Uranium and the Nuclear Fuel Cycle

The question

Is the uranium complex priced as a late-cycle commodity trade that mean-reverts once Kazakh pounds return, or as a decade-long structural deficit in which Western utilities, the DOE HALEU program, and hyperscaler PPAs compete for a pool of pounds that producers spent fifteen years dismantling?

Base rates

The reference class is prior commodity super-cycles in which supply was cut before demand inflected: thermal coal in 2003-2008, iron ore in 2004-2011, lithium in 2020-2022. The incentive price to bring marginal supply online ran 50-100% above spot for three-to-five years, producer equities compounded 25-40% annually through the first four years, and the junior cohort outperformed the majors by roughly two because reserve-life optionality re-priced as the forward curve steepened.

After Fukushima in 2011, Cameco idled McArthur River, KazAtomProm guided 20% below subsoil entitlements, and the Western conversion and enrichment base contracted to a handful of operators. The cumulative production deficit between 2018 and 2025 was roughly 150 million pounds against secondary supply that has now been drawn down. Sprott Physical Uranium Trust, launched in 2021, sequesters spot pounds from the term market at a cadence that producers cannot match with restarts; the trust's net asset accumulation is a structural bid that did not exist in prior cycles.

The imputed forward price embedded in sell-side models is roughly $75-85/lb U3O8 through 2028. The incentive price to bring Athabasca and Namibian tier-two projects online is closer to $90-110/lb. The consensus forward does not clear the incentive curve.

Why consensus is wrong

Consensus models the deficit as a price problem. It is a permitting and capital-cycle problem. A new conventional mine from discovery to first pound runs 12-18 years; an ISR restart on a permitted asset runs 18-36 months. The universe of near-term restartable pounds in the United States is a handful of names, and most of them sit in the cohort we own. When the binding constraint is permitted capacity rather than price, the economics of the junior with a restart-ready asset improve faster than the economics of the major, because the junior captures the entire incremental curve without legacy contract drag.

The second miss is the fuel cycle itself. Enriched uranium for the existing light-water fleet and HALEU for the coming advanced reactor fleet are distinct bottlenecks. The DOE LEU production line at Centrus Piketon and the GAIN Act fuel qualification program have made HALEU a sovereign priority, and Russian enrichment via TENEX is no longer a policy-acceptable source. The market treats enrichment as a service business with modest multiples; it is priced as if the separative work capacity constraint of 2026-2030 were not real.

Third, AI data-center demand has turned nuclear from a stranded-cost problem into a scarce-baseload asset. The Amazon-Talen deal for Susquehanna capacity and the Microsoft-Constellation agreement to restart Three Mile Island Unit 1 are not one-offs; they are the first two contracts in a pipeline of behind-the-meter PPAs that require firm 24/7 carbon-free generation. That demand pulls through to fuel procurement on a five-to-seven year cadence the sell-side has not fully modeled.

Position construction

The book has three 20% anchors and three sub-books.

Anchors (~58.3%). UEC at 20% is the US ISR restart book with permitted production and the cleanest sovereign-pounds exposure. LEU at 20% is the HALEU enrichment monopoly at Piketon, the fuel-cycle name that captures policy dollars regardless of which reactor design wins. NXE at ~18.3% is the Athabasca development asset -- Rook I at Arrow is among the highest-grade undeveloped deposits in the world and owns the tier-one reserve-life optionality in the cohort.

US production restart and development (~28.6%). UUUU at ~18.5% is the conventional mill and rare-earth byproduct optionality at White Mesa, the only operating conventional mill in the United States. DNN at ~9.9% is the Wheeler River ISR project plus a physical uranium holding that compounds with spot. URG at ~2.2% adds the Lost Creek ISR producer with permitted pounds on a faster restart cadence than most of the cohort.

Fuel services and SMR adjacency (~6.9%). LTBR at ~2.3% is the metallic fuel design for existing and advanced reactors, sized as optionality rather than core. EU at ~2.3% is the enCore Energy ISR portfolio across Texas and South Dakota. BW at ~1.3% is the SMR-adjacency position -- BWXT-heritage engineering, nuclear services, and a BWRX-300 and mPower design lineage that sits upstream of utility fleet decisions.

Antimony and critical-minerals adjacency (~5.1%). UAMY at ~5.1% is the domestic antimony and fuel-cycle adjacency; the position is sized to reflect its strategic-minerals optionality rather than direct uranium exposure.

The cohort deliberately excludes CCJ, BWXT, and OKLO; those names are owned in the AI Power Grid book where their utility-fleet and SMR-design exposure is the primary thesis rather than uranium pounds.

Asymmetric payoff

If the Sprott trust continues net accumulation at recent cadence, US utility term contracting runs through 2027 at current volumes, and the DOE HALEU program funds a second Piketon cascade, the weighted book returns roughly 28-42% annualized over three years. If Kazakh production restores to subsoil entitlements earlier than expected and two or three planned reactor restarts slip, the book returns roughly -10% to -20%. If a behind-the-meter nuclear PPA pipeline converts to contracted capacity at the pace of the Amazon-Talen and Microsoft-Constellation precedents, the right tail is 60-90% with multiple expansion on the junior cohort.

At 50% base, 25% bear, and 25% bull, expected value is roughly +22 to +32% annualized against an SPY base rate near +8%. The payoff is asymmetric because the supply curve is locked by permitting lead times while the demand curve has three independent drivers -- existing-fleet relicensing, SMR build-out, and hyperscaler PPAs -- that each contribute pounds demand without requiring the others.

Three things that would change our mind

1. KazAtomProm restoring guidance to full subsoil entitlements with two consecutive quarters of actual production clearing 28 thousand tonnes annualized, signalling the deficit math has a near-term supply release the consensus forward already prices.
2. The Sprott trust entering sustained net redemption -- three consecutive months of unit holder outflows that force physical sales into the spot market -- which would remove the structural bid that differentiates this cycle from prior ones.
3. A congressional reversal of the Russian enriched uranium import ban or a carve-out extension that materially re-opens TENEX supply to US utilities, which would collapse the sovereign-pounds premium embedded in the US junior cohort.

What we are explicitly NOT betting on

We are not betting on any single SMR design -- NuScale, BWRX-300, Natrium, or mPower -- clearing NRC certification on a specific timeline. We are not betting on a particular spot uranium price target. We are not betting on any single Athabasca development asset clearing permitting; NXE is sized to survive a permitting slip. We are not betting on CCJ or BWXT, which live in a different book. We are not betting on lithium, copper, or rare-earth cross-substitution; UUUU's rare-earth optionality is a free option, not a thesis driver. The thesis requires only that the structural deficit persists, that the Sprott trust continues to bid, and that Western utilities keep contracting pounds forward. All three are strictly weaker claims than picking the next Cameco.