Uranium's supply gap is bigger than the market thinks

In 1954, Lewis Strauss, then chairman of the Atomic Energy Commission, told a room of science writers that electricity from nuclear fission would one day be 'too cheap to meter.' He was mocked for it, and rightly so, but read the speech carefully and Strauss wasn't actually predicting free power. He was predicting that the fuel cost would become a rounding error against the capital cost of the plants themselves. On that narrower point, he was correct. A pressurized water reactor in 2026 spends roughly 5% of its operating budget on the uranium that goes into it. The other 95% is steel, concrete, labor, and the long tail of regulatory paperwork.

That 5% figure is why nobody in nuclear engineering loses sleep over uranium prices. A utility signing a 20-year power purchase agreement with a hyperscaler doesn't care whether U3O8 trades at $80 or $120 per pound. What they care about, deeply, is whether the pounds will physically be there in 2031, 2034, and 2039 when the fuel assemblies need to be loaded.

And that question, the availability question, has become the most important unsolved problem in next-gen energy.

Here's the situation as of this month. Global reactor demand for uranium is running around 180 million pounds per year. Primary mine supply is roughly 145 million. The gap, about 35 million pounds, has been filled for the last decade by secondary supplies: decommissioned weapons material, government stockpiles, reprocessed tails, and inventory drawdowns from utilities themselves. Those buckets are nearly empty. The Megatons to Megawatts program ended in 2013. The DOE stockpile sales have slowed to a trickle. Russian enrichment exports are restricted under the 2024 Prohibiting Russian Uranium Imports Act. And the utility inventories that cushioned the market through the 2010s have been drawn down to the point where most US utilities now hold less than two years of forward cover, against a historical norm of three to four.

Now add the demand side. Vistra signed its data center deal with a hyperscaler last year. Constellation reopened Three Mile Island Unit 1, with the entire output contracted to Microsoft through 2044. Amazon bought the Talen campus next to Susquehanna. Oracle has publicly said it's evaluating SMR deployments alongside its Stargate buildout. None of this existed in the demand models five years ago. The World Nuclear Association's 2025 Nuclear Fuel Report revised its 2040 reactor uranium demand upward by 28% from the 2023 edition. That's not a forecast tweak. That's an admission that the model was broken.

And the supply response? Cameco is running flat out. Kazatomprom downgraded its 2025 production guidance citing sulfuric acid shortages and wellfield development delays. NexGen's Rook I project in Saskatchewan is permitted but won't pour first pounds until 2028 at the earliest. Paladin's Langer Heinrich restart in Namibia has been hit by water supply issues. There are perhaps a dozen genuine new mines globally that could matter to the supply picture this decade, and almost all of them are years behind their original timelines.

This is the part where most investors reach for the obvious trade, which is to buy the miners. Cameco, Denison, NexGen, Paladin. And those are fine businesses. But the miners come with operational risk, capital expenditure cycles, geopolitical exposure (a lot of the world's uranium sits in Kazakhstan, Niger, and Namibia, which are not boring places), and the constant overhang of equity raises to fund development.

There is a different way to play this. A way that gives you exposure to the pounds without the drill bits and the diesel and the dust. A structure that was used for decades in gold and silver and oil to brilliant effect, and which now exists in uranium in exactly one publicly listed vehicle.

The royalty model. Pay upfront for a slice of future production from a portfolio of mines. Take a piece of the pounds, or the revenue, in perpetuity. No mining costs. No capex blowouts. No payroll. Just exposure to whatever the market price of uranium happens to be, multiplied by however many pounds the underlying operators actually produce.

In the gold sector, Franco-Nevada (NYSE:FNV) used this template to compound shareholder returns at roughly 16% annually for two decades while the average gold miner returned almost nothing. Wheaton Precious Metals did the same in silver. Both companies trade at premium multiples because the model genuinely deserves them: low overhead, no operating leverage to costs, pure exposure to price and volume.

There is one company doing this in uranium. The market cap is under $1 billion. Most generalist energy investors have never heard of it. And it has been steadily building a portfolio of royalty interests across some of the most important uranium projects in North America while everyone else was busy debating whether nuclear was making a comeback.