TL;DR
The physical buildout is accelerating faster than the equity complex can process: Amazon signed a 20-year, 690 MW Calvert Cliffs deal with Constellation, the DOE lined up a $4.2 billion uprate loan for Vistra, the NRC issued the first-ever US construction permit for a BWRX-300, and Korea committed a framework worth $120 billion toward Westinghouse reactors on US soil — all inside a single week. Yet pre-revenue SMR developers made new lows into that news barrage. The theme has split in two: capital (hyperscaler, federal, and foreign) is flooding into the existing fleet, while public equity demands cash flow, not frameworks.
Hyperscaler Capital Chooses the Existing Fleet — Again
The uprate model that Google pioneered at Hatch/Vogtle is now a pattern, not an experiment. Amazon and Constellation signed a 20-year agreement covering 690 MW at Calvert Cliffs, funding "approximately 190 megawatts in total of new, emissions-free generating capacity coming online between 2030 and 2032" plus "more than $3 billion in Maryland infrastructure investment" (Constellation press release). Constellation's Dardis put the scale in industry terms: "690 megawatts is a big contract for us — and generally speaking, in the industry" (Maryland Matters). The revenue certainty is explicitly what unlocks relicensing and future new builds at the site — and per the Wall Street Journal, Amazon is already exploring new reactors at Calvert Cliffs (WSJ).
Two hyperscaler-funded uprates in under two weeks means tech buyers have found the cheapest, fastest firm-clean megawatts available: reactors that already exist. The federal government is chasing the same megawatts — see below.
What to watch: Whether a third hyperscaler pairs with a different utility on the same template before year-end.
Federal Money Joins the Uprate Trade
DOE capital is now explicitly funding the same strategy as the hyperscalers. Reuters reports the US will lend Vistra about $4.2 billion to uprate at least three of its four nuclear stations — work that "can be done without requiring new licenses from the Nuclear Regulatory Commission" (Reuters via Mining.com). Energy Secretary Chris Wright is scheduled to announce the loan Monday at Vistra's Perry plant on Lake Erie, and the report situates it in a demand-driven push where Wright has said the bulk of DOE loan office capacity "will support nuclear power plants" (Reuters via Mining.com).
VST rose 4.93% on the report, closing at $140.02 — still 33.2% below its 52-week high (Vistra market view). This is the third and most demand-proximate tranche of DOE Energy Dominance Financing, and it confirms the direction: public capital is buying the cheapest megawatts, not just first-of-a-kind builds.
What to watch: Monday's formal announcement — final terms and which stations are named.
Korea Puts a Price Tag on the AP1000 Pipeline
The largest new-build signal of the period is foreign-financed. On September 30, the US and Korea announced a framework under which Korea anticipates investing $120 billion toward "up to eight large nuclear reactors in the United States, including six Westinghouse AP1000® reactors and up to two Korean APR1400 reactors," deployed on federal sites beginning with two AP1000s (Cameco release). Each APR1400 built under a one-time IP-settlement waiver would generate "approximately US$2 billion of value per APR1400 reactor" to Westinghouse, and the framework includes a potential Korean equity investment in Westinghouse (Cameco release).
The caveats matter: everything is non-binding, and the market has priced none of it in — CCJ closed at $85.18, down 6.9% YTD and 36.5% below its high (Cameco market view). But it stacks with DOE's supply-chain conditional commitment, under which Cameco says AP1000 timelines "are expected to accelerate by as much as three years."
What to watch: Definitive agreements or named federal sites — the framework is worth nothing until one of those lands.
The NRC Turns Paper Into a Buildable Reactor
The regulatory bottleneck cracked, at least for one design. On September 29, the NRC issued the construction permit for TVA's Clinch River Unit 1 — a 300 MW BWRX-300 — "the first US construction permit ever issued for a BWRX-300," and per Neutron Bytes "only the second advanced reactor construction permit ever issued in the nation" (ANS; Neutron Bytes). The commercial caveat from World Nuclear News: "although no construction date has been announced, it is now the first US utility permitted to construct the BWRX-300 small modular reactor" (World Nuclear News).
This gives GE Vernova Hitachi's design a first-mover regulatory advantage over US developers still in pre-permit stages, including Oklo and Nano Nuclear — and TVA keeps consolidating its role as the deployment bellwether.
What to watch: TVA naming a construction date or a financing partner for Clinch River.
Uranium Makes History While the Miners Bleed
The term market keeps setting records with no equity bid underneath. Sprott confirms "the long-term uranium price increased to an all-time high of $96/lb in August, surpassing the previous cycle high reached nearly two decades ago," while spot crossed $90 shortly after month-end (NucNet). Spot remains well below its $107 cycle peak and the 2007 record of $136 — meaning the market where utilities actually contract is the one making history (NucNet).
Cameco still can't catch a bid: $85.18, RSI 27, after a Q2 with net income down 92.1% year over year and a temporary Cigar Lake suspension (Cameco market view). The Korea framework and the uprate wave are exactly the demand signals that should tighten term contracting further — the metal and the miner are diverging harder than ever.
What to watch: Year-end contracting volume at record prices, and whether Q3 earnings (early November for CCJ) explain or deepen the disconnect.
The Equity Split Is Now the Story
The market is no longer trading "nuclear" as one asset class. Into the most policy-positive week the theme has seen, NuScale closed at $7.75 (down 85.5% from its high), Oklo at $35.87 (down 79.4%), and NNE at $15.62 (down 72.4%) — all making new lows — while Constellation gained 9.34% in a quarter where utilities fell 13% (NuScale/Oklo/NNE market views; CEG market view
+1). The macro explains the mechanism: a 5.24% 10-year Treasury against a 3.75% fed funds rate is punishing pure duration, and pre-revenue developers are exactly that (macro view
).
This revises last cycle's "first tentative stabilization" read — it didn't hold for the developers. Meanwhile NNE is buying real assets: a definitive $13.5 million deal for Radnostix's NRC-licensed fuel processing facility, which would make it "one of ten NRC-licensed fuel cycle facilities in the United States" (NNE press release) — a rare developer move the market can verify.
What to watch: Q3 earnings calls from SO, CEG, VST and CCJ for hyperscaler demand and uprate commentary.
What surprised us
- The uprate consensus formed in weeks, not years. Google (Sept 21), Amazon (Sept 30), and the DOE (Oct 5) all converged on squeezing more from the existing fleet in the same fortnight — the fastest capital-alignment this theme has shown (Finding: Amazon PPA
+1; Finding: Vistra loan
).
- New lows into the best news week on record. The developers didn't just lag — they hit 52-week lows (NuScale at 0.3% of its range) while an NRC permit, a $4.2B loan, and a $120B framework all printed. The market is telling you frameworks are worthless until they're revenue (Finding: equity split
).
- The Saudi vote got scheduled, not shelved. Senators secured a floor vote before December 13 — after the midterms — keeping Westinghouse's potential anchor Middle East customer in limbo through the election (Reuters).
- NNE's Radnostix deal is the first developer move that buys a regulatory shortcut instead of a headline. Amending an existing NRC license (SUB-1011) is a materially faster path than greenfield licensing (Finding: NNE fuel cycle
).