BWX Technologies, Inc. (NYSE: BWXT) — Betting on the Race, Priced as if It Already Won

Institutional research memo · Report date: 2026-07-02 · Sector: Industrials — Aerospace & Defense / Nuclear Components · Independent equity research.

This report discusses valuation only as embedded expectations and scenarios. It contains no buy/sell recommendation and no price target anywhere except the clearly-labeled “Author’s Take” block below, which is a subjective view.

⚡ Author’s Take

This is the author’s own subjective opinion and general information, not investment advice. The analysis that follows (sections 1–15) takes no position and carries no price target.

Verdict: HOLD — a genuinely great business at an ungreat price. Not a short; accumulate on weakness, not here. Directional zone: I’d want to own it around ~28–32x forward non-GAAP EPS / ~24–28x adjusted EBITDA — roughly $135–160 on FY26 numbers (about where the 52-week low sat) — and I would not chase it above ~$210. Conviction: medium.

Strip away the noise and BWXT is one of the best-positioned industrial franchises in the market. It is the sole/near-sole manufacturer of the reactors that power every US Navy submarine and aircraft carrier — a government-designated monopoly with a moat measured in decades and clearances no entrant can buy — sitting on a record $8.7B backlog (up 77% year-on-year) into the most budget-protected corner of the defense ledger. Bolted onto that annuity is real, not-imaginary optionality: BWXT owns the only large heavy-nuclear component manufacturing base in North America precisely as the SMR/AP1000/data-center-power build begins, it is the only at-scale producer of TRISO reactor fuel, and it has a ~20%-CAGR medical-isotope business. The economics are healthy, not “prime-like-bad”: return on invested capital is ~12% all-in and ~17% ex-goodwill (above cost of capital), cash conversion is strong, and — refreshingly — 40% of long-term incentive pay is tied to ROIC. CEO Geveden’s line that BWXT is “betting on the race, not the horse” is exactly right: it sells the picks and shovels to whichever reactor design wins.

So why only HOLD? Because the price already assumes it wins the race. At ~$191 the stock trades at ~47x trailing non-GAAP earnings, ~33x adjusted EBITDA, and roughly 2–3x the multiple of every defense prime (LMT/NOC/GD at 18–22x) — despite margins and returns that are good, not exceptional, and a single monopsony customer that caps the upside on three-quarters of the business. Roughly three-quarters of the five-year, ~5x share-price move came from multiple expansion, not earnings; revenue compounded ~11% while the stock ran ~260%. The naval annuity deserves a premium to the primes. It does not obviously deserve forty-times-forward. The commercial/SMR layer that justifies the rest of the multiple is real but small, years from generating FCF at scale, and riding a crowded capital cycle where the market has a habit of capitalizing the boom into perpetuity (see GEV). Factor-wise the tape confirms it: BWXT loads on Momentum and Aerospace-Defense with negative Value and Quality tilts, its factor twin is Curtiss-Wright, and it is a theme/momentum trade now digesting a ~20% drawdown — a high-flyer correcting from an extended base, not a falling knife.

Framing: quality-compounder-at-the-wrong-price / priced-for-perfection. Flip bullish if: firm SMR/AP1000 component orders convert the optionality into hard backlog and segment margins inflect back toward the high-teens as the capex supercycle rolls off — ideally on a de-rate into the $140s. Flip bearish if: a Navy schedule slip or continuing-resolution stalls the core, commercial growth decelerates off the +121% Q1 base, or the multiple simply completes its mean-reversion toward the prime complex. Tag: The best moat in nuclear, wearing a nuclear-renaissance multiple.

📈 Stock Price Action — Five-Year Event Map

Over the trailing ~60 months BWXT ran a ~4.7x round-trip-and-then-some: from a five-year low of $40.72 (2022-02-22) to an all-time high of $238.10 (2026-04-15), closing $191.06 on 2026-07-02 — −19.8% off the high, inside a 52-week range of $135.69–$238.10. A sleepy naval-component compounder was re-rated into a nuclear-renaissance growth stock; the most recent leg is a high-multiple name digesting a ~20% correction while still holding above its 200-day EMA (~$191.5).

Cycle narrative. (1) The 2021–early-2022 leg was a naval-component compounder de-rating alongside the whole growth complex, no company-specific catalyst. (2) Russia’s invasion of Ukraine and the AUKUS submarine pact lifted the entire defense book and nearly doubled the stock through 2023. (3) The decisive change of character came when the nuclear-renaissance / SMR / data-center-power narrative attached itself to BWXT’s genuinely unique asset base — a strong Q4-23 print in February 2024 lit the fuse. (4) Microreactor (Project Pele) and medical-isotope optionality carried 2024 to a then-peak near $132. (5) A growth-stock rotation took it back below $90 into spring 2025. (6) The defining re-rate was April–October 2025 (+140%), bracketed by the $525M Kinectrics acquisition close (May-20-25) and the White House’s four nuclear executive orders (May-23-25) targeting 400 GW of new nuclear by 2050 and designating AI data centers as critical infrastructure — with backlog up 50% underneath. (7) A Q3 wobble and AI-unwind trimmed ~16%. (8) Momentum then drove a fresh all-time high of $238.10 (Apr-15-26) on the record 2025 year and a +121% Commercial-Operations surge in Q1-26, before (9) a ~20% correction into July. Price moves are FACT; attributed drivers are INTERPRETATION. No price target, no support/resistance levels.

1. Executive Summary

BWX Technologies is the United States’ sole-source manufacturer of naval nuclear reactors, cores and fuel — the beating heart of every US Navy submarine and aircraft carrier — and, increasingly, the picks-and-shovels supplier to the commercial nuclear renaissance. The company reports through two segments: Government Operations (GO), ~73% of FY25 revenue and the overwhelming majority of profit, and Commercial Operations (CO), ~27% and growing fast on acquisitions and the nuclear-power build-out. FY2025 revenue was $3,198.4M (+18.3%), GAAP diluted EPS $3.58 (non-GAAP ~$4.01), adjusted EBITDA $574.3M, and free cash flow $295.3M. Backlog closed the year at $7.26B (+50%) and reached $8.7B (+77%) by Q1-2026.

The business is genuinely high-quality. The naval franchise is one of the most durable competitive positions in the industrial economy — a government-designated monopoly protected by classified know-how, security clearances, ASME nuclear certifications and multi-decade qualification cycles, with essentially no entrants. Returns are healthy: ROE ~27%, ROIC ~12% all-in and ~17% ex-goodwill, both above an ~8–9% cost of capital. Cash conversion is strong (OCF/NI ~1.46x). Crucially — and contrary to a first read of third-party data — the margin “collapse” narrative is overstated: reported operating margin is 12.6% (not the ~10% some aggregators show, which wrongly exclude $74.9M of naval-JV equity income), and the real compression over three years is a modest ~2.7 points, driven mostly by the mix effect of two lower-margin acquisitions in an active capacity-investment phase. Management’s incentives are well-designed, with a 40% ROIC weight in long-term pay.

The debate is therefore not about business quality — it is about price. At ~$191 BWXT trades at ~47x trailing non-GAAP earnings, ~41x forward, ~33x adjusted EBITDA and ~5.9x revenue — roughly a 2–3x premium to the defense-prime complex (LMT/NOC/GD at 18–22x P/E) and at the 82nd–92nd percentile of its own history. About three-quarters of the five-year, ~5x price move was multiple expansion, not earnings growth. The market is correctly underwriting the durability of the naval monopoly and the record backlog; it is aggressively underwriting a full capture of the commercial/SMR total addressable market, a margin re-expansion, and the persistence of a 30x-plus EBITDA multiple on a business with a single monopsony customer and a ~1.7% free-cash-flow yield. The risk skew embedded in that multiple is symmetric-to-negative. This memo lays out the mechanism, the numbers, and the falsification tests for both sides.

2. Business Overview

BWX Technologies, headquartered in Lynchburg, Virginia and founded in 1867, is the successor to the historic Babcock & Wilcox Company (it took the BWXT name in June 2015 when the legacy power-generation business was spun off). It manufactures nuclear components and fuel, and provides nuclear services, through two operating segments.

Government Operations (GO) — $2,350.1M FY25 revenue (73.5% of total), 16.8% segment operating margin. This is the crown jewel. Through GO, BWXT engineers, designs and manufactures precision naval nuclear components, reactors and nuclear fuel for the US Department of Energy / National Nuclear Security Administration’s Naval Nuclear Propulsion Program — the reactors that power US Navy Virginia-class and Columbia-class submarines and Ford-class aircraft carriers. BWXT has supplied this program since the 1950s and has delivered roughly 420 reactors over that history. GO also encompasses: special materials (downblending government uranium stockpiles, uranium processing, and a growing defense-fuels enrichment and high-purity depleted uranium — “HPDU” — franchise); technical services (management-and-operations roles at DOE sites, recognized largely as equity income from unconsolidated joint ventures); and advanced/emerging programs — microreactors (Project Pele transportable reactor, the Janus Army program), TRISO reactor fuel, and space nuclear propulsion. GO is largely cost-reimbursable and fixed-price-incentive work; the 10-K explicitly notes that cost-plus structures “reduce our overall risk as projects increase in scale.”

Commercial Operations (CO) — $853.1M FY25 revenue (26.7% of total), 6.8% segment operating margin. CO comprises three businesses. Commercial nuclear power designs and manufactures steam generators, heat exchangers, pressure vessels, reactor components and spent-fuel containers — anchored by the CANDU refurbishment supercycle in Canada (Bruce, Darlington, Pickering life-extensions) and, increasingly, SMR/large-reactor component contracts. BWXT’s Cambridge, Ontario plant is the only large heavy-nuclear component manufacturing facility in North America. BWXT Medical produces medical radioisotopes and radiopharmaceuticals (TheraSphere, strontium, germanium, actinium-225, and a pipeline including lead-212 and technetium-99m). Kinectrics (acquired May 2025) adds a full nuclear lifecycle-services layer — testing, inspection, engineering — plus grid high-voltage testing and an entry into fusion (the UK tritium loop facility). CO carries firm-fixed-price, milestone-billed contracts with higher execution risk than GO.

Customer concentration. US Government revenue was 68% of the consolidated total in FY25, down from 76% in FY24 and 75% in FY23. The decline is mechanical — Kinectrics’ Canadian commercial revenue diluting the government share — not evidence of reduced reliance on the naval franchise, which remains the profit engine. No non-US-government customer exceeds 10% of revenue.

Revenue quality. The bulk of GO is recurring, program-of-record, multi-year work tied to congressionally-appropriated shipbuilding; CO nuclear is project-based but underpinned by decade-long refurbishment programs; medical is a recurring, consumable-like annuity. The company employs roughly 10,400 people (6,700 US, 3,500 Canada), ~3,200 unionized, across facilities including Lynchburg VA, Mount Vernon IN (home to a 1,000-metric-ton crane on the Ohio River), Euclid and Barberton OH, Erwin TN (Nuclear Fuel Services), and Cambridge/Peterborough/Kanata in Ontario.

Verdict: A high-quality, mostly-recurring revenue base anchored by an irreplaceable naval franchise, with a fast-growing but lower-margin commercial layer. The FY25 revenue jump was roughly two-thirds acquired (Kinectrics + A.O.T.) and one-third organic — a nuance the headline growth rate obscures.

3. Industry Dynamics

BWXT operates in two structurally different worlds, and conflating them is the single most common analytical error on the name.

Naval nuclear propulsion — one of the best industry structures in existence. The buyer is a monopsony (the US Navy, via DOE-NNSA). The barriers to entry are as high as any in the industrial economy: security clearances, classified reactor designs, ASME N/NPT/U certifications, Naval Reactors qualification measured in decades, and a capital base no new entrant would rationally build for a single customer. Competition is, in the 10-K’s own word, “limited.” The demand outlook is exceptional and budget-protected: the Navy’s 30-year shipbuilding plan shows growth under all three funding profiles, with a cadence of roughly two Virginia-class submarines per year, one Columbia-class per year, and Ford-class carriers at ~5-year intervals — plus genuine upside from additional Columbia units under discussion, the AUKUS submarine pact (Australia), and potential Korean nuclear-submarine fuel demand. This is a protected annuity that sits outside the ordinary capital cycle. The one structural caveat is the classic monopsony trade: extreme barriers do not translate into extreme margins, because the government buyer captures much of the surplus (the diversified primes earn only ~10–11% operating margins and low-teens ROIC for exactly this reason). BWXT’s sole-source position lets it escape some of that ceiling, but not all of it.

Commercial nuclear — a real demand renaissance drawing a flood of capital (a Marathon warning), where BWXT is deliberately the merchant, not the bettor. The tailwinds are legitimate and numerous: the US-Japan announcement of up to $40B / 3 GW of GE-Hitachi BWRX-300 SMRs; plans for up to 10 AP1000 and up to 10 X-300 reactors in the US; the on-time, under-budget CANDU refurbishments in Canada; AI-data-center power demand; the ADVANCE Act and DOE loan programs; and the May-2025 executive orders targeting 400 GW by 2050. But this is precisely the kind of demand surge that, in Marathon’s capital-cycle framework, pulls in excess supply — and the reactor-developer field is now crowded with capital (NuScale, X-energy, TerraPower, Oklo, Kairos, plus SPAC/IPO issuance). BWXT’s strategic answer is the correct one: rather than bet on which reactor design wins, it sells the scarce, capacity-constrained input every design needs — heavy N-stamp component manufacturing, the Cambridge plant, the 1,000-ton crane, and TRISO fuel. As CEO Geveden puts it, “betting on the race, not the horse.” The manufacturing bottleneck is itself a moat; the reactor-design lottery is not one BWXT is exposed to.

Medical isotopes are a smaller, genuinely competitive market (Lantheus, Curium, NorthStar, SHINE), where BWXT is a sub-scale but fast-growing participant with real optionality in radiopharmaceuticals.

Verdict: structurally good, but bifurcated. The naval segment is a near-ideal industry (monopsony-capped margins the only blemish); the commercial segment enjoys a powerful demand cycle that is simultaneously a supply-side capital-cycle risk — mitigated, in BWXT’s case, by its merchant-supplier positioning and its ownership of a genuine manufacturing bottleneck.

4. Competitive Position

Naval: a durable, government-granted monopoly — the strongest Greenwald advantage type. In the Competition Demystified taxonomy, BWXT’s naval franchise combines all three genuine advantages at once: economies of scale (it is the only qualified manufacturer, so it amortizes a fixed cost base no competitor can match), customer captivity (switching costs measured in decades — the Navy cannot re-qualify an alternative supplier on any relevant horizon), and intangibles (classified designs, security clearances, and nuclear certifications an entrant cannot buy). The government’s explicit designation of BWXT as sole/near-sole source is the capstone. The moat is not theoretical: it shows up in the financials as a sustained ~20% GO adjusted-EBITDA margin and in the naval joint ventures’ managed revenue of ~$7.2B. Absent this franchise, GO would be commodity metal fabrication earning commodity returns. This is the load-bearing element of the entire investment case, and it is as real and durable as moats get.

Commercial: a real but narrower and more contestable edge. Here BWXT competes against Framatome, Westinghouse, Doosan Enerbility, Japan Steel Works and Sheffield Forgemasters. Its advantage is North American localization — Cambridge is the only large heavy-component plant on the continent, and management (correctly) argues that as the US and Europe favor local supply chains for nuclear projects, that geography becomes a durable competitive advantage. Kinectrics adds a services moat (embedded, recurring relationships with plant operators). But this edge is regional and erodable, which is exactly why BWXT is racing to lock it in with the PCG acquisition and the planned Mount Vernon greenfield — an attempt to build the same manufacturing bottleneck in the US before competitors do.

TRISO fuel: an early near-monopoly worth watching. Management states BWXT is the “only producer of TRISO at any scale” (hundreds of kilograms per year), supplying its own Pele reactor plus third parties (Antares, a collaboration with Kairos). A large-scale Wyoming plant is under consideration. If the microreactor market develops, this is a genuine pick-and-shovel monopoly in the making — though it is early and small today.

Medical: sub-scale and competitive — real growth, no moat.

Verdict: durable advantage in naval (and emerging in TRISO); narrow, contestable differentiation in commercial and medical. The critical point for valuation is that the genuinely durable moat sits under the lower-multiple part of the story (naval), while the higher-multiple narrative (commercial/SMR) rests on a real-but-erodable manufacturing-capacity edge that the company is spending heavily to defend and extend.

5. Growth History and Forward Opportunities

History. Revenue compounded from $1,894.9M (FY19) to $3,198.4M (FY25), a ~9% CAGR, accelerating to +18.3% in FY25 — but that headline flatters the organic trend. Of the FY25 increase, roughly two-thirds was acquired: Kinectrics ($434.5M net, May-2025) and A.O.T. ($101.1M, January-2025). Organic growth was closer to ~7% for the full year, though the Q1-2026 print ran hotter at +26% total / +11% organic, with Commercial Operations up an eye-catching +121% (39% organic + Kinectrics) on Pickering steam-generator throughput and better-than-expected Kinectrics performance.

High-quality, backlog-supported growth (the base case earns this):

• Naval ramp — Virginia + Columbia + Ford, funded by a 30-year plan; a $1.4B naval-reactor pricing agreement and long-lead procurement booked in Q1-26 drove backlog to $8.7B.
• Defense-fuels enrichment and HPDU — a ~$1.6B contract, with over half of FY26 GO growth attributed to defense fuels and HPDU; the HEU enrichment facility in Erwin TN and HPDU facility in Jonesborough TN are ramping.
• CANDU refurbishment — the Pickering life-extension drove the Q1-26 CO beat; a multi-year program.
• Medical — ~20% CAGR over the last three years, with high-teens growth guided for FY26.

Speculative optionality (priced into the multiple, not yet earned):

• SMR components — reactor-vessel supplier on the first GE-Hitachi BWRX-300 at Darlington; but most SMR work to date is design awards, not production orders. The US-Japan 3 GW program and up-to-10-unit AP1000/X-300 programs are prospective.
• Microreactors — Pele → BANR (20 MWe commercial derivative) → Janus (Army), in procurement, not yet awarded at scale.
• Space nuclear — NASA nuclear-electric/thermal propulsion and national-security space; one-off, early.
• Technetium-99m — explicitly not in the 2026 forecast; still pre-commercialization with product-quality questions.
• Golden Dome / SHIELD — a large IDIQ with undefined scope for BWXT.
• Data-center nuclear — “in the windshield,” per management.

Forward guidance (FY26, raised at Q1, excludes PCG): revenue ≥$3.75B (~high-teens growth), adjusted EBITDA $650–665M, non-GAAP EPS $4.60–4.75, FCF $315–330M.

Verdict: mixed quality. The contracted core is high-quality and visible (record backlog, budget-protected demand). But the valuation prices a thick additional layer of SMR / AP1000 / microreactor / data-center / isotope optionality that is real in possibility but small, largely pre-revenue-at-scale, and years from generating meaningful free cash flow. The growth you can bank is good; the growth you’re paying for is a call option.

6. Financial Quality

This is the section where third-party data most badly misleads, and where the reconciliation to the 10-K changes the thesis. The apparent margin-and-returns “collapse” in aggregator data is largely an artifact; the real deterioration is modest and mix-driven.

Revenue and margins. FY25 revenue $3,198.4M. Gross margin compressed from 24.9% (FY23) to 24.2% (FY24) to 22.9% (FY25) — a real but shallow ~2-point move, driven by the mix effect of lower-gross-margin Kinectrics/medical revenue and GO program mix (new programs booked at conservative initial profit recognition). Reported operating margin is 12.6%, not the ~10% that appears in some feeds — the discrepancy is that the 10-K’s operating income of $404.5M includes $74.9M of equity income from the naval/DOE unconsolidated joint ventures (a genuine and growing profit stream: $50.8M → $55.9M → $74.9M over three years), which some aggregators strip out. The real operating-margin path is 15.3% → 14.1% → 12.6%, a ~2.7-point decline over two years attributable to gross-margin mix, higher deal-related SG&A/amortization, and unallocated corporate rising to $48.1M. Segment margins: GO 18.4% → 17.3% → 16.8% (mild erosion) and CO 8.0% → 8.9% → 6.8% (Kinectrics-dilutive in its integration year). Importantly, the enormous economic value in the naval JVs — ~$7.2B of managed revenue — never touches consolidated margin except through the growing equity-income line, so consolidated margin understates the franchise’s true profitability.

The “pension loss” that wasn’t. A naïve read of the non-operating line suggested a ~$114M pension drag. The 10-K shows the opposite: “Other — net” was a +$34.0M gain, and the pension/OPEB mark-to-market (immediate-recognition accounting; salaried accruals frozen since 2015) was a +$15.2M gain in FY25. There is no pension drag to normalize out; if anything, reported earnings were mildly flattered.

Normalized earnings. Reported net income $328.9M, GAAP diluted EPS $3.58 on 91.86M shares, effective tax rate 17.1% (below 21% on federal R&D credits). Management’s non-GAAP EPS was $4.01. Adjusting for ~$60.6M of pretax one-time costs (restructuring/transformation $29.6M + acquisition costs $31.0M ≈ +$0.55/share after tax) and removing the $15.2M MTM gain (≈ −$0.14/share), normalized EPS is ~$3.95–4.00 — underlying earnings power slightly exceeds the GAAP headline. Adjusted EBITDA (BWXT’s own measure) was $574.3M (FY24 $498.7M, FY23 $466.5M).

Cash flow and working capital. Operating cash flow was $479.8M — 1.46x net income, a strong quality signal. Capex of $184.6M (5.8% of revenue) is down from the FY20/FY21 peak (~$255M/$311M) but remains elevated as the capacity supercycle continues. FCF of $295.3M was ~90% of net income — depressed only by that normalizing capex, not by earnings quality — and should inflect higher as capex trends toward a maintenance level nearer ~$130M. Working capital is favorable: advance billings rose to $305.3M (customers prepaying naval work) and the cash-conversion cycle is only ~14 days.

Balance sheet. Total stockholders’ equity $1,232.5M; shares out 91.43M → book value $13.48/share (confirming the ~$13.93 in the AZI feed and refuting a ~$27.57 figure in some data). Tangible book is $401.8M ($4.39/share) after goodwill $500.9M + intangibles $329.9M = $830.7M (not the ~$2B some data implied). The real capital base is physical: PP&E net $1,585.1M. Debt is $2,016.0M against $499.8M cash → net debt $1,516.2M — net-debt/adjusted-EBITDA ~2.6x (3.5x on stricter reported EBITDA). The stack is investment-grade quality: a $1.25B 0%-coupon convertible due 2030 (conversion ~$262.51, capped-call-hedged, issued November 2025 to fund M&A), plus $400M 4.125% notes due 2028 and $400M 4.125% due 2029. Pension is modestly underfunded (~$78M liability + ~$79M OPEB) and manageable.

Returns. ROE is ~27% (flattered by ~$1.43B of accumulated treasury stock shrinking the equity base). ROIC is ~12% all-in (NOPAT ~$335M / invested capital ~$2,749M) and ~17.5% ex-goodwill/intangibles — both comfortably above an ~8–9% WACC. The “10% ROIC / value-destructive” read that appears in some feeds is the same equity-income-exclusion artifact; it is wrong.

Verdict: positive but nuanced — economics do NOT deteriorate the way the headline suggests. This is a strong-cash-conversion, above-WACC business in a heavy but contract-backed investment phase. The modest margin compression is real, mostly mix-driven, and should reverse as the two acquisitions season and the capacity build fills. The financial quality is genuinely good; it is the price that is the issue, not the business.

7. Capital Allocation

Capex — contract-backed, not speculative. The multi-year capacity supercycle (naval/DOE capacity, the Cambridge expansion) is underwritten by a $7.26B backlog growing 50%, with book-to-bill above 1. This is the right kind of capital intensity: building capacity against contracted and highly-visible demand, not on spec. Management guides FY26 capex to ~6% of sales, possibly stepping to ~7% if the Mount Vernon greenfield proceeds — explicitly rejecting a return to the ~9–10% of the prior decade.

M&A — a disciplined, on-strategy spree. Two 2025 deals: Kinectrics ($434.5M net, Canadian nuclear lifecycle services + radioisotopes) and A.O.T. ($101.1M, from L3Harris, special materials); plus PCG announced April 2026 (~$200M, ~$125M revenue, low-double-digit EBITDA margins, 400 N-stamp-certified employees) to establish US commercial-nuclear manufacturing capacity. Prices look reasonable (roughly single-digit-to-low-double-digit EBITDA multiples), the strategic logic is coherent (build/buy the manufacturing bottleneck ahead of the commercial build), and there have been no impairments to date. Kinectrics is reportedly exceeding its acquisition business case. Goodwill rose to $500.9M — the integration and CO-margin drag is the item to watch, but the record so far is competent. The mPower SMR license to Applied Atomics (June 2026) is a smart, capital-light monetization of a shelved reactor design (BWXT keeps the IP, exclusive manufacturing rights, and royalties).

Shareholder returns — deliberately minimal. The dividend is $92.5M (~$1.01/share, ~28% payout, ~0.5% yield) — conservative and reinvestment-first. Buybacks are token ($30M in FY25, merely offsetting $26.1M of SBC; share count essentially flat). This is a reinvest-for-growth allocation policy, appropriate given the opportunity set, not a capital-return story.

Incentive alignment — genuinely well-designed (and the “empire-building” suspicion is refuted). CEO Rex Geveden (President & CEO since 2017) has a target package of ~$8.0M, ~69% equity. The annual bonus is 70% operating income / 25% free cash flow / 5% safety, and — critically — the long-term PSU plan is 40% adjusted cumulative EBITDA / 40% average return on invested capital / 20% relative TSR. A 40% ROIC weight in long-term pay is exactly the discipline one wants to see in a company deep in a capex-and-M&A cycle; it materially lowers the risk that management grows the empire at the expense of returns. This is a meaningful positive and distinguishes BWXT from lower-quality “transformation” names whose comp rewards revenue and bookings alone.

Verdict: capital allocated intelligently on balance. Contract-backed capex, sensibly-priced and on-strategy M&A, and ROIC-linked incentives are the strengths; rising leverage (to ~2.6x) and the absence of any insider open-market buying are the offsets. The debt-funded growth is prudent given the backlog, but it does remove balance-sheet slack and is worth monitoring if the commercial cycle disappoints.

8. Changes and Headwinds — Last Two Years

Strategic and portfolio changes (net thesis-strengthening on the core):

• M&A spree (~$0.7B in 18 months): A.O.T. (Jan-2025, GO/special materials), Kinectrics (May-2025, CO lifecycle services + isotopes), and PCG (announced Apr-2026, US commercial manufacturing capacity, closes 2H26).
• US capacity build-out: planned Mount Vernon, Indiana greenfield (~100k sq ft, ~2x the cost of the Cambridge expansion, 2–3 year build) for large components (steam generators, reactor pressure vessels); the first deliberate step to replicate the North American manufacturing bottleneck on US soil.
• mPower SMR license to Applied Atomics (June 2026): exclusive land-based commercial rights in the US/Canada, capital-light, IP-and-royalty-retaining.
• ATI five-year naval material supply agreement through 2030 (June 2026), reinforcing the naval supply chain.
• Commercial firsts: the Kozloduy (Bulgaria) AP1000 owner’s-engineer win (Dec-2025) — a first meaningful AP1000 award — and RPV supplier status on the first GEH BWRX-300 at Darlington.

Balance-sheet and leadership:

• $1.25B 0% convertible issued (Q4-2025) to fund the M&A, plus a refinanced/expanded credit facility; ~$1.7B liquidity.
• CEO Rex Geveden remains in seat (succession not disclosed — an open question given his age, ~65); CFO is now Mike Fitzgerald. RADM McCoy (ex-head of GO) was seconded to DoD to help fix shipyard throughput. Next Investor Day is Fall 2026.

Headwinds and watch-items:

• Margin compression — gross margin 28.2% (FY19) → 22.9% (FY25) on commercial/services mix and ramp costs; the market needs this to inflect.
• Rising leverage to ~2.6x adjusted EBITDA and the removal of balance-sheet slack.
• Commercial integration (Kinectrics dilution) and the execution risk of the capacity build.
• Macro/policy: the FY27 ~$1.5T defense budget is a request, not an appropriation; a continuing resolution or Navy schedule slip is the shared near-term risk across the whole complex, compounded here by extreme single-customer concentration. A defense-contractor executive order (EO 14372) referenced in the filings carries potential authority to restrict distributions/buybacks/exec pay for underperforming contractors — low probability of bite, but a tail policy risk.

Verdict: net thesis-strengthening. The changes deepen the naval moat, extend the commercial optionality, and lock in scarce manufacturing capacity — but they also raise capital intensity, leverage and integration risk, and none of them changes the core valuation tension.

9. Risk Analysis

The dominant near-term risk is #1 — this is a valuation risk more than a business risk, and it is partly a factor-crowding risk (see the relevant section). The dominant long-term risks are #2 (the single-buyer dependence that underpins three-quarters of profit) and #4 (whether the commercial optionality the multiple pays for actually converts). Note what is not on this list at high severity: there is no accounting red flag, no restatement, no material-weakness — a clean contrast to some lower-quality “transformation” defense names.

10. Valuation Discussion (Embedded Expectations)

Where the multiple sits. At the 2026-07-02 close of $191.06, with ~91.4M shares (mkt cap ~$17.5B) and ~$1.5B net debt (EV ~$19B):

The comparison that frames everything. BWXT trades at a 2–3x premium to the entire defense-prime complex — LMT ~18x forward P/E / ~13x EBITDA; NOC ~20x / ~15x; GD ~22x / ~16x; RTX ~26x / ~19x; and, most tellingly, submarine co-builder HII at ~14.6x P/E / ~13x EBITDA — despite BWXT’s ~10.8% consolidated operating margin and ~10–12% ROIC being prime-like, and its single-buyer concentration being worse than a diversified prime. Against the nuclear-renaissance names, BWXT sits between the merchant/regulated operators (VST ~13x, CEG ~21x forward) and the hyper-growth extreme (GEV at a mirage multiple on a ramp-to-2028 story). The clear read: BWXT’s premium is a growth-optionality multiple (SMR/medical/space), not a returns-superiority multiple. The cleanest structural cousin, GD (via Electric Boat), earns a higher ROIC (~13%) on cleaner books and trades at less than half BWXT’s multiple.

The re-rate did the work. BWXT’s own-history P/E ran from ~14.8x (2021) to ~48x (2025); EV/EBITDA from ~15x to ~40x (stricter) or ~11x to ~30x (on adjusted EBITDA). Over that window revenue rose only ~51% (~11% CAGR) while the stock ran ~260% — roughly three-quarters of the total return was multiple expansion, not earnings. The AZI own-history percentile is 82nd composite, 92nd on P/E, 92nd on P/S — richest-ever territory. (The P/B percentile, 64th, is unreliable given the book-value data discrepancy and should be discounted.)

Reverse-DCF / embedded expectations. At EV ~$19B on ~$300M of current FCF, a Gordon model at ~8.5% WACC implies ~6.9% perpetual FCF growth with zero transition period — heroic. On normalized post-capex-peak FCF (~$450–500M, once the naval/isotope capacity build rolls off), the embedded perpetual growth eases to ~4–6% — still a full price that requires the SMR/commercial TAM and a margin recovery to materialize, not merely the naval annuity to persist. The market is simultaneously underwriting continued double-digit growth AND the persistence of a 30x-plus EBITDA multiple on a business with a ~1.7% FCF yield.

Scenario analysis (to ~FY2029; implied EV vs the current ~$19B):

The multiple is the dominant swing variable, and the skew is symmetric-to-negative (−42% / ~flat / +65%). What the market is underwriting correctly: naval monopoly durability, sole-source pricing, the $8.7B backlog, and ownership of the only large-reactor manufacturing base in North America. What it is underwriting aggressively: commercial/SMR TAM capture at scale (still small, years from FCF), margin re-expansion from the compressed trough back toward high-teens, an isotope ramp, and a durable 30x-plus EBITDA multiple on prime-like returns and a thin FCF yield. No price target — this is embedded-expectations analysis only.

11. Variant Perception

Consensus belief (what ~41x forward embeds). BWXT is the irreplaceable sole-source manufacturer of US naval nuclear reactors and the picks-and-shovels supplier to the entire commercial nuclear/SMR renaissance — a multi-decade shipbuilding super-ramp plus an emerging AI-data-center-power growth engine — deserving a secular-compounder multiple. The sell side is broadly constructive: consensus is a Buy with a median price target around $225 (range ~$182–278); Deutsche Bank upgraded to Buy at $255, Seaport to Buy at $245.

Strongest bull case. A genuine government-granted monopoly — the only large-reactor and naval-nuclear manufacturing base in North America — with a record $8.7B backlog (+77%), a shipbuilding ramp funded by the most budget-protected corner of the defense ledger, plus real optionality as the merchant supplier to every SMR developer (BWRX-300 for TVA/OPG, BANR/TRISO for data centers), microreactors (Pele), and a ~20%-CAGR medical-isotope business. Kinectrics adds a lifecycle-services layer. If the commercial pipeline converts, growth accelerates, margins inflect as the capex cycle rolls off, and today’s multiple is defensible on a longer runway.

Strongest bear case. Defense-contractor economics dressed in a growth-stock multiple. Cost-plus naval work, EBITDA margins compressed from ~19% to ~13.6%, ~10–12% ROIC, a ~1.7% FCF yield, and a single monopsony buyer — the fundamentals look like HII (which trades at ~14x P/E), but the price is ~41x forward. The commercial/SMR layer is still small and years from FCF at scale, and the theme is a crowded capital cycle (a Marathon signature: high returns and hot narratives attract the capital that competes them away). Strip the narrative and you own a good-but-capital-hungry defense supplier at 2–3x the prime multiple.

The 3–5 assumptions that matter most (with falsification tests):

1. Naval/consolidated margin re-expands as capex normalizes. Falsify (bear): two more years of sub-15% EBITDA margin and FCF conversion stuck below net income. Confirm (bull): a visible path back toward high-teens margins with rising FCF.
2. Commercial/SMR converts from narrative to revenue at scale. Falsify: BWRX-300/BANR slip, orders fail to firm, Commercial decelerates hard off the +121% Q1 base. Confirm: firm SMR/AP1000 component production contracts and data-center reactor awards entering backlog.
3. The 30x-plus EBITDA multiple persists. Falsify: a de-rate toward the prime/HII range (this is the base/bear driver). Confirm: the multiple holds through delivery.
4. Backlog growth is durable, not a pull-forward. Falsify: book-to-bill drops below 1x and backlog rolls over. Confirm: continued >1.2x book-to-bill.
5. Single-buyer concentration stays benign. Falsify: a Navy schedule slip or budget continuing-resolution. Confirm: multi-year appropriated funding.

The factor-positioning read — where consensus may be offsides. FactorsToday models BWXT as a Market (beta 1.17) + Aerospace-&-Defense (+0.73) + Industrials (+0.58) + Momentum (+0.30) long, with a Canada (+0.30, Kinectrics) and gold-price (+0.27) tilt and — tellingly — negative Value (−0.48) and negative Quality (−0.12) loadings: the market pays up for it and it does not screen as high-return-quality. This is a growth/theme trade, not a value or quality-compounder loading. Its risk-adjusted record is strong (y3 +41%/yr, Sharpe 1.06; y1 +37%; alpha +0.18; 12-month relative strength ~37) but it is now in a ~20% drawdown with a 3-month return of ~−34% annualized — a high-flyer digesting a correction from an extended base, not a falling knife (5-year max drawdown only −32.9%, still above the 200-day EMA). Idiosyncratic volatility is high (~32.5% annualized, ~half of total variance), so much of the move is stock-specific narrative rather than factor beta. The factor twin is Curtiss-Wright (similarity 0.96), then aerospace-defense ETFs (PPA/XAR/ITA/SHLD) — the model reads BWXT as an A&D-momentum name, not a utility/power name. That is precisely where the variant sits: consensus is paying a nuclear-renaissance / SMR multiple, but the tape and factor loadings still own it as a crowded momentum/defense trade. The de-rate risk is therefore a factor-crowding risk as much as a fundamental one — and crowded momentum names de-rate faster than their fundamentals when the theme cools.

12. Fact vs. Interpretation Table

13. Open Questions

1. Exact organic growth split — how much of FY25’s +18.3% was A.O.T. (Kinectrics is disclosed at ~$231M of the contribution)? The organic core rate (~7% FY25, ~11% Q1-26) is the number that matters for the durable trajectory.
2. When does capex normalize and FCF inflect? The whole “FCF yield is thin” bear point depends on whether capex rolls toward ~$130M maintenance in 2–3 years or the Mount Vernon greenfield keeps it elevated through the decade.
3. CEO succession. Geveden (President & CEO since 2017, ~65) has no disclosed successor; the franchise is people-and-relationship-intensive.
4. Commercial margin trajectory. Does CO margin recover from the 6.8% integration trough back toward the ~9–16% it has shown, and how fast — the single biggest lever on consolidated margin re-expansion?
5. How much SMR/AP1000 work is production vs. design? The multiple pays for production-at-scale; the backlog today is heavily naval + defense-fuels. When do commercial production orders (not owner’s-engineer or design roles) enter backlog?
6. Kinectrics purchase multiple and durability of its outperformance — is the “exceeding the business case” a timing effect or a run-rate?

14. What Must Be True

For the bull case to be right (and today’s price to prove cheap):

• The commercial/SMR optionality must convert into hard, at-scale production backlog within ~2–3 years — not just design awards and owner’s-engineer roles. Falsification test: if, by end-2027, Commercial Operations backlog is still dominated by CANDU refurb and medical with no material SMR/AP1000 production contracts, the growth-optionality premium is unearned.
• Consolidated margins must inflect back toward the high-teens as the capex cycle rolls off and the acquisitions season, lifting FCF conversion above net income. Falsification test: adjusted EBITDA margin still below ~15% and FCF below net income in FY2027.

For the bear case to be right (and today’s price to prove expensive):

• The multiple must mean-revert toward the defense-prime/HII range as the market re-classifies BWXT from “nuclear-renaissance compounder” to “sole-source defense supplier with a good but capital-hungry commercial arm.” Falsification test: the stock sustains a >30x EBITDA multiple through a full year of decelerating commercial growth — which would show the market has permanently re-rated the franchise.
• Single-buyer/commodity dynamics must reassert: a Navy schedule slip, a continuing resolution, or a commercial deceleration off the +121% base. Falsification test: multi-year appropriated naval funding and firming commercial orders together would break the bear.

The elegant thing about BWXT is that both falsification tests are observable on a ~2-year horizon: watch the composition of backlog growth (naval/defense-fuels vs. commercial production) and the direction of segment margins. Those two series will settle the debate.

15. Source Appendix

See the Source Appendix (Appendix B) for the full citation list. Primary sources include: BWXT FY2025 Form 10-K (filed 2026-02-23), FY2021–FY2024 Form 10-Ks, the Q1-2026 Form 10-Q (filed 2026-05-04), the DEF 14A proxy, the Q1-2026 and Q4-2025 earnings-call transcripts, the FY2025 results release and FY2026 guidance, 8-K material-event filings (A.O.T./Kinectrics/PCG closings, the $1.25B convertible, contract awards), and the Form 4 insider corpus. Third-party/aggregated quantitative data (ROIC.ai, AZI, FactorsToday) were used for cross-checks and factor positioning and are reconciled to the filings throughout; where they disagreed with the 10-K on material figures (operating margin, book value, ROE/ROIC, non-operating items), the filing governs and the discrepancy is noted in the relevant section.

APPENDIX A — Standard Diligence Questionnaire — BWX Technologies, Inc. (NYSE: BWXT)

Report date: 2026-07-02. Supplemental to the research memo; grounded in the underlying analysis. Fact / Interpretation / Assumption labels applied where material.

General

What thoughtful questions have other investors asked about this company? The recurring, load-bearing questions are: (1) Is the ~47x non-GAAP / ~33x adjusted-EBITDA multiple defensible for a business whose profit engine is a cost-plus, single-customer naval franchise? (2) How much of the SMR/nuclear-renaissance narrative is real production backlog versus design awards and optionality? (3) When does the multi-year capex supercycle roll off and free cash flow inflect? (4) Is the margin compression (gross margin 28%→23% since 2019) structural or a transient mix/integration effect? (5) How dependent is the whole thesis on the US federal budget, and what does a continuing resolution or Navy schedule slip do? These are the right questions; the memo answers each below.

Cyclicality & Earnings Nature

Are earnings at a cyclical high or low? Interpretation: neither extreme — earnings are on a secular up-ramp (naval shipbuilding + defense fuels + commercial nuclear), but margins are cyclically depressed by the acquisition-integration and capex-investment phase. Normalized margins are above reported.

Driven by external environment or internal actions? Both: external (defense budgets, nuclear-renaissance demand, CANDU refurbishment schedules) and internal (M&A, capacity build, throughput initiatives — “DPX/Driving Performance Excellence”). The naval core is externally-appropriation-driven; the commercial upside is internally-execution-driven.

How stable are revenues? Fact: highly stable and visible — a $7.26B (FY25) / $8.7B (Q1-26) backlog covers multiple years, ~40% recognizing within 12 months, book-to-bill above 1. Naval is program-of-record. This is among the most visible revenue bases in industrials.

Outlook for products/services? Strongly positive on demand (naval ramp, SMR/AP1000 build, medical isotopes, defense fuels); the question is margin and multiple, not demand.

How big will this market be? Interpretation: the naval-nuclear TAM grows steadily with the 30-year shipbuilding plan and AUKUS; the commercial-nuclear TAM is potentially very large (hundreds of large reactors, thousands of SMRs globally per management) but with wide timing uncertainty. Domestic and international (Canada is now a major base via Kinectrics; Bulgaria/UK/global SMR prospects).

Business Quality & Competitive Moat

Is the industry getting more or less competitive? Naval: essentially non-competitive (sole/near-sole source) — stable. Commercial: a demand renaissance is drawing in capital (more competitive over time in reactor design), but the manufacturing-capacity bottleneck BWXT occupies is scarce and getting scarcer relative to demand.

How profitable is the business? Fact: ROE ~27%, ROIC ~12% all-in / ~17.5% ex-goodwill (above WACC), GO segment operating margin 16.8%, adjusted EBITDA margin ~18% (management basis). Good, not exceptional; monopsony caps the naval ceiling.

How profitable is the industry — competitors, barriers? Defense primes earn ~10–11% operating margins / low-teens ROIC; barriers are extreme (clearances, certifications, capital, qualification). BWXT’s sole-source position earns above the diversified-prime average on its naval work.

Can the business be easily understood? Interpretation: the segments are understandable, but the naval accounting (equity-method JVs, managed vs. consolidated revenue, immediate-recognition pension MTM) creates traps that mislead casual readers (see the Financial Quality section).

Undermined by foreign low-cost labor? No — security-cleared, domestically-mandated nuclear manufacturing is structurally protected from offshoring; localization is in fact a tailwind.

Do brands matter? Nature of competition? Switching costs? Brand per se is minor; what matters are certifications, clearances and a multi-decade track record — which function as an un-buyable brand. Naval switching costs are effectively infinite on any relevant horizon; commercial switching costs are moderate (qualified-supplier status, localization).

Financial Condition & Balance Sheet

Assets not fully recognized on the balance sheet? Interpretation: yes — the naval franchise, security clearances, certifications, and the JV interests are worth far more than their carrying value; managed JV revenue (~$7.2B) dwarfs the equity-income line that reaches the P&L.

Off-balance-sheet liabilities? Modest — operating leases, a manageable pension/OPEB underfunding (~$78M + ~$79M), and standard nuclear-materials handling obligations (Price-Anderson indemnified).

How conservative is the accounting? Interpretation: reasonably conservative — strong OCF/NI (1.46x), no restatements, no material weakness, favorable working capital (customer advances). The immediate-recognition pension MTM adds non-operating lumpiness but is transparent.

How CapEx-hungry? Fact: very — FY25 capex 5.8% of revenue (down from ~9–10% a decade ago), and a multi-year supercycle to build naval and commercial capacity. This is the main reason FCF (~$295M) sits below net income (~$329M). Contract-backed, but a real cash drag until it normalizes.

Capital Allocation & Management

How much FCF, and how is it used? ~$295M FY25 FCF (guided $315–330M FY26); used for a conservative dividend (~28% payout), token buybacks (offsetting SBC), and — primarily — reinvestment and M&A. Philosophy is reinvest-first, appropriate to the opportunity set.

Significant acquisitions recently? Fact: A.O.T. ($101.1M, Jan-2025), Kinectrics ($434.5M, May-2025), PCG (~$200M, announced Apr-2026). Sensibly priced, on-strategy, no impairments; integration is the watch-item.

Buying back shares? Minimally — $30M FY25, essentially just offsetting dilution. Not a return-of-capital story.

Issuing shares to insiders? ~$26M SBC/year; share count roughly flat. No egregious dilution.

Compensation policy / motivations of management? Fact: well-designed — annual bonus 70% operating income / 25% FCF / 5% safety; long-term PSUs 40% adjusted cumulative EBITDA / 40% average ROIC / 20% relative TSR. The 40% ROIC weight is a genuine capital-discipline signal and refutes an “empire-building” read. CEO Rex Geveden (since 2017); CFO Mike Fitzgerald. Caveat: zero insider open-market buying 2024–2026 (neutral-to-slightly-negative signal).

Valuation & Market Data

ADR, MLP, or K-1 issuer? No — a standard US C-corporation, NYSE-listed common stock, 1099 (no K-1).

Dividend policy? ~$1.01/share (~28% payout, ~0.5% yield); modest and growing, secondary to reinvestment.

How profitable is the business? See above — good absolute returns, above WACC, but monopsony-capped and not exceptional relative to the multiple.

Is net income diverging from cash from operations? Fact: OCF exceeds NI (1.46x) — a positive quality signal. FCF is below NI, but only because of elevated (contract-backed, normalizing) capex, not earnings quality.

Risks & Downside

What would cause the stock to decline? Primarily a multiple de-rate (the stock is at the 82nd–92nd percentile of its own history and 2–3x the prime complex); secondarily a Navy budget/schedule disruption, a commercial-growth deceleration off the +121% Q1 base, an M&A-integration stumble, or a broad growth/momentum unwind (the name loads on Momentum with negative Value/Quality).

Risk of catastrophic loss? Interpretation: low-probability but high-severity — a serious nuclear-safety/handling incident would be franchise-threatening (mitigated by Price-Anderson indemnity and a decades-long safety record). Financial catastrophic loss (insolvency) is very unlikely given the backlog and investment-grade balance sheet.

Chance of a total loss? Remote. The bear case is a large de-rating (−40%+ from here in the scenario table), not a zero.

Recent News & Events

Has the business environment changed recently? Fact: materially in BWXT’s favor on demand — the May-2025 nuclear executive orders (400 GW by 2050, AI data centers as critical infrastructure), the US-Japan 3 GW SMR announcement, additional Columbia-class discussion, and AUKUS momentum. Backlog jumped to $8.7B (+77%) by Q1-26.

Significant acquisitions / accounting changes / new markets or facilities? Kinectrics/A.O.T./PCG (above); the planned Mount Vernon greenfield; the mPower license to Applied Atomics; the ATI naval supply agreement through 2030; the Kozloduy AP1000 owner’s-engineer win. No accounting-policy changes of note. CFO transition to Mike Fitzgerald.

APPENDIX B — Source Appendix — BWX Technologies, Inc. (NYSE: BWXT)

Report date: 2026-07-02. Primary sources first. Third-party aggregated data used for cross-checks and factor positioning are reconciled to the filings; where they disagreed on material figures, the filing governs.

Primary — SEC filings (EDGAR, CIK 0001486957)

1. BWXT FY2025 Form 10-K — filed 2026-02-23 (period end 2025-12-31). Segment revenue/operating income (Note 3), consolidated income statement, balance sheet, cash flow, backlog, debt, pension, equity income of investees. https://www.sec.gov/cgi-bin/browse-edgar?action=getcompany&CIK=0001486957
2. BWXT Q1-2026 Form 10-Q — filed 2026-05-04 (period end 2026-03-31). Q1 revenue $860.2M, gross margin, debt, equity.
3. BWXT FY2021–FY2024 Form 10-Ks — filed 2022-02-22, 2023-02-23, 2024-02-27, 2025-02-24 (+ 10-K/A 2025-03-19). Multi-year trend basis.
4. DEF 14A proxy statement — executive compensation structure (bonus 70% operating income / 25% FCF / 5% safety; LTI PSUs 40% adjusted cumulative EBITDA / 40% average ROIC / 20% relative TSR); board and NEO detail.
5. 8-K material-event filings (2024–2026) — A.O.T. close (Jan-2025), Kinectrics close (May-2025), $1.25B 0% convertible + capped-call issuance (Nov-2025), PCG acquisition announcement (Apr-2026), quarterly earnings releases, contract awards.
6. Form 3/4/5 insider corpus — 191 Form 4 filings swept 2024-01-01 to 2026-07-02; zero code-P open-market purchases identified.

Primary — earnings calls & investor materials

7. Q1-2026 earnings call transcript — 2026-05-05. FY26 guidance (revenue ≥$3.75B, adjusted EBITDA $650–665M, non-GAAP EPS $4.60–4.75, FCF $315–330M); backlog $8.7B (+77%); PCG (~$200M); Mount Vernon greenfield; segment detail (GO +4%, CO +121%); TRISO/microreactor/medical commentary; “betting on the race, not the horse.”
8. Q4-2025 earnings call transcript — 2026-02-23. FY2025 record year; segment results; capex guidance.
9. FY2025 results release / FY2026 guidance — BWXT investor relations (bwxt.com/investors); adjusted EBITDA $574.3M FY25; non-GAAP EPS $4.01; FCF $295.3M.
10. BWXT company profile / investor presentation — segment descriptions, facility footprint, program detail.

Primary — policy / program / industry

11. US Navy 30-year shipbuilding plan (March 2024) — Virginia/Columbia/Ford cadence and funding profiles.
12. White House nuclear executive orders — 2025-05-23 (four EOs targeting 400 GW by 2050; AI data centers as critical infrastructure).
13. US-Japan SMR announcement — up to $40B / 3 GW of GE-Hitachi BWRX-300 SMRs (2026).
14. ADVANCE Act and DOE loan-program materials — commercial-nuclear support framework.
15. ATI naval material supply agreement (June 2026); mPower license to Applied Atomics (2026-06-18); Kozloduy AP1000 owner’s-engineer award (Dec-2025) — company/trade-press releases (World Nuclear News, American Nuclear Society).

Third-party / aggregated (cross-check only; reconciled to filings)

16. ROIC.ai — income statement, balance sheet, cash flow, profitability/valuation ratios, enterprise value (accessed 2026-07-02). Note: several fields (operating margin excluding JV equity income, book value per share, ROE/ROIC, non-operating “other”) disagreed with the 10-K and were corrected to the filing in the relevant section.
17. AZI (azitrading.com) — 5-year daily price/OHLCV CSV; valuation-index own-history percentiles (composite 82nd, P/E 92nd, P/S 92nd); news feed (n=11).
18. FactorsToday (factorstoday.com) — factor loadings (Market 1.17, A&D +0.73, Industrials +0.58, Momentum +0.30, Canada +0.30, negative Value −0.48 / Quality −0.12), leaderboard (y3 +41%/yr, Sharpe 1.06; m3 −34% ann.), related-stocks (factor twin Curtiss-Wright), specific volatility (accessed 2026-07-01/02).
19. Sell-side actions — Deutsche Bank (Buy, $255), Seaport Research Partners (Buy, $245), consensus median PT ~$225 (range ~$182–278), via MarketBeat/TipRanks/Benzinga aggregation.
20. Peer public filings — LMT, NOC, RTX, GD, LHX (defense primes); CEG, VST, GEV (nuclear power/renaissance); AVAV (defense growth analog). Public 10-Ks and investor materials used for comparative multiples/margins/ROIC and industry framing.