Westinghouse

Phase A — Understand the business

Lens 1 · Company Overview

Westinghouse is the original-equipment manufacturer (OEM) to more than half the global nuclear reactor fleet and a services provider to roughly half of the world's operating nuclear generation. Founded 1886; it built the first US commercial reactor (Shippingport, 1957) and is the progenitor of the pressurized-water reactor (PWR) — the dominant reactor architecture worldwide. The business today runs on three legs:

1. Operating Plant Services & Parts (the annuity). Outage services, maintenance, engineering, instrumentation & controls (I&C), and spare parts for the installed PWR fleet. Long-term service agreements and refuelling outages recur on a fixed clock and are effectively non-discretionary for any utility running a Westinghouse plant. This is the cash engine.
2. Nuclear Fuel (the annuity's twin). Fuel-assembly fabrication at Columbia, SC (US); Springfields (UK) — 80+ years of continuous fuel manufacture; and Västerås (Sweden). Critically, Westinghouse is the only supplier of a fully Western (Russia-independent) VVER fuel for Russian-designed reactors across Eastern Europe and Ukraine (Lens 2/3).
3. New Plant / AP1000 (the option). The Gen-III+ AP1000 large reactor (1,117 MWe), plus derivatives AP300 (300 MWe SMR) and eVinci (5 MWe heat-pipe microreactor). This is the growth/optionality leg — lumpy, project-based, high-beta to policy.

Revenue mix. Total 2025 revenue ≈ US$5.0B. The "core business" (services + fuel + parts + engineering) was ≈ US$4.3B, ~84% of 2025 revenue, with recurring fuel + services described as ~50% of earnings — the annuity that makes a project company bankable. New-plant/AP1000 is the remaining ~16% of revenue but nearly all of the upside variance.

Contract structure. Two very different books: (a) recurring, high-visibility, high-margin services/fuel contracts (multi-year, outage-cycle-driven, sticky); and (b) milestone-based EPC/engineering contracts on new builds (Poland, Bulgaria, Ukraine, Czech Dukovany participation) that recognize lumpy revenue — the Q2 2025 spike was a Dukovany milestone (Lens 5).

Lens 2 · Supply Chain

Westinghouse sits in the middle of the nuclear value chain — it is both a buyer of enriched fuel/heavy components and a seller of fuel + reactor systems + services. Named stakeholders:

Upstream (inputs to Westinghouse):

• Uranium & conversion: ties to owner Cameco (uranium miner/converter — a vertical-integration logic behind the 49% stake); enrichment from Urenco, Orano, Centrus (and historically Russia's Rosatom/TVEL — the dependency Westinghouse's Western VVER fuel is displacing).
• Heavy forgings / reactor pressure vessels & components: Japan Steel Works, Doosan Enerbility (Korea), and a deliberately localized AP1000 supply chain — Westinghouse has placed first purchase orders with Bulgarian suppliers (MTG-Dolphin AD, Aug 2025) and Polish suppliers (Oct 2024) to build indigenous content.
• EPC / construction partners: Bechtel (US + Poland), Hyundai Engineering & Construction (Bulgaria Kozloduy).

Westinghouse → downstream (end customers):

• Utilities running Westinghouse PWRs worldwide (services + fuel annuity) — e.g. Southern Company (Vogtle 3&4, Georgia), US/European/Asian PWR operators.
• New-build sovereign customers: Poland (PEJ) — 3 AP1000 at Lubiatowo-Kopalino; Bulgaria (Kozloduy NPP New Build EAD) — 2 AP1000; Ukraine (Energoatom) — up to 9 AP1000 + all VVER fuel; US utilities/data-center offtakers under the $80B federal programme.
• Fuel customers: Ukraine's Energoatom (VVER-440/1000 fuel + a July 2025 agreement for in-country final-assembly capability), and an EU consortium Westinghouse leads to supply Russia-independent VVER fuel across Eastern Europe.

Chokepoints / single-source tells: (1) Westinghouse is the single Western source of VVER fuel — a chokepoint that works in its favour. (2) Heavy-forging capacity globally is scarce (Japan Steel Works, Doosan) — a genuine constraint on how fast an AP1000 fleet can be poured. (3) Skilled nuclear construction labour and first-of-a-kind (FOAK) engineering is the binding constraint the Google-AI partnership (Lens 3/8) is explicitly aimed at.

Lens 3 · Competitive Advantages (moats)

Westinghouse has one of the widest moats in the energy-industrial complex, but it is a moat around a slow, cyclical, politically-hostage business.

• Installed base + switching costs (the deepest moat). OEM to >50% of the global fleet; a utility running a Westinghouse plant is a near-captive services/fuel customer for the plant's 60–80-year life. Fuel qualification and I&C integration create multi-year switching costs. This is why ~50% of earnings recur.
• IP estate / process moat. Seven-plus decades of proprietary PWR data, now being codified into the HiVE generative-AI platform on a custom LLM ("bertha," launched Sept 2024). The APR1400/APR1000 designs of competitor KHNP derive from Westinghouse's licensed System-80 technology — the KHNP IP settlement (Jan 2025) monetizes that lineage (Lens 10).
• Geopolitical moat (the sharpest edge right now). The only fully Western VVER fuel — the vehicle by which the EU and Ukraine de-Russify their nuclear fuel supply away from Rosatom/TVEL. Selected by an EU consortium for secure European VVER supply. Russia is a pariah supplier in the West; China's CAP1000 (an AP1000 derivative) cannot sell into NATO markets. That leaves Westinghouse and France's Framatome as the two Western reactor+fuel houses — and Westinghouse is the only one with a proven Gen-III+ unit operating in the US (Vogtle).
• Bargaining power: HIGH over installed-base services/fuel customers (captive); MODERATE-to-LOW as an EPC contractor on new builds, where sovereign customers, EPC partners (Bechtel/Hyundai), and heavy-forging suppliers hold leverage — and where FOAK cost risk historically flowed to Westinghouse (the 2017 bankruptcy).

Moat verdict: durable and arguably widening (VVER-fuel geopolitics + AI-enabled construction + a US government now underwriting demand). The risk is not that a competitor takes the moat — it is that the new-build leg's cost/schedule risk re-impairs the whole (Lens 12/13).

Lens 4 · Segments (revenue by line & geography)

Westinghouse does not publish audited segment financials (private). Best-sourced structure:

Geography: US (Vogtle base + the $80B federal fleet), UK (Springfields fuel), Sweden (Västerås fuel), and a fast-growing Central/Eastern European book (Poland, Bulgaria, Czech Dukovany participation, Ukraine). The 2025 growth delta was disproportionately European — the Dukovany construction participation drove the year's revenue and adjusted-EBITDA acceleration (Lens 5). Trend: accelerating on the new-build leg, steady on the annuity. Exact per-segment EBITDA and per-geography splits: n/a — not disclosed (private).

---

Phase B — Measure performance

+private overlay: Lens 5 → Funding & valuation trajectory; Lens 7 → Cap table & marks; plus a Traction & unit-economics block. Lens 6 → founder/owner commentary & Cameco earnings calls (no Westinghouse calls exist); Lens 8 → funding/policy/product catalysts.

Lens 5 · Funding & Valuation Trajectory (+ traction / unit economics)

Westinghouse's "financing history" is a private-equity ownership chain, not venture rounds:

Traction / unit economics (via Cameco's 49% equity-method disclosure — the hard numbers):

• 2024 (FY): Cameco's 49% share of Westinghouse adjusted EBITDA = $483M (within its $445–510M guide) → 100% adj. EBITDA ≈ $986M. Net loss $218M (Cameco's share) — driven by purchase-accounting (inventory step-up revaluation + non-operating acquisition/transition costs), not operations.
• 2025 (FY): Westinghouse adjusted EBITDA +30% vs 2024 → Cameco share ≈ $625M → 100% ≈ $1.28B; net earnings improved ~$276M vs 2024 (approaching GAAP breakeven as purchase-accounting drag rolls off). Total revenue ~$5.0B; ~26% adj-EBITDA margin. Driver: Dukovany construction participation.
• Q1 2026: Cameco share adj EBITDA $122M (vs $92M Q1 2025, +33%); net loss $46M (Cameco share), improved from $62M loss Q1 2025. (Q1 is seasonally the lightest quarter; Q2 is the lumpy peak.)
• Guidance signal (critical): Cameco raised its 2025 Westinghouse share guide from $355–405M to $525–580M mid-year on a ~$170M one-off Dukovany revenue increment; it had guided 6–10% CAGR over five years ex-that-one-off. Then in Q1 2026 it eliminated the five-year Westinghouse outlook, citing "significant variability in timing and magnitude of new-build projects" from the $80B deal, and now gives current-year guidance only. Read this two ways at once: the annuity got more valuable (a US government is now underwriting the build option), and the visibility that made it modelable was simultaneously withdrawn.

Lens 6 · Owner/Operator Commentary (sentiment trend)

No Westinghouse earnings calls exist; the read-through is Cameco's quarterly calls and management posture:

• Tone shift 2024 → 2026: escalating confidence, then a deliberate re-framing to "optionality we can't yet size." Through 2024–2025, Cameco framed Westinghouse as a steady, growing annuity beating deal-model expectations. By late 2025 (post-$80B), the language shifted to transformational / generational ("transformational partnership with the United States Government").
• The withdrawal of five-year guidance in May 2026 is the single most telling sentiment datapoint: management is signalling the upside is now too large and too lumpy to forecast — a bullish tell wrapped in a caveat. They repeatedly stress the outlook is "dependent on the ability of the executive branch of the US Government to obtain funding and support for the deployments" — i.e., they are explicitly fencing off political/appropriations risk.
• Recurring phrases: "full-cycle value," "disciplined strategy," "positive outlook for nuclear energy." What they stopped saying: precise multi-year Westinghouse EBITDA targets.

Lens 7 · Cap Table & Secondary Marks (comps)

Cap table: Brookfield ~51% (via Brookfield Renewable + institutional partners — Brookfield's direct economic interest is smaller than 51% because much is held through funds/LPs) / Cameco 49%. The extraordinary incoming "crossover investor" is the US federal government — the strongest IPO-proximity tell imaginable: under the Oct 2025 term sheet the government may take warrants for a 20% equity stake (or, per CNBC's read of the blended structure, end up an ~8% shareholder) if it makes an investment decision by Jan 2029.

Implied Westinghouse multiples vs listed nuclear peers (Westinghouse is private — these are implied marks against sourced peer multiples):

Read: the $30B IPO floor (~23x 2025 adj EBITDA) is conservative-to-fair against the nuclear-supplier peer set (BWXT ~45x, Cameco ~37x). If Westinghouse simply held its peers' multiples on ~$1.3–1.5B of adj EBITDA, a $40–50B+ mark is arithmetically defensible — which is the crux of the bull case and why Cameco/Brookfield may resist an early, floor-priced IPO. Peer multiples not pulled this run are honestly marked n/a; do not infer them.

Lens 8 · Catalysts (funding / policy / product events that move the mark)

Westinghouse isn't listed, so "price" moves show up in Cameco's (CCJ) stock and the implied private mark. The pattern reveals the market reacts almost entirely to policy and new-build catalysts, not the annuity:

• May 2025 — Trump executive orders to quadruple US nuclear capacity + reform the NRC; Westinghouse signals US AP1000 deployment.
• Jul 2025 — Google Cloud AI partnership (repeatable AP1000 construction); Energoatom VVER final-assembly agreement.
• Oct 27 2025 — $80B US government partnership — the defining catalyst; CCJ re-rated sharply; announced by Trump in Japan alongside the US-Japan framework (drawing on Japan's $500B US-energy pledge).
• Jan 2025 — KHNP/KEPCO IP settlement — removed the Dukovany overhang and created a framework for Westinghouse to participate in KHNP builds outside Korea.
• Nov 2024 / Aug 2025 — Bulgaria Kozloduy engineering contract → first purchase orders; Oct 2024 Poland supplier contracts.
• Ongoing — AP300 MoUs (New Brunswick, European data centers); eVinci at Malmstrom AFB (ANPI, Apr 2026) + DOE FEEED. Pattern: the market pays for sovereign demand and policy underwriting, and largely ignores the steady annuity — which means the mark is high-beta to Washington and thin-skinned to any $80B-deal slippage.

---

Phase C — Judge people & books

Lens 9 · Management

• CEO — Dan Sumner. 14-year Westinghouse veteran; former CFO and former President of Operating Plant Services (the annuity engine). Became Deputy CEO Jan 2025, interim CEO Apr 1 2025, confirmed permanent CEO in 2025. Archetype: operator/finance lifer, not a founder — appropriate for a capital-intensive, execution-risk business at the moment it must industrialize a repeatable AP1000. His services/CFO background is reassuring for the annuity and for cost discipline; the open question is FOAK new-build execution.
• Prior CEO — Patrick Fragman (2019–Mar 2025) led the successful post-2018 Brookfield turnaround; departed for family reasons in Europe. A clean, non-distressed handover — not a red flag.
• Owners as de-facto board. Brookfield (deep infrastructure/renewables capital-allocation track record; Connor Teskey / renewable leadership) and Cameco (CEO Tim Gitzel — uranium/fuel-cycle expertise, the vertical-integration logic). This is a sophisticated, aligned, deep-pocketed ownership with a clear monetization path (IPO/spin by ~2029).
• Capital allocation: Brookfield's 2018 purchase (~$4.6B) → 2023 sale into the consortium (EV $7.9B) was a strong PE round-trip; the current owners are reinvesting into AP1000 industrialization + AI (HiVE/Google). Skin in the game: $2.1B of Cameco's balance sheet is in the 49% stake — very real alignment.
• Red flags (governance): related-party dynamics are structural — Cameco is both owner and uranium/fuel-cycle counterparty; and the incoming US-government participation interest is an unusual overhang on minority economics (Lens 12/13). None are fraud-flavoured; all are structural.

Lens 10 · Forensic Red Flags & Regulatory

Accounting (private, unaudited — all ``):

• Purchase-accounting distortion. Reported GAAP net losses in 2024 (–$218M Cameco share) are dominated by inventory step-up revaluation and acquisition/transition cost expensing — non-cash/non-operating. Adjusted EBITDA (~$986M '24 → ~$1.28B '25 at 100%) is the truer operating signal, but "adjusted" is the owners' definition — treat the ~50-point gap between adj EBITDA and GAAP earnings with appropriate skepticism until purchase-accounting fully rolls off.
• Revenue lumpiness / recognition. New-build EPC milestone revenue (the Dukovany +$170M) makes headline growth jumpy and forecasting hazardous — precisely why Cameco pulled five-year guidance. Watch for milestone pull-forwards flattering any given period.
• The FOAK cost-overrun scar. The 2017 Chapter 11 was caused by fixed-price AP1000 EPC exposure at Vogtle/VC Summer. The relevant forensic question for the $80B fleet is contract structure — how much FOAK/nth-of-a-kind cost risk sits with Westinghouse vs the US government/utilities/EPC partners. Not disclosed; this is the single most important unknown in the file.

Regulatory findings (regulatory/regulatory-findings.md, generated 2026-07-10):

• SEC (EDGAR EFTS — LR + AAER): zero findings. Westinghouse has no CIK — private, not an SEC filer; no EDGAR enforcement search possible.
• Non-SEC / litigation (web): The KHNP/KEPCO APR1400 IP dispute — Westinghouse alleged infringement of System-80-derived technology; a US court dismissed the suit on jurisdiction, and the parties settled Jan 17, 2025, clearing KHNP's Czech Dukovany signing and framing future Westinghouse participation in KHNP builds abroad. Resolved and monetized. No material FTC/DOJ/FDA/consent-decree hits surfaced.
• Nuclear safety/regulatory posture: NRC (US), ONR (UK), and multi-national licensing are the operative regulators; eVinci received the first-ever DOE preliminary safety design approval for a microreactor — a positive regulatory signal.
• Summary: No material SEC/AAER findings (private, no CIK — verified via EDGAR EFTS 2026-07-10). The only material legal matter (KHNP IP) was settled favourably Jan 2025. The defining historical risk is operational/financial (2017 FOAK bankruptcy), not enforcement.

---

Phase D — Project & stress-test

Lens 11 · IPO-Readiness & Path-to-Tradeable (+private Lens 11)

This is the be-early payoff lens. Westinghouse is not directly investable today; the path to a tradeable security is now unusually well-defined:

• The deal writes the IPO into a term sheet. Under the Oct 2025 partnership, if the US government makes an investment decision on/before January 2029 and an IPO would value Westinghouse at ≥$30B, the government may require an IPO and receive warrants for up to a 20% equity stake. Separately, the government takes 20% of Westinghouse cash distributions above a $17.5B cumulative threshold.
• Cameco has signalled it would consider spinning Westinghouse out as an independent company in 2029.
• Window: an S-1 becomes plausible 2028–2029, gated on (a) the US government's investment decision, (b) reaching the $30B valuation trigger, and (c) new-build execution de-risking enough to support a public multiple.
• Milestones that unlock the S-1: first US new-build FIDs under the $80B programme; AP1000 fleet order conversions (Poland/Bulgaria/Ukraine moving from engineering to construction); AP300 design certification (targeted 2027); continued adj-EBITDA compounding through ~$1.5B.

Implied valuation math (all ``): at the $30B floor on ~$1.28B 2025 adj EBITDA → ~23x — below Cameco (~37x) and BWXT (~45x). Hold peer multiples on a 2028 adj EBITDA of, say, ~$1.6B at even a 25–35x nuclear-supplier multiple → ~$40–56B. The $30B is a floor, not a fair value — which is the entire tension of the setup.

Forecast log: per the skill's --watchlist rule, no forecast.ts binary logged in unattended breadth mode (no committed base case). A future thesis pass could log: "Westinghouse IPO or spin announced on/before 2029-01-31, p≈0.55."

+private write-back: Westinghouse is not currently in research/private-watch.json. Adding it (stage: pre-IPO; ipo_readiness: high/term-sheet-defined; catalyst: US-gov investment decision by Jan 2029; dossier: this file) is a recommended follow-up — not executed this run to respect the wave's state-file boundaries.

Lens 12 · Bull vs Bear

Bull. Westinghouse is the only pure-play, Western, proven Gen-III+ reactor OEM at the exact moment three secular tailwinds converge: (1) AI-datacenter power demand, (2) energy-security de-Russification (its VVER-fuel monopoly), and (3) an explicit US-government demand underwrite worth ≥$80B. Beneath the option sits a ~$4.3B, ~50%-of-earnings annuity growing mid-single-digits with near-zero churn (fleet is captive for decades). The AI/Google construction partnership attacks the one thing that has historically killed nuclear economics — FOAK construction cost (60% of reactor cost). The $30B IPO floor is below peer multiples; the fair mark is plausibly $40–55B. Owners are sophisticated and aligned with a defined 2029 monetization. The market is being handed a nuclear supermajor's re-IPO with a government backstop.

Bear (permanent-impairment risks). (1) FOAK cost risk, reloaded. The same AP1000 EPC exposure bankrupted the company in 2017; a US fleet of 10 units carries 10x the FOAK/schedule risk, and the contract risk-allocation is undisclosed. One Vogtle-style overrun on Westinghouse's balance sheet re-impairs everything. (2) The $80B is a non-binding term sheet — "the U.S. Government has not made a final investment decision or entered into any binding financial commitment"; it is hostage to appropriations, a future administration, and NRC throughput. Cameco pulling five-year guidance is the honest admission that none of this is yet bankable. (3) Government as counterparty and claimant — the 20%-above-$17.5B participation and the 20% IPO warrant dilute minority upside and complicate the eventual public equity story. Pre-mortem (18 months out, thesis broken): the executive branch fails to appropriate funding, the first US FID slips past 2027, a European build hits a cost dispute, and the "generational" narrative deflates back toward the annuity's ~$1.3B EBITDA — at which point the private mark compresses from the $30B+ hope toward the 8x entry multiple ($10–13B).

Contrarian view (what the market refuses to see): the consensus treats the $80B headline as the thesis; the real, underpriced asset is the VVER-fuel geopolitical monopoly + the services annuity — a Russia-independent, de-facto-regulated, inflation-linked cash machine that is valuable even if not one new US AP1000 is ever poured. The new-build leg is a free call option on top of a bond-like core.

Lens 13 · Devil's Advocate (short-seller)

Dismantling the bull case:

• Revenue concentration in a promise. Strip the annuity and the "growth" is a single non-binding US term sheet plus a handful of European engineering contracts at the earliest, riskiest stage. If the US FID slips or shrinks, the entire re-rating thesis is a policy headline with no cash behind it.
• The moat doesn't stop the thing that kills it. Switching costs protect the annuity, but they do nothing against construction cost/schedule blowout — the actual failure mode. Bulls point at the moat; the bankruptcy came from EPC math.
• Most dangerous competitor bulls underrate: KHNP. Korea's APR1400 is built on time and on budget (Barakah, UAE) — the one thing Westinghouse has not proven at scale. KHNP already beat Westinghouse at Dukovany. In a world that rewards delivery over pedigree, the settlement that "monetized" Westinghouse's IP also legitimized its most capable rival.
• Capital-allocation / incentive misalignment: Cameco is owner and fuel-cycle counterparty; the US government is about to be a claimant on distributions and a warrant-holder. Minority public shareholders in a 2029 IPO would sit behind a government participation interest and beneath two controlling owners with their own agendas.
• What breaks it permanently: a first US new-build FID that carries material FOAK cost risk on Westinghouse's balance sheet, followed by a Vogtle-style overrun. Plausibility: non-trivial — it has literally happened before, to this exact company, on this exact reactor.
• If growth disappoints 20–30%: the mark reverts from a $30B+ IPO hope toward the ~$8–13B private range (6–8x on a flat ~$1.3B adj EBITDA). The downside to the hype is large; the downside to the annuity is small — which is why the honest expression is a proxy long (CCJ) sized for a wide outcome distribution, not a levered bet on the $80B print.

Lens 14 · Fifteen Questions for the CEO (ordered by information value)

1. On the US $80B programme, what share of first-of-a-kind cost and schedule risk sits on Westinghouse's balance sheet vs the government, utilities, and EPC partners — and how does that compare to the Vogtle/VC Summer contracts that caused the 2017 bankruptcy?
2. The US government's commitment is a non-binding term sheet; what specific, dated milestones convert it into binding, funded orders, and what is your base case for the first US new-build FID?
3. How does the 20% government participation above $17.5B in distributions and the 20% IPO warrant affect the economics available to public minority shareholders in an eventual IPO?
4. Quantify the services + fuel annuity: contracted backlog, average contract life, renewal rate, and its through-cycle EBITDA margin — the bond beneath the option.
5. What is the realistic AP1000 build cadence given global heavy-forging capacity (Japan Steel Works, Doosan) and nuclear-labour constraints — how many units can actually be under construction by 2030?
6. On the Google/HiVE AI partnership: quantify the construction-cost reduction you underwrite per AP1000, and how much is proven vs projected?
7. What is the hard order backlog today (firm contracts, not MoUs) across Poland, Bulgaria, Ukraine, and the US, in dollars and units?
8. On VVER fuel: what is your realistic peak share of the European/Ukrainian VVER market as Rosatom is displaced, and what capacity investment does capturing it require?
9. When purchase-accounting fully rolls off, what is the normalized GAAP net-income and free-cash-flow conversion of the business?
10. What is the base-case IPO/spin timeline and valuation you would accept — and would you resist an early IPO forced at the $30B floor if you believe fair value is higher?
11. How exposed is the new-build pipeline to a change in US administration or NRC throughput, and what is your contingency if federal support weakens?
12. How do you keep KHNP — which delivers on time and on budget — from winning the next tranche of European and third-country tenders?
13. AP300 SMR and eVinci: which has a credible path to firm, financed orders before 2030, and what de-risks the FOAK unit?
14. Cameco is both owner and fuel-cycle counterparty — how are related-party fuel/uranium transactions priced and governed at arm's length?
15. What is the single risk that most keeps you up at night, and what would we see first if it were materializing?

---