The uranium operator's checklist: separating the mines from the narratives

Uranium has traded on narrative for almost two decades. The narrative is real — nuclear restart, AP1000 build, data-center power demand, post-Fukushima inventory exhaustion — but the narrative is not the business, and the operators worth owning are read on the mines, not the slides.

The Halvren uranium checklist is short. Six questions. Four of them eliminate most uranium names without controversy. The remaining two are where the work happens.

1. Does the operator produce, or is the operator a developer with a press release?

The single most important sorting question. Uranium developers have outnumbered producers for a decade. Many of them have spent more on investor relations than on drilling. The check is mechanical: in the most recent reporting year, did the operator produce more than two million pounds of U₃O₈? If yes, the operator is a producer. If no, the operator is a developer, and the framework for reading a developer is different (resource quality, permitting status, capital plan against the equity base, dilution velocity).

Most pieces written about "the uranium opportunity" are written about developers. Most uranium investors lose money on developers. Both facts are reconcilable.

2. What does the per-pound cash cost look like, audited, at the worst recent price?

For producers, the cash cost question is the structural read. We track the all-in mining cost — operator-reported, audited annually — and check it against the trough uranium price of the last decade (~US$22/lb in 2016–2017 for spot, with realized prices lagging meaningfully because of long-term contract floors).

Cameco (CCO) at McArthur River and Cigar Lake reports per-pound cash costs in the low US$20s, with sustaining capex layered on top. The McArthur River shutdown from 2018–2022 was the cleanest single capital-discipline decision in modern uranium history: the operator mothballed the world's highest-grade uranium mine rather than produce into a depressed price, then restarted it when contract prices supported the economics. The marginal Canadian and African producers that did not mothball destroyed value on every pound.

Kazatomprom's in-situ recovery in southern Kazakhstan reports the lowest disclosed per-pound cash costs in the cohort. The operator is the largest source of structurally cheap pounds. The honest reader has to engage with the Kazakh cost structure because the price-deck math runs through it; ignoring Kazatomprom because it is uncomfortable does not change the marginal-tonne dynamics.

3. What is the contract book, and what does the realized-price lag look like?

Uranium is sold predominantly under long-term contracts with three- to ten-year terms, market-reference pricing, floors, and ceilings. The reported selling price of any uranium operator trails the spot tape on the way up, and trails it on the way down. The lag is a feature, not a bug, if the operator's average contract terms are favourable.

Cameco's long-term contract book and average realized price are the single most important numbers in the quarterly release, in our read. A rising contract book at rising average prices is the clean signal that the business will earn meaningfully more over the next three to five years, regardless of what spot does next month. A flat contract book at falling average prices is the opposite read.

We track these numbers every quarter for the senior uranium producers that disclose them. Cameco, Kazatomprom, and Orano report enough detail to make the read possible. Most of the rest do not, which limits how confidently any of those names can be underwritten.

4. What happens to the operator if the spot price falls 40%?

The fourth check is a stress test. Uranium spot has fallen by 40% from a recent peak in three separate windows over the last fifteen years (post-Fukushima 2011, the 2014–2016 cycle, and a smaller 2018 reset). The check is: at the operator's current cost structure, contract book, and balance sheet, does a 40% spot drawdown leave free cash positive?

For Cameco, the answer is yes, with the contract book providing a multi-year buffer and the Westinghouse contribution adding non-spot-correlated EBITDA. For the marginal producers, the answer is no, and the response in past windows has been a mix of mine mothballing, equity issuance, and corporate restructuring. The 2014–2016 cohort attrition was substantial.

The check is not whether the spot will fall 40%; it is whether the operator has a business if it does. The first question is unanswerable. The second is a fact-pattern.

5. What is the operator's reinvestment record?

Capital allocation matters more in uranium than in most mining sectors because the cycle is so volatile that the temptation to over-build at the peak is unusually strong. The 2007 cycle peak left a generation of half-built mines stranded for fifteen years. The 2024–2025 cycle is the second uranium cycle in living memory where producers have spent meaningful capital on capacity additions; the test of whether the additions earn their cost of capital is still in front of us.

We read the reinvestment record by tracking, over a ten-year window, the operator's cumulative capital expenditure versus the cumulative free cash flow over the same period. For Cameco, the post-2018 record is exemplary: capital was deployed selectively, the McArthur restart was disciplined, and the Westinghouse acquisition was the major strategic bet whose verdict is still being written but whose early returns are encouraging. For the rest of the cohort, the records are more uneven and, in some cases, dependent on the next two years of spot prices to look good.

6. Is there a non-mining EBITDA contribution, and is it real?

The sixth check is specific to a small subset of uranium operators. Cameco's 49% stake in Westinghouse, Energy Fuels' rare-earth and vanadium contributions, certain royalty-and-streaming exposures: these are non-mining EBITDA streams that can materially change how the operator earns through the cycle.

Westinghouse is the most important single example. The 2023 acquisition added roughly C$170–200M of annual adjusted EBITDA contribution on Cameco's 49% share in 2025, with line of sight to growth as the AP1000 pipeline in Poland, Bulgaria, Ukraine, and the US converts from talking points to construction. The unsettled question is the durable mix between fuel and services (high-margin, repeatable) and new-build engineering (project-driven, lumpy). The answer changes the appropriate multiple by several turns. We do not have a confident answer yet.

Uranium has been a narrative asset. The mines that survive the cycle are operating businesses. Reading them as the second thing is the only honest way to underwrite the first.

The developer trap, in three filings

The uranium developer cohort has a recognizable pattern in the regulatory record. Three filings tell the story.

The first is the prospectus or the annual information form. A developer's prospectus typically describes a deposit, a metallurgical study, and a path to production. The path to production reliably involves another two to four years of permitting, engineering, and capital deployment. The reader who is paying attention notes the timeline carefully; it will slip. Almost every uranium developer's published timeline has slipped by years against the original schedule. The slippage is structural in the regulatory environment for nuclear-grade uranium production, not a flaw of any single operator. A developer whose deck claims first production in two years should, in our framework, be read as a developer whose first production is four or more years away.

The second is the equity-issuance history. Developers raise equity. The cadence and the pricing are public. A clean developer raises equity at the milestone moments — preliminary economic assessment, feasibility study, permit approval — and at prices consistent with the disclosed value-creation. A less clean developer raises equity at every share-price weakness, regardless of project progress, with the per-share dilution accumulating quietly. We pull the share-count history for every uranium developer that has been on the public market for at least five years. The cumulative dilution number, in many cases, exceeds 200% over five years. The producer cohort, by comparison, has share counts that are flat or modestly growing.

The third is the management compensation disclosure. A producer's compensation plan ties bonuses to production volumes, unit costs, safety statistics, and per-share metrics. A developer's compensation plan, in many cases, ties bonuses to milestone achievement — completion of a feasibility study, receipt of a permit, achievement of a financing. The milestone bonuses can be substantial relative to the operating cash flow of the developer (which is, by definition, near zero), and the milestone structure has historically been a poor proxy for ultimate shareholder value creation. Many developers have achieved every milestone and never produced an economic ounce. The framework's bias toward producers is, in part, a bias toward this compensation structure being aligned with what the shareholder actually wants.

The combined effect of the three filings is that the uranium developer cohort is, in our framework, structurally lower-quality than the producer cohort, with the exceptions being developers operating in supportive jurisdictions, on already-permitted ground, with disciplined equity issuance and per-share-aligned compensation. The exceptions are rare. The producer cohort, with the same six questions applied, is also small. Both observations are deliberate; the desk universe is built to be small.

What the checklist eliminates

Run the six questions across a representative uranium developer or junior producer. The first eliminates most of the universe. The second narrows further. By the time you have run all six, the remaining names are a short list: Cameco, Kazatomprom, Orano, BHP's Olympic Dam (a copper mine with uranium by-product), and a handful of post-restart North American operators whose checklist passes are not yet supported by multi-year operating data.

That list is shorter than the universe of names being discussed at any given uranium conference. It is also the list whose share-price performance over the last cycle has been less spectacular than the developers'. The first observation is structural; the second is the consequence of an entire investor cohort having confused operators with options. Both can be true. The Halvren framework rewards Pillar I and II over the Pillar III commodity bet; that bias produces the smaller universe and the more patient compounding.

The compare for Canadian Natural (CNQ) in oil and gas, with its sustaining-cost discipline and dividend record, is closer to Cameco than it sounds. Both are operating businesses in narrative commodities. The narrative is the part that prices the stock in any given quarter. The operating business is the part that earns the position over a decade. The reader who can hold the second sentence steady through the first is the reader the framework is written for.