Contrarian Codex — WNA report 2026

Intro — a rough two weeks, and the mindset shift goes up a level

• Written after flying home a day early for his grandfather's funeral (the report is dedicated to him); still a full reporting of the conference.
• Last year's bi-annual fuel report "got a snowball rolling" and uranium is up over $15 since. A prominent buy-side participant: the conversation has shifted from terms and pricing to "delivery security."

World Nuclear Outlook — records, but zero Western construction starts

• 2025 generation a record 2,702 TWh (+35), ~9% of world electricity (vs ~17% mid-1990s); fleet capacity factor 83.7% (from 82.9%); 400 GWe net operable. 11 construction starts (9 China, 2 Russia), 3 grid connections, 7 shutdowns — "the entire Western world contributed exactly zero construction starts."
• Tripling needs start rates ~6x current by the mid-2030s — "11 starts against something like 66 required" is a gap policy language cannot close. Zhangzhou 2 took 62 months concrete-to-grid; Rajasthan 7 took 163: serial standardized build is the whole difference.
• 2050 projection 1,457 GWe (vs 1,446); 559 GWe (~38%) exists only as government ambition with no project attached.

The pounds don't need the tripling

• 401 GWe of today's 423 GWe gross is projected still running in 2050 (230 GWe from 60–80-year reactors); mean fleet age 33, 46 reactors past 50, retirement age rising (48.7 vs 47.3), and no age-related capacity-factor decline even past 40 years.
• Under-construction reactors add 54 GWe by 2030 and 28 GWe more by 2035. "Uranium demand does not need the tripling target to be met."
• The report asks governments to expand fuel supply, HALEU/LEU+ and reprocessing — but "the capacity side of this report gets two dozen charts… and the fuel side gets adjectives." Left to the market, it prices "firmly in the triple digits."

Finance Summit — cost of capital as big as cost of construction

• The WNA Director General opened on cost of capital; Cameco's Grant Isaac asked why a 60-year contracted asset is treated differently from a toll road — "Banks price contracted cash flows, they do not price reactors."
• Capital is available ("from whether to when"), but pre-FID money is hardest and first-of-a-kind still needs sovereign de-risking. Amir Adnani's $6 trillion super-cycle figure: "we also need to consider who is saying this."
• ENEC: the fear is hesitation, not technical unknowns. The closing panel wanted a visible project pipeline and a shared developer-investor marketplace.

The floor — sellers hold the strongest hand in "many" years

• An experienced participant: anyone with significant pounds for sale is in the strongest position in many years; sellers dictate terms and some are holding material back.
• Buyers are having a "wake up and smell the coffee" moment, disappointed by the lack of supply response; miners "are not just going to sell their production for $5 over their C1 cash costs," and juniors that contracted too early are a lesson others learned.
• A large US utility fuel manager: "coverage on paper is not the same as coverage in the reactor." A third buyer still anchored to a sub-market price is "your marginal buyer" who will chase term into triple digits.

Duke's RFP and careful buying

• Duke came to market for 400,000 lb/yr for 5 years into the 2030s, with a 300,000 lb/yr, 3-year option — meaningful against today's thin term volumes, and the terms "would absolutely have raised eyebrows a few years ago."
• One buyer would bring large demand to market piecemeal "as to not spark a strong price rally" — itself a tell on market state.

Replacement-rate contracting is not a precondition for price

• 2023's ~160m lb included ~40m lb of the one-off Energoatom deal; 2024 ~110m, 2025 ~116m (part of it the India contract), 2026 tracking under — while annual requirements approach ~200m lb. "All the while, the term price kept climbing, what does that tell you?"
• When volumes return toward replacement, "the effect on term pricing is going to be outsized."

Who held spot up through summer

• Traders, but also utilities and producers, all seeing value in the mid-$80s; the Uzbek-offtake traders who usually slam month-end found buyers "right there waiting." Contacts describe a floor in the mid-$80s that is rising; spot ~$86–87 at writing.
• With base-escalated contracts in triple digits, the trader carry trade (sign a forward sale, buy spot, pocket the spread) keeps spot demand in place — though the window narrows as spot rises.
• Utilities are signing market-referenced floors in the $80s and ceilings at $150+ they refused 18 months ago; Paladin's call lines up (term ~$97, utility talks at $100+).
• The market is opaque by design — RFP responses private, contracts under NDA — so published data points argue with each other.

Inventories: a 2027 cushion, not a 2034 one

• US + European utility inventories ~229m lb, +34m lb from the 2022 low — built by post-Ukraine double-buying and upflexing cheap legacy contracts. Coverage to ~2030 is adequate; after that, 52% of the next 10 years of US demand is uncovered, 186m lb unfilled through 2035 (EIA).
• UxC counts ~38m lb of long-term contracting this year excluding the two India deals. Procurement works 3–10 years ahead: "a utility carrying a hole in 2033-2035 is a buyer today."
• Strategic reserves are becoming structural and price-insensitive; a Chinese contact confirmed China's large inventory will not come to market.

Supply: Arrow is the right tail, the rest are slide decks

• If Arrow delivers ~16m lb by 2033–34 rather than ~30m lb by 2030–31, "a massive shockwave through the market."
• Much of early/mid-2030s supply "comes from projects that do not exist" — not permitted, financed or built; development timelines have stretched from 8–15 to 10–20 years; secondary supply is maybe ~15m lb/yr.
• A sector veteran puts the incentive price well above the high $90s the buy side thinks is enough.

Hyperscalers and BHP — new balance sheets

• Hyperscalers are "price insensitive and ruthless" about powering assets for decades, talking to enrichers (Centrus, possibly GLE) and expected to work down the fuel cycle to physical uranium: "maybe they will buy a mine, these guys don't mess around." Google's 22-year nuclear PPA in Finland cited.
• BHP's M&A representative was on the floor; the land staking was "a minor diversion" and BHP is after "something bigger."

KHNP's 800,000 SWU enrichment tender

• Enough for roughly a quarter of the Korean fleet; his and others' estimates put it at 26–31m lb U3O8-equivalent. Language hints at a bundled response; few can bid at that size.
• Recalls KHNP's earlier uranium tender with a "laughably low" ceiling that drew no offers — a watershed. Korea faces 25–30 GW of data-center/fab-driven demand growth with 79% public support for new nuclear.

Kazatomprom — western utilities back at the table, costs running away

• Management says western utilities are back negotiating on higher prices after a two-year stand-off. (The acid-plant delay was a 28-million-year-old Paraceratherium, not fish.)
• H1 C1 $24.48/lb (+37%), AISC $38.45 (+25%); guidance raised to C1 $25.50–27.00, AISC $39.00–40.50. Acid cost per tonne +46% y/y (~+132% vs 2022). He remains skeptical on acid availability; Zarechnoye depleting faster than plan, Inkai the standout, Karatau/SMCC backwards.

The NNSA RFI — ~40m lb of US-origin uranium

• ~4m lb/yr unobligated, ~1,500 tU as UF6, acquisition possibly from the early 2030s into the early 2040s — against 2.1m lb of total US production in 2025 (basically all Energy Fuels). The government now competes with its own utilities; possibly for naval inventories.
• Who fills it: UEC's licensed capacity vs its sub-70,000 lb quarter; Energy Fuels' White Mesa plus a pipeline that could add 5m+ lb/yr if built; Cameco's Smith Ranch has capacity but not resource. GAO calls NNSA supply adequate into the early 2040s — "Adequate until the need date is precisely why procurement starts a decade early."

SMRs — Rolls-Royce, GE-Vernova, shipping, the IAEA

• Rolls-Royce SMR: "we will buy as many as you can make and more"; first reactor ~2035 with nine more to decade-end; the Dutch PM flew over to get in line. The constraint is capital, not demand.
• GE-Vernova: Darlington is the de-risking event; ~10 units / 3 GW by 2035, maybe 4x by 2040. Fuel standardization is a key risk.
• The IMO secretary general reports significant shipping-company interest in nuclear propulsion — possibly "severely underestimating the demand from this side." The IAEA's RDS-1 has SMRs at 23–28% of new capacity by 2060 (~60% in North America).

Talking to SPUT

• 81.7m lb, never sold or lent; ~7m lb bought this year vs the 9m lb cap (Sprott reads deliveries beyond 12 months as outside the cap). Premium days: 12 of 251 in 2025, 16 vs 46 in Q1, 6 vs 57 in Q2.
• Sale question to institutions: nods start "around $150"; a takeover needs >66% of 344m largely retail units.
• Trading houses are re-staffing uranium desks and bigger balance sheets are showing up on physical trades — deeper markets.

Enrichment and Section 232

• Urenco New Mexico adds 2.1m SWU (~+50%), construction 2029, first production 2032; Orano's $5bn Oak Ridge plant targets early-2030s deliveries — the enrichment valve opens "exactly as uncovered uranium requirements go vertical," a second bottleneck.
• The Section 232 critical-minerals proclamation named uranium but imposed nothing; minimum import prices / tiered pricing / DOE equity-for-supply are available tools, already used on polysilicon. He expects more US strategic reserve buying.

India — a buyer on US/China scale within 20 years

• 24 reactors, 8,780 MWe gross (~3% of power); targets 22.48 GW by 2031–32 (mostly PHWR-700) and 100 GWe by 2047, plus ~$2.1bn for 5+ indigenous SMRs by 2033. Kudankulam adds 4 more VVER-1000s on lifetime Russian fuel.
• The SHANTI Act replaced the 1962 and 2010 acts, letting private Indian companies build, own and operate plants and removing the supplier-liability bar that kept Westinghouse, GE-Hitachi and EDF out — foreign vendors can supply but not own. NPCIL's 220 MWe Bharat Small Reactor tender drew Hindalco, Tata Power, Reliance, JSW, Adani and others.
• PHWRs use ~370,000–410,000 lb per GWe-year (~80% of an LWR): ~3.5–4.5m lb/yr today, ~8.8–9.4m lb in the early 2030s, ~45m lb/yr at 100 GWe. The PFBR reached criticality 22 years after construction began — apply that ratio to every target.
• Domestic ore is high-cost (Tummalapalle ~0.04%, under 1m lb/yr). Stated need 9,000 tU (23.4m lb) over 2025–33, yet Cameco's ~22m lb plus his implied 40–50m lb from Kazatomprom's >$4bn deal is 2–3x that — "before Uzbekistan, before Australia" — a strategic reserve on the Chinese playbook.

Positioning — another third of the cash into DNN and UUUU

• Equity price action is "undeniably frustrating"; sentiment is heading back to depression levels and could go lower as risk assets, but fundamentals should prevail.
• Denison: add back 10% at $2.85 (after selling 20% at ~$4.10 in January). Energy Fuels: back to a full position, plus Jan-27 $15 calls at $1.30 as a small speculative bolt-on. "This is just what I am personally doing right now."