Tearing apart the uranium fuel pipeline

Uranium.

Wall Street discovered the nuclear trade about eighteen months ago. Every bank deck from here to Singapore has a slide titled “Nuclear Renaissance” with a chart that goes up and to the right. CNBC books a uranium CEO once a week now. The consensus take: AI needs power, nuclear is clean baseload, uranium goes up. Buy the miners.

They’re not wrong about the destination. They’re wrong about the road. And the road is where you lose the money.

The street is pricing uranium like a momentum stock. It’s not. It’s an industrial fuel with a supply chain so constrained and so time-lagged that most of the people pitching it couldn’t diagram the path from ore body to fuel rod if you spotted them three steps. It’s like buying tickets to a power plant and forgetting to check whether the fuel exists.

Mine production currently covers about 75% of global reactor fuel requirements. The gap — roughly 45 million pounds a year — has been plugged by secondary supplies. Government stockpiles. Utility inventory drawdowns. Enrichment underfeeding. Sources that are depleting every quarter with no replacement anywhere in the pipeline.

The last big buffer, the Megatons to Megawatts program that converted Russian warheads into reactor fuel, ended in 2013. Nobody built the next one.

Strip the story. Here’s the machine.

Global reactor requirements: approximately 180 million pounds U₃O₈ per year. Global mine production: approximately 135 million. The delta is ~45 million pounds, every year, filled by drawdowns from a secondary supply pool that’s been shrinking for twelve straight years.

Now add the demand nobody modeled two years ago. Microsoft signed to restart Three Mile Island Unit 1 — 835 MW of dedicated nuclear capacity for data centers. Amazon has nuclear power purchase agreements across multiple sites. Google is contracting for small modular reactor output that doesn’t physically exist yet. Meta issued proposals for up to 4 GW of new nuclear capacity. Each deal locks in fuel purchasing commitments 5–10 years forward.

Sixty-five reactors under construction globally. China alone is building north of twenty. Each one needs initial core loading — roughly 1 to 1.5 million lbs U₃O₈ for a standard PWR — plus annual reloads around 400,000–500,000 lbs. That’s committed demand on a timeline the mine supply chain cannot match.

New greenfield uranium mine, discovery to first production: 10 to 15 years. Permitting eats four to seven of those. There are very few projects in the pipeline that will meaningfully move global production before 2030. Denison’s Phoenix and Fission’s PLS are advancing, but the aggregate tonnage doesn’t close the gap.

Kazatomprom — the world’s largest producer, ~21% of global mine output — lowered its 2025 production guidance last August. This isn’t a labor dispute you can settle over a table. Their in-situ recovery mining runs on continuous sulfuric acid injection, and the acid supply chain itself is constrained. When the input to the input tightens, output drops. The geology doesn’t negotiate.

The spot market, where roughly 15% of uranium trades, shows $64–$79/lb. The term contract market, where 85% actually changes hands, prices 15–30% higher. Headlines quote spot. Supply decisions run on term. That disconnect is the first thing the Renaissance crowd misses.

Everyone’s watching mine supply. Almost nobody is watching enrichment.

You can’t load yellowcake into a reactor. Mined uranium has to be converted to uranium hexafluoride, then enriched from 0.7% U-235 to 3–5% for light-water service. Each step has its own constrained supply chain. At enrichment, the bottleneck has a Russian flag on it.

Western utilities have been stepping back from Russian enrichment services following Russia’s invasion of Ukraine. But replacement capacity doesn’t build on a sanctions timeline. Urenco and Orano are expanding centrifuge capacity. Five- to seven-year builds. They are not done.

The financial layer — Sprott Physical Uranium Trust, uranium ETFs, the futures market — has absorbed billions from investors who think they own the nuclear renaissance. What they actually own is exposure to a spot price that represents a sliver of the real fuel market. When the enrichment queue backs up, delivered fuel costs will diverge from spot in ways the paper instruments cannot express.

I’ve seen this pattern in other constrained supply chains. The physical bottleneck always arrives later than the bulls expect and hits harder than the bears modeled. You can be right about the thesis for two years and still blow up on the timing. I’ve done it myself.

The Renaissance narrative makes the uranium trade sound simple. Mine more, price goes up, everyone wins. But between the mine and the reactor sit conversion plants, enrichment cascades, and fuel fabrication facilities — each one a valve that can close. Right now, the enrichment valve is the one nobody’s checking.

Walk the sequence.

A utility with a reactor coming online in 2028 needs enrichment services contracted now. If Russian SWU is politically off-limits and Western capacity is sold forward through 2030, that utility has three options. Pay a massive premium for whatever separative work exists. Delay the reactor start and eat billions in construction cost overruns. Or scramble for underenriched material on spot and hope for the best.

Option one reprices the entire term market. Option two creates deferred demand that hits later and harder. Option three is what happened to European gas utilities in 2021. I don’t need to describe how that ended.

Where capital goes: uranium-focused producers and advanced developers with significant Western reserves — Cameco, Paladin, NexGen, Denison. Companies whose valuations are mechanically tied to term-contract repricing. Physical uranium vehicles on any spot dip, because the floor under spot rises every quarter that secondary supplies deplete.

Where capital doesn’t go: diversified miners where uranium rounds to zero in the revenue mix. Broad nuclear ETFs stuffed with engineering firms and utility holdcos that give you the narrative without the mechanical exposure.

The street has the thesis right. Nuclear is coming back. But they’re trading the bumper sticker. The actual trade is in the fuel pipeline — the mine-to-reactor chain that is structurally short at every single link. Spot at $71 isn’t the ceiling. It’s not even the incentive price for new greenfield development.

The enrichment bottleneck hasn’t started to bite yet. When it does, the repricing won’t be gradual. These things never are.