17.0 | Rook I licensed for construction March 2026 | Uranium developer |
| CW | 15.9 | Reactor coolant pumps; control rod drives; AP1000 | Components supplier |
| DNN | 15.1 | Phoenix ISR FID Feb 2026; production target mid-2028 | Uranium developer |
| BEP | 11.0 | 51% Westinghouse ownership via Brookfield | Indirect exposure |
| CEG | 10.5 | Largest U.S. nuclear operator; Microsoft TMI PPA | Power now |
| VST | 10.5 | 2.1 GW Meta nuclear agreements; AI power play | Power now |
| FLR | 10.2 | EPC for Centrus enrichment + X-Energy Dow project | EPC contractor |
| NNE | 4.8 | KRONOS MMR; University of Illinois permit filed | Early stage / hype |
| BIP | 1.8 | Too indirect; Westinghouse exposure cleaner via BEP | Avoid |
ETFs: URA (broad sector confirmation) · URNM (uranium miner sentiment) · URNJ (junior uranium beta gauge)
The drone issue made one argument: the hardest component to replace matters more than the product it goes into. The grid issue made the same argument for transformers and core steel. Nuclear makes it more starkly than either of those.
Three names sit at the center of this. LEU owns the bottleneck today. ASPI is the bet that a different enrichment technology breaks the supply-side constraint. LTBR is the bet that better fuel for existing reactors reduces how much new enrichment capacity the market actually needs. One chokepoint operator. Two disruptors coming at it from opposite directions.
The reactor names will have their run. Advanced nuclear is getting built. But the money in this trade moves upstream, not downstream. Fuel first. Enrichment before reactors. The constraint before the product. I ran the model across the whole sector. The bottleneck is not the reactor.