Centrus Energy Corp. (NYSE: LEU) — A Russian-Arbitrage Broker Priced as America’s Enrichment Monopoly

Report date: 2026-07-18 · Price: $156.05 (2026-07-17 close) · Shares outstanding: 19,671,587 · Market capitalization: ~$3.07B · Enterprise value: ~$2.38B Sector: Energy · Nuclear Fuel Cycle / Uranium Enrichment · CIK: 0001065059

⚡ Claude’s Take

The author’s own independent opinion, offered as general information and not investment advice. The analysis that follows takes no position and contains no price target; this block is the sole exception.

AVOID here / accumulate only in the $85–105 zone. Conviction: medium.

“A money-market fund stapled to a Russian trading desk, wearing a monopoly’s clothes.”

The call. Centrus is not a uranium enricher. It is a broker that buys enrichment from Russia’s TENEX and France’s Orano and resells it to Western utilities, and the entire gross margin comes from a price-reset provision that locked its input cost at 2018 trough prices. The 10-K says so in plain language: the reset “reduced the cost for our purchases from 2019 through 2028.” That is not a moat — it is a wasting arbitrage with a printed expiry date. Meanwhile the company’s own production is a 16-centrifuge demonstration cascade making 900 kg of HALEU a year: roughly 0.35% of the world enrichment market and about 4% of Urenco’s single New Mexico plant. The tell that nobody wants to look at is that gross profit has been dead flat at $110–120M for five consecutive years ($117.9M in FY2022, $117.5M in FY2025) while revenue grew 53%. Six years of “growth” produced no incremental economics whatsoever.

The valuation is where this becomes actionable. At $156 the enterprise is $2.38B against $50.2M of FY2025 operating income — 47x EV/EBIT, 40x EV/EBITDA. And even that flatters it: roughly two-thirds of FY2025 pretax income was non-operating — $44.7M of interest earned on the cash pile the company raised by selling stock and converts, plus an $11.8M debt-extinguishment gain. Strip the treasury income and normalize the 9.4% tax rate and the operating business earns somewhere between $1.05 and $1.95 a share — the stock is at 80–150x. The most damning single datum: after a 64% drawdown from the October-2025 high, LEU still trades at the 95th percentile of its own ten-year P/E history and the 90th percentile on P/S. The drawdown de-rated the price, not the valuation, because earnings fell faster than the stock. (The 0.7th-percentile P/B reading is a denominator artifact — book equity was negative for eight straight years and only exists now because of the 2025 raises. It is not cheapness.)

Why not a short, and why the accumulate zone is real. Three things here are genuine and non-replicable inside a decade: the only NRC licence held by a US-owned entity to construct and operate a commercial enrichment plant; the only deployment-ready US-origin enrichment technology, which is why NNSA notified an intent to sole-source from Centrus (Urenco and Orano are legally barred); and a real, signed, firm-fixed-price $900M DOE contract under which Centrus takes title to the capacity it builds. Management also deserves genuine credit for capital allocation: they sold ~$533.6M of stock into the melt-up (the November-2025 tranche at an average of $269.21/share, 73% above today) and issued $805M of 0% converts — then declined to touch the ATM in Q1-2026 because it “didn’t feel it provided the right shareholder value.” That is opportunistic financing of a high order, and it is why there is $1.87B of cash and $689.6M of net cash ($35.06/share) behind the stock. This is a binary policy option, and options are worth something. They are just not worth 47x EBIT on the run-off business that funds them.

The framing, evidence-based. This is a falling knife, not a value name — and the factor data makes that precise rather than rhetorical. LEU carries a positive momentum loading (base-model beta 1.18) into a market where momentum is working (+15.5% over 252 days, z-score 0.67), and is still down 47.6% over six months, with m3/m6 Sharpe at −0.71/−0.84. A high momentum beta with deeply negative realized momentum in a friendly momentum regime is the signature of a crowded thematic trade losing its crowd. Two-thirds of the variance is idiosyncratic (R² 0.31–0.35; specific vol 68%), and five of the six factor-nearest peers are uranium ETFs — the tape prices LEU as nuclear-theme beta, not as a company. Note also that S&P SmallCap 600 inclusion on 2026-07-14 was sold into: 3.5x normal volume, −8.8% on the day. And the tail is not theoretical — the predecessor entity, USEC, took this equity to zero in a 2014 Chapter 11; the lifetime max drawdown is −99.98%.

Valuation zone. My base-case scenario values the enterprise around $0.9–1.2B on present value, versus $2.38B today. Backing in net cash of $35/share, an entry around $85–105 puts the operating business plus the policy option at roughly $1.0–1.4B of EV — a price at which you are paid for the 2028 air-pocket rather than paying through it. Below $85 the option is close to free.

Triggers. Flips me bullish: a second, materially larger federal award (>$3B) or a binding, priced, multi-year HALEU offtake from a developer that has actually broken ground — plus any credible bridge over the 2028–29 gap. Flips me bearish: TENEX/Orano roll-off arriving with no margin replacement, a large dilutive equity raise to fund capex, first-cascade slippage past 2029, or SWU spot printing below ~$160.

📈 Stock Price Action — Five-Year Event Map

Over five years LEU ran from $18.63 (2022-05-12) to $436.00 (2025-10-15) — a 23-fold move — and has since given back roughly two-thirds. It closed at $156.05 on 2026-07-17, inside a 52-week range of $146.61–$436.00, −64.2% off the high and −42.7% year to date, trading below its 21-day ($165.67), 50-day ($175.53) and 200-day ($202.56) exponential moving averages.

Narrative. (1) The 2021 leg was the first retail uranium wave finding a small, newly profitable ticker. (2) It round-tripped entirely as rates rose and the thematic bid vanished. (3) The 2022–23 recovery was the first genuinely fundamental leg — the invasion made Western fuel security a policy priority and DOE handed Centrus the HALEU demonstration contract. (4) The 2024 drawdown reflected an awkward truth: the company that benefits from banning Russian uranium is also sourcing from Russia; TENEX declared force majeure and waiver mechanics dominated the tape. (5) The autumn-2024 spike was pure AI-power narrative, and management sensibly monetised it with a $402.5M convertible. (6) The Jan–Apr 2025 unwind was the AI-capex scare. (7) The 2025 melt-up is the defining event of the story: a 7x in six months on narrative, during which management raised roughly $1.33B in converts and equity — the entire cash pile that now underpins the balance sheet. (8) The subsequent de-rate has been relentless and, notably, fundamentally uneventful: even the $900M DOE contract on 2026-06-30, the largest commercial validation in company history, produced no durable re-rating, and S&P SmallCap 600 inclusion on 2026-07-14 was met with the heaviest down-volume day of the year (3,061,035 shares, −8.8%).

1. Executive Summary

Centrus Energy is the corporate successor to USEC, the privatised US government enrichment monopoly that took its equity to zero in a 2014 Chapter 11. Today it reports two segments. The LEU segment (77% of FY2025 revenue) sells the separative work unit (SWU) component of low-enriched uranium to nuclear utilities — but it does not, in any commercially meaningful sense, make SWU. It buys enrichment under a 2011 supply contract with TENEX, a Rosatom subsidiary, and under a 2018 agreement with France’s Orano, and resells it. The Technical Solutions segment performs the DOE HALEU demonstration and operations contracts at Piketon, Ohio, where a single cascade of 16 AC100M centrifuges produces about 900 kilograms of high-assay low-enriched uranium a year.

The economics follow directly from that structure and are not what the headline growth suggests. Revenue grew 53% from FY2022 to FY2025 ($293.8M to $448.7M). Gross profit over the same span went from $117.9M to $117.5M — flat, to within a rounding error, through the largest bull market in enrichment prices in modern history. Consolidated gross margin fell from roughly 40% to roughly 26%. The reason is disclosed in the 10-K: a market-related price-reset provision in the TENEX contract took effect in 2019, when SWU prices were near historic lows, and “reduced the cost for our purchases from 2019 through 2028.” Centrus’s margin is a long-dated purchase arbitrage — buy at 2018 prices, sell into a market that has since roughly tripled — and it expires with the TENEX contract in 2028 and the Orano agreement in 2030.

That expiry does not arrive alone. The Prohibiting Russian Uranium Imports Act’s waiver authority terminates at end-2027, after which the import prohibition is absolute. Russia’s own export decree licensing TENEX shipment-by-shipment runs only through 2027-12-31. The Russian Suspension Agreement quota set aside for Centrus runs through 2028. The Canadian ocean-shipper’s sanctions permit expires in March 2027. And Centrus’s first new cascade of centrifuges does not come online until 2029. There is, on the company’s own disclosed timeline, a structural earnings hole in 2028–29 between the death of the arbitrage and the birth of the plant. Management has been asked about it directly on three separate calls and has declined to answer; the CEO’s response — “You’re asking me to speculate. I would not be able to answer that… these are the years that I’ve put a question mark on as well” — is the most candid statement in the file.

Reported earnings are further removed from operations than they appear. FY2025 pretax income of $85.9M exceeded operating income of $50.2M because of $44.7M of investment income — interest on the cash raised by selling stock and convertible notes, which the 10-K describes as “interest earned on operating cash… primarily held in money market accounts” — plus an $11.8M gain on debt extinguishment. Across five years, 41% of cumulative pretax income was non-operating. In Q1-2026, more than 100% was: operating income was $0.8M. Add a 9.4% effective tax rate sheltered by $636.6M of federal NOLs and the gap between GAAP EPS of $3.90 and normalized operating EPS of roughly $1.05–1.95 is the single most important adjustment in this report.

Against that, the enterprise value is $2.38B — 47x FY2025 EBIT, 40x EBITDA, 20x gross profit. Reverse-engineering what the market is underwriting, Centrus would need to reach roughly 9 million SWU per year by 2032 at world-class economics to justify today’s EV, against a funded expansion of about 0.49 million SWU and an implied capital requirement of $11–17 billion on its own contracted cost benchmark. The $900M DOE award — genuinely valuable capital on which Centrus takes title — funds about 5% of that. Meanwhile the capital cycle is turning against it: DOE deliberately awarded $900M each to three competitors (Centrus, Orano, and the start-up General Matter) plus $28M to a substitute laser technology, while Urenco USA is expanding from 4.3M to over 7M SWU. Centrus will enter as the smallest, newest and highest-cost producer in that cohort.

The offsetting case is real and should not be dismissed. Centrus holds the only NRC licence held by a US-owned entity to construct and operate a commercial enrichment plant (granted 2007, running to 2037, amended in 2021 for HALEU), the only deployment-ready US-origin enrichment technology eligible for national-security work — which is why NNSA has notified an intent to sole-source — and a five-to-eight-year operating head start over any new domestic entrant. Management has financed opportunistically and well. But this is a politically granted option, not a competitive advantage in any Greenwald sense: there is no cost advantage (Centrus is the highest-cost first-of-a-kind producer), no customer captivity (SWU is a physics-defined commodity bought on price in competitive RFPs), and no scale (it is smaller than its rivals by more than an order of magnitude). And it is an option the granting authority is actively diluting.

No recommendation and no price target appears anywhere in the analysis below; the single labelled exception is the opinion block above.

2. Business Overview

2.1 What Centrus actually does

Centrus describes its LEU segment as supplying “various components of nuclear fuel to commercial customers from our global network of suppliers.” That clause carries the entire business model. Utilities operating light-water reactors need uranium enriched from its natural 0.711% U-235 content to roughly 3–5% U-235. The enrichment service is measured in separative work units. Under the prevailing commercial structure, a utility typically supplies its own natural uranium hexafluoride feed and pays the enricher only for the SWU component. Centrus sells that SWU component — but it buys it wholesale rather than producing it.

The Technical Solutions segment is a government-contracting business. Under the 2019 HALEU Demonstration Contract (DOE-funded up to $173.0M) and the successor HALEU Operation Contract, Centrus built and operates a 16-centrifuge cascade at the DOE’s Piketon, Ohio site. Enrichment began 2023-10-11 and the first HALEU was delivered 2023-11-07. The Phase 2 production rate is 900 kilograms of HALEU UF6 per year at nominal 19.75% U-235; cumulative production through Q1-2026 exceeds 1.6 metric tons.

2.2 Segment economics — the table that tells the story

Two observations dominate. First, total gross profit has not grown in six years: $97.6M, $114.5M, $117.9M, $112.1M, $111.5M, $117.5M. From FY2022 to FY2025 that is +3.8% cumulative while revenue rose 53%. Anyone underwriting this business on the revenue line is reading the wrong row. Second, the Technical Solutions margin collapsed to 5.9% in FY2025, and the reason is specific and important: the CFO disclosed on the Q4-2025 call that “Phase two costs incurred subsequent to November 2024 have not been subject to a fee as this portion of the contract remains undefinitized.” Centrus has been performing its flagship HALEU work essentially fee-free for over a year.

2.3 Customers, contracts and concentration

LEU-segment contracts are “primarily medium and long-term, fixed-commitment contracts under which customers are obligated to purchase a specified quantity of the SWU component of LEU.” International customers are roughly a third of LEU-segment revenue. The average individual SWU order fulfilled in 2025 was about $10.2M, which means the timing of a handful of shipments swings quarterly results violently — Q1-2026 saw SWU volume fall 47% and revenue fall 19% on essentially no change in the underlying business.

Concentration is severe and worsening. In FY2025 the ten largest customers were 77% of total revenue, with four individual customers at 16%, 15%, 13% and 11% — over 40% of the company in four relationships. In FY2022 the named large customers were Kyushu Electric (15%) and Synatom (13%). Add the DOE, which is effectively the entire Technical Solutions segment, and the customer base is a handful of utilities plus one sovereign.

2.4 The customer float is reversing

Centrus has historically been part-financed by customer prepayments — deferred revenue and advances that fund working capital ahead of delivery. That source is now draining. Deferred revenue and advances fell from $216.4M at Q1-2025 to $131.1M at year-end 2025 to $112.8M at Q1-2026. In FY2025 the company added only $17.3M of new advances while recognising $102.6M into revenue; deferred revenue net of deferred costs was a −$90.2M cash drag in FY2025 and −$152.5M in FY2024. New advances are running at roughly one-sixth the recognition rate. A structural negative-working-capital tailwind has become a headwind, and it is being obscured by the $1.87B of raised cash sitting on the balance sheet.

2.5 Backlog — headline $3.9B, deliverable ~$0.7B

At Q1-2026 total backlog was $3.9B ($3.1B LEU, $0.8B Technical Solutions), extending to 2040. Against ~$450M of revenue that reads as 8.7x coverage. The composition tells a different story. The CEO broke it down on the Q1-2026 call as “$700 million of broker-dealer backlog and $2.4 billion in contingent LEU enrichment sales.” The 10-Q states that the contingent commitments “depend on our ability to secure substantial public and private investment necessary to build new enrichment capacity.” Independently, the ASC 606 remaining performance obligations in the LEU segment were only ~$0.6B at year-end 2025. The Technical Solutions backlog is disclosed to include “funded amounts, unfunded amounts and unexercised options” exercisable at DOE’s sole discretion.

INTERPRETATION. The honest revenue-visibility figure is roughly $0.6–0.8B — about 1.6x revenue, not 8.7x. The other $2.4B is a conditional order book against a plant that does not exist, financed by capital not yet raised. “Definitive agreement” is not the same as unconditional; the contingency is a condition precedent inside the definitive agreement. Asked directly when contingencies convert to firm, the CEO declined: “this is not something that, at this point, we will get into a lot of contractual details.”

3. Industry Dynamics

3.1 The value chain and where Centrus sits

The nuclear fuel cycle runs: mining and milling (U3O8) → conversion (to UF6) → enrichment (SWU) → deconversion (for HALEU, UF6 to oxide or metal) → fuel fabrication → reactor. Centrus occupies only the enrichment node, and attempts to add deconversion. It owns no mines, no conversion capacity and no fabrication. Cameco’s own disclosures establish the upstream structure: Port Hope is one of only three Western UF6 conversion plants at roughly 18% of world capacity, and Russia holds approximately 22% of global conversion and 44% of enrichment. Cameco’s observation that enrichment tightness feeds back into uranium demand — because enrichers run lower tails assays when SWU is scarce — is the mirror image of the Centrus story.

3.2 Market structure: an oligopoly of sovereigns

The 10-K states the position with unusual frankness: “All of our current competitors are owned or controlled, in whole or in part, by foreign governments, and operate enrichment technologies developed with the financial support of foreign governments. These competitors may make business decisions… influenced by political or economic policy considerations rather than exclusively by commercial considerations.”

This is the defining feature of the industry and it cuts against Centrus. Barriers to entry are genuinely enormous — capital intensity, decade-long lead times, NRC licensing, non-proliferation regimes — which would normally imply excellent economics for incumbents. But the incumbents are sovereigns that do not require a commercial return on capital, built their plants with public money, and depreciate them accordingly. The profit pool that theory says should exist behind those barriers is largely captured by states, not shareholders. Being the smallest, newest, highest-cost, non-sovereign participant in such a structure is not an attractive position.

3.3 Pricing — and why it does not reach Centrus’s P&L

SWU prices bottomed near $35–50 after the post-Fukushima demand collapse and have risen roughly threefold since. Spot indicators reached approximately $188–220/SWU in 2026 with term around $166/SWU. The CEO cited a “24% compound annual growth rate from 2019 to 2025” in the LEU pricing curve, and noted spot “soared to $220” in late 2025.

That headline does not translate into Centrus’s earnings on any near horizon, for two disclosed reasons. First, the 10-K states plainly that “the average SWU price billed to customers typically lags published price indicators by several years” — and indeed FY2025’s realised SWU price was down 1% in a year when indicators rose sharply. Second, and more fundamentally, Centrus is a buyer as well as a seller. The Q1-2026 disclosure is the cleanest illustration: the average price of SWU sold rose 52% while the average cost of SWU sold rose 45%. The realised spread barely widened. An analyst caught this on the call and asked whether gross margin should therefore keep rising; the CFO deflected to “look at this on an average basis over several quarters.”

INTERPRETATION. “SWU prices are at records, therefore Centrus margins soar” is the most common error in the bull case and it is wrong on the company’s own numbers. As a reseller of third-party material, Centrus’s cost of goods rises alongside its selling price. Pricing leverage becomes real only when it owns production — which is to say, 2029 at the earliest.

3.4 The demand case, honestly stated

The bull case on demand is not fabricated. If Russia genuinely exits Western markets, non-Russian capacity of roughly 36M SWU must serve roughly 50M SWU of world demand — a ~14M SWU structural deficit. The CEO quantifies the domestic piece: “Near-term domestic LEU demand alone is set to increase by approximately 6.5 million SWUs, stemming from Russia’s exiting the market and the additional demand from restarts, upgrades, and new pledged reactors.” Reactor life extensions, restarts (Palisades, Three Mile Island) and datacenter-driven power procurement are real.

HALEU is a genuinely separate and more speculative market. Enrichment to 5–20% U-235 is required by Oklo’s Aurora, TerraPower’s Natrium, X-energy’s Xe-100, Kairos, Radiant and Westinghouse’s eVinci. Centrus cites industry surveys in which HALEU access is the single largest concern of advanced-reactor developers. Two caveats matter enormously. First, DOE’s HALEU Availability Program has made conditional allocations to Kairos, Radiant, TerraPower, TRISO-X and Westinghouse from down-blended government HEU stockpiles, not from Centrus — the near-term HALEU demand is being partly satisfied by free government inventory, which the 10-K itself warns “could exert downward pressure on prices.” Second, none of these reactor developers has deployed a commercial unit. Oklo’s first commercial operation date has slipped to 2028, with the Idaho National Laboratory unit producing nuclear heat rather than commercial power, and NuScale has not signed a binding order, with a first module expected in 2031 or beyond. The customers for Centrus’s future product are themselves pre-revenue.

One genuine structural gap favours Centrus: commercial deconversion of HALEU UF6 to oxide or metal does not exist at all. The CEO: “there is a hole in the fuel cycle right now… that process does not exist commercially.” Centrus is exploring a JV with Oklo to fill it.

3.5 The capital cycle — Marathon’s lens, and it is unambiguous

High prices plus subsidy attract capital; capital erodes returns. This industry is midway through exactly that sequence, and the supply response is already committed.

Congress appropriated roughly $3.4B to jumpstart domestic LEU and HALEU production. On 2026-01-05 DOE announced $2.7B across ten years: $900M each to General Matter, American Centrifuge Operating (Centrus) and Orano Federal Services, plus $28M to Global Laser Enrichment. Orano is deploying its award against “Project IKE,” a roughly $5B greenfield plant at Oak Ridge, Tennessee, targeting first LEU in 2031. General Matter — a start-up with no operating history — has secured DOE land at Paducah, Kentucky. Separately, Urenco USA is expanding Eunice from 4.3M SWU/yr, completing +0.7M SWU in 2027 and announcing a further +2.1M SWU on 2026-06-02 (first production 2032, installations to 2036), taking a single site to over 7M SWU/yr.

Announced Western additions plausibly total 6–8M SWU/yr arriving 2029–2036, against the ~6.5M SWU US demand hole. That is a balanced market if every project lands on time, and a materially oversupplied one if reactor demand disappoints or if Russian material returns following any Ukraine settlement.

The CEO’s rebuttal deserves quoting because it is precisely the argument every management makes at the top of a capital cycle: “most of the announced expansions… is based on contracted SWUs. There is really nobody out there that is building what I call just-in-case.” It is contradicted by Centrus’s own build-out, which is government-funded against a contingent order book.

The asymmetry that cuts hardest against Centrus specifically: all of this capacity arrives in exactly the window when Centrus’s first cascade does. Centrus will be a sub-scale, first-of-a-kind producer entering a market simultaneously absorbing Urenco’s mature, largely amortised 2.8M SWU and Orano’s greenfield. And the subsidy that enables it is deliberately structured to fund three rivals at once — which is how policymakers ensure no one earns excess returns.

3.6 Verdict — Industry

Structurally poor for a sub-scale participant; temporarily and artificially attractive. The barriers are real but the profit pool behind them belongs to sovereigns. The current tightness is a geopolitical event, not a structural demand shift, and it is reversible by a diplomatic outcome entirely outside Centrus’s control. The government subsidy is being deployed to create competition, not to protect a champion. Enrichment is a good business to own if you are Urenco; it is a difficult business to enter if you are Centrus.

4. Competitive Position

4.1 The moat test, run honestly

The barriers-to-entry framework admits three genuine competitive advantages: supply-side (cost), demand-side (customer captivity), and economies of scale reinforced by captivity. Centrus has none of them.

(a) Supply / cost advantage — NO, and demonstrably the reverse. Centrus is the highest-cost potential producer in its competitive set. Urenco operates a mature, multi-generation, largely depreciated 17M SWU fleet and expands at incremental cost. Orano has run Georges Besse II commercially since 2011. Rosatom operates 27M SWU on Soviet-legacy assets. Centrus is building a first-of-a-kind AC100M fleet from a standing start while simultaneously constructing the centrifuge factory itself at Oak Ridge. The CEO’s own framing — that reaching “nth-of-a-kind cost” is an achievement still ahead — is an admission that Centrus operates today at first-of-a-kind cost. The one cost edge it possesses is the 2018 TENEX/Orano price reset, which is a purchased input advantage on someone else’s product, expiring 2028 and 2030.

(b) Demand-side captivity — NO. SWU is a physics-defined, fungible commodity: one SWU from Piketon is identical to one from Eunice or Seversk. The 10-K states that “Global LEU suppliers in our highly competitive industry compete on the basis of price and reliability of supply.” There are no switching costs, no network effects, no search costs, no habit. Utilities run competitive RFPs; management confirms “a steady flow of requests for pricing, requests for quotations.” The only stickiness is contract duration — a legal artifact, not captivity. The market-share stability test fails comprehensively: Centrus’s share is under 5%, and that share is a function of a supply contract it did not engineer rather than any competitive position it built.

© Economies of scale — NO. Centrus is the sub-scale participant by more than an order of magnitude. Scale economics in enrichment are real, and they belong to Urenco and Rosatom.

4.2 What Centrus does have — three real, qualified assets

1. The only NRC licence held by a US-owned entity to construct and operate a commercial enrichment plant. Granted April 2007, expiring 2037-04-13, amended in June 2021 to permit HALEU production. Centrus is, on its own accurate description, “the only company with a licence from the NRC actively enriching up to 20% U-235 assay HALEU and that is operating a small scaled HALEU production facility.” This is genuinely scarce and cannot be replicated quickly.
2. US-origin technology free of foreign entanglement. “Longstanding U.S. policy and binding nonproliferation agreements prohibit the use of foreign-origin civilian enrichment technology for U.S. national security missions. Our AC100M centrifuge currently is the only deployment-ready U.S. uranium enrichment technology that can meet these national security requirements.” This is why NNSA notified Centrus in Q4-2025 of an intent to sole-source certain enrichment activities — an opportunity for which Urenco and Orano are legally ineligible.
3. An operating cascade, a licensed site and a cleared, trained workforce — a five-to-eight-year head start over General Matter.

4.3 But call it what it is

A moat is a mechanism the company controls that would cause financial outcomes to deteriorate if removed. Centrus’s franchise is granted by the US government — the NRC licence, the non-proliferation rules, the DOE contracts, the NNSA sole-source eligibility, the import ban itself — and can be modified by the same government, which has in the same breath funded two competitors at $900M each and a substitute technology at $28M. A franchise your regulator is simultaneously diluting is not durable in the Greenwald sense.

The “US-owned” attribute is likewise a legal status, not an operating capability. It protects Centrus in the national-security niche — which is small, opaque and unquantified — but does nothing in the commercial LEU market, where Urenco USA already enriches on American soil at roughly 24 times Centrus’s scale.

The honest formulation: Centrus’s moat is a licence and a passport, not a cost curve or a captive customer. It is worth something real. But it is an option on policy continuity, and options should be valued for their volatility and their strike, not assumed into perpetuity.

4.4 The ROIC test — and why the conventional number is meaningless

At Q1-2026, equity of $775.2M plus debt of $1,178.6M less cash of $1,868.2M leaves roughly $85M of genuine invested capital. ROIC computed on that denominator whipsaws absurdly — third-party data shows 117.5% (2022), 35.3% (2023), 3.4% (2025) — which is the signature of a broken denominator, not a volatile business. No conventional ROIC is reported for Centrus here, and readers should distrust any that they see.

The correct analogs are simpler and more damning. Gross profit has been flat at $110–120M for five years against an inventory and PP&E base that has grown. And the forward test is arithmetic: to earn a 10% return on the roughly $2B of new capital contemplated, operating income must go from $50.2M to over $250M — a fivefold increase — against five years of flat gross profit and falling operating income.

4.5 Head-to-head

4.6 Verdict — Competitive Position

No durable competitive advantage in the Greenwald sense; a narrow, real, but government-dependent and government-dilutable franchise attached to a commodity reseller whose margin expires in 2028. The falsification test for any moat claim is: name the financial outcome that deteriorates if the moat disappears. Here the answer is uncomfortable — remove the political franchise and you have a sub-scale, high-cost enricher with no plant; remove the TENEX contract and you have almost no gross profit at all.

5. Growth History and Forward Opportunities

5.1 What actually drove the growth

Revenue: FY2019 $209.7M (with an operating loss) → FY2020 $247.2M → FY2021 $298.3M → FY2022 $293.8M → FY2023 $320.2M → FY2024 $442.0M → FY2025 $448.7M. On the surface, a respectable trajectory. Decomposed, it is not.

The FY2023 → FY2024 jump of $121.8M (+38%) was not SWU pricing. SWU revenue rose only $38.6M. Uranium revenue rose $42.3M on what the CFO later described as “a large one-time uranium sale in 2024.” Technical Solutions rose $40.9M of cost-plus government reimbursement. Roughly two-thirds of 2024’s growth was a one-time commodity trade plus reimbursed government cost — neither of which is a franchise.

FY2025 (+1.5%) saw SWU revenue rise 21% driven entirely by volume (+23%) against a −1% price, offset by uranium falling 54% as the 2024 one-timer did not repeat.

Q1-2026 saw SWU volume fall 47%, price rise 52%, and net SWU revenue fall 19%.

Over six years SWU revenue compounded at roughly 14.5% — which sounds respectable until you note it is a volume-and-mix story on resold material that produced zero gross-profit growth. Technical Solutions growth is cost-plus reimbursement at a 5.9% gross margin, currently fee-free; that is activity, not economics.

5.2 Contracted and funded versus announced

The single most useful discipline available to a reader of this story is to separate what is contracted from what is announced.

5.3 The $900M DOE contract — better than a service contract, worse than a windfall

The 8-K filed 2026-07-02 (event date 2026-06-30) discloses that American Centrifuge Operating LLC signed a definitive contract with these terms: a firm fixed price of $900 million, paid incrementally as performance-based milestone payments; scope to “deploy a specified amount of enrichment capacity and deliver, by March 2032, one metric ton of uranium enriched as HALEU UF6 to a nominal 19.75 wt% U-235”; two options at DOE’s sole discretion for up to five MTU each at $17M/MTU; and — importantly — “Upon completion of its obligations under the Contract, ACO would obtain title to the deployed enrichment capacity.”

This is genuinely good. It is capital, not revenue: DOE funds a plant that Centrus ends up owning. Management’s characterisation is fair — “another pool of low-cost capital… neither debt nor equity.”

But two qualifications matter. First, it is firm fixed price on a first-of-a-kind six-year industrial build; every dollar of overrun is Centrus’s. Second — and this is the finding that should trouble any underwriter — the 8-K of 2026-04-16 discloses that ACO engaged Geiger Brothers, Inc. as primary construction contractor through 2031-01-30 on a time-and-materials pricing structure, with the aggregate amount payable “capped at $900 million.” Centrus therefore holds a firm-fixed-price $900M receivable from DOE against a time-and-materials $900M-capped payable to a single contractor. Fixed price in, cost-plus out, on the largest line item. That is an adverse risk transfer, and it means the $900M cannot simultaneously fund the project and absorb its overruns.

Add Fluor as EPC, Palantir Foundry (with a $5M prepayment already made), the $560M Oak Ridge centrifuge factory and $60M of Piketon groundwork, and the $900M is plainly a partial contribution to a build management itself calls “multi-billion-dollar.”

5.4 What management will not tell you

The total capital cost of the build-out has never been disclosed. The CEO, asked directly: “As always, I’m very careful. We are not disclosing full costs, and so I will stop there.” Asked about hurdle SWU pricing: “I will not be able to name that because that will point to our cost structure.” FY2026 capital spend is guided to $350–500M — roughly equal to full-year revenue guidance of $450–500M — and the CFO says the first year “is not indicative of the linear spend that I anticipate seeing in the twenty-seven through twenty-nine period.”

The only management-sourced capital-intensity anchor in the entire record is the former CFO’s aside on the Q2-2025 call: “a competitor in Europe who has announced that they’re expanding roughly 2.5 million SWU for the equivalent of USD 2 billion” — approximately $800 per SWU of annual capacity. At that benchmark, a 3.5M SWU Centrus plant implies roughly $2.8B. On Centrus’s own contracted rate (see the valuation section below), the implied figure is materially higher.

Separately, the CEO claims “approximately $300 million in potential cost savings” identified since January 2026 — disclosed against a total project cost that is not disclosed. It is an unquantifiable percentage of an unstated number, and should be treated as an unverifiable assertion.

5.5 Verdict — Growth

Low-quality growth today; genuinely high-optionality growth tomorrow, but unfunded, unpriced and unproven. Six years of revenue growth produced no gross-profit growth. The FY2024 spike was a one-time uranium trade plus cost-plus reimbursement. FY2026 guidance is flat — the midpoint of $450M against FY2025’s $448.7M, in the year management describes a demand super-cycle. The forward case rests almost entirely on items that are contingent, discretionary or non-binding. The one genuinely new, contracted and valuable item of the last twelve months is the $900M DOE contract with title transfer — and it arrives wrapped in fixed-price execution risk.

6. Financial Quality

6.1 The central quality-of-earnings problem

GAAP net income at Centrus is not operating earnings, and the gap is large, growing, and disclosed if you look for it.

The FY2025 10-K states verbatim that investment income “represents interest earned on operating cash… primarily held in money market accounts,” driven by the 0% convertible issuance and the ATM programme. The hypothesis is confirmed by the filing itself: Centrus earned $44.7M in FY2025 by holding the money it raised from selling its own stock and notes.

The magnitudes are stark. Across FY2021–FY2025, $185.5M of $447.1M of cumulative pretax income — 41% — was non-operating. In Q1-2026, more than 100% was: operating income was $0.8M against $12.5M of pretax income. And the pension tailwind that flattered FY2021 is now structurally dead: the projected benefit obligation fell from $527.3M (2022) to $31.0M (2025) via two group-annuity buyouts transferring roughly $420M, and the line has flipped from +$14.7M of income to −$6.8M of cost.

That last point deserves emphasis because it reframes the historical record. FY2021’s headline pretax income of $135.9M was roughly half pension credits. Reported diluted EPS of $12.46 for FY2021 — a figure that appears in third-party databases — is in fact the company’s non-GAAP number, which adds back $37.6M of “distributed earnings allocable to retired preferred shares.” GAAP diluted EPS for FY2021 was $9.75, and even that was inflated by a $39.1M deferred-tax valuation-allowance release. FY2021 is not a run-rate under any construction.

6.2 Normalized earnings

Two defensible normalizations, both far below the headline:

• Operating income only, taxed normally: $50.2M × (1 − 23%) ÷ ~19.9M diluted shares ≈ $1.94/share.
• Operating income less run-rate convertible interest ($14.6M) and the pension cost ($6.8M), taxed at 21%, on ~21.5M fully diluted shares ≈ $1.06/share.

Against reported GAAP diluted EPS of $3.90 for FY2025, roughly 50–73% of reported earnings is not operating earnings. Trailing-twelve-month diluted EPS reconciles to approximately $2.75 (third-party sources showing $3.02 do not tie to the filings).

6.3 The tax shield

Federal NOL carryforwards stand at $636.6M ($505.2M expiring in 2037, $131.4M with no expiry); state NOLs of $491.3M are fully reserved. Cash federal tax has been zero for three consecutive years. The effective rate was 9.4% in FY2025, approximately 0% in FY2023 and FY2024, and a $39.1M benefit in FY2021.

Two observations. First, the FY2025 valuation-allowance release was justified, per the filing, because “the positive evidence of increased future investment income supported the release” — the auditors credited the money-market account, not the business. Second, and more telling, a $355.4M valuation allowance still stands, which is management’s own formal assertion that it does not expect to generate enough taxable income to use the remaining NOLs.

Normalizing to a 21–25% rate is a material haircut to reported earnings, and the shield will not survive the transition to a capital-intensive operating company indefinitely.

6.4 Margins — why they fell

Consolidated gross margin fell from 40.1% (FY2022) to 25.2% (FY2024), recovering to 26.2% (FY2025). Three causes, in order of importance:

1. FY2022’s 55.4% LEU-segment margin was the arbitrage at its widest — 2018-struck input prices against a market repricing violently upward post-invasion. It was a windfall, not a baseline.
2. Mix shift to Technical Solutions, which carried a 5.9% gross margin in FY2025 because the Phase 2 fee remains undefinitized and unpaid.
3. Input cost re-pricing. FY2025’s margin recovery came from cost, not price — SWU volume +23% against a −1% price and a 13% decrease in average unit cost. Q1-2026 reversed: price +52%, cost +45%.

Management’s own forward steer, from the Q2-2025 call, is not for expansion: “it is reasonable to expect margin levels to remain within the range we’ve seen over the past few years.”

6.5 Cash flow, balance sheet and liquidity

Cumulative five-year operating cash flow was $167.7M against cumulative net income of $462.6M — a 36% conversion rate. That gap is the deferred-revenue drawdown and working capital, and it is the clearest signal that reported earnings have not been cash.

The balance sheet is, on its face, fortress-like: cash of $1,868.2M, inventories of $336.0M, total assets of $2,433.2M against total liabilities of $1,658.0M, equity of $775.2M, and net cash of $689.6M. The current ratio is 5.7x.

But three qualifications matter. First, the cash is borrowed and issued, not earned: roughly $1.73B was raised in fifteen months through two convertible issues and two ATM programmes. Second, the converts are a hard $1.2B cash claim ahead of equity — principal settles in cash, maturing 2030 and 2032, straddling the 2029 plant startup. Third, net fixed assets are only $59.5M (and were $29.5M at year-end 2025): Centrus owns almost no plant, because the Piketon centrifuges are owned by the DOE and the site is leased from DOE on a term currently running only to 2027-06-30.

Working capital is heavy — inventories of $336M and a cash conversion cycle around 375 days — which is normal for the fuel cycle but consumes cash as volumes grow.

6.6 The new non-GAAP measure

In Q1-2026 — precisely the quarter build-out spending began and GAAP operating income collapsed 96% to $0.8M — management introduced “adjusted net income” and “adjusted EPS.” The adjustment converts GAAP net income of $10.0M / $0.45 per share into $23.5M / $1.05 per share, a 135% uplift, by adding back “$17 million of growth expenses in our advanced technology costs and $400,000 in stock-based compensation.”

The terms being added back are real, recurring cash operating costs — the CFO’s own examples include “manufacturer readiness and security training” — and he concedes they persist: “Going forward, we can expect to have a certain level of these types of expenses flow through our income statement.” An adjustment you expect to make every quarter for six years is not an adjustment; it is the cost of the business. The measure appears five times in the Q1-2026 10-Q and zero times in the FY2025 10-K, the FY2024 10-K or the Q1-2025 10-Q. This memo uses GAAP.

6.7 Verdict — Financial Quality

Economics have not improved with scale; they have deteriorated. Revenue up 53% from FY2022 to FY2025 produced gross profit up 3.8% and operating income down from $59.7M to $50.2M. Cash conversion is 36%. Two-thirds of the most recent year’s pretax income was non-operating. Retained earnings stand at $11.5M after twenty-eight years of corporate existence. On today’s filings this is a finite trading spread on two expiring supply contracts, plus a currently fee-free cost-plus government services contract, with a $1.9B money-market fund attached. The bull case is a venture-stage construction option and should be underwritten as one.

7. Capital Allocation

7.1 The financings — genuinely well executed

This is the strongest part of the story and it deserves full credit.

Issuing $390.4M of equity at an average of $269.21 per share — 73% above today’s $156.05 — is excellent capital allocation. So is raising $805M at a 0% coupon. So is the decision not to issue in Q1-2026: “We don’t feel any pressure to be actively out there raising capital in a down market… we just didn’t feel it provided the right shareholder value.” Only 7,022 shares were issued in Q1-2026. Management read its own security correctly, monetised a narrative-driven melt-up, and stopped when the price no longer justified it. That is rarer than it should be.

Historic capital allocation is also clean: the Series B preferred was retired cheaply in 2021 for $43.3M, there has been no value-destructive M&A, and there are no buybacks or dividends — appropriate for a company that needs every dollar for capex.

Dilution. Full conversion of both notes would add 7,634,057 shares, or +38.8%. However, principal settles in cash, so economic dilution at $156 is limited to the in-the-money spread on the 2030 notes — roughly 1.55M shares, or +7.9%. Notably, a thorough search across the FY2024 10-K, FY2025 10-K and Q1-2026 10-Q found no capped-call transactions on either issue, which is unusual for a $1.2B convertible programme and means there is no purchased hedge above the strikes.

7.2 But 100% of the deployment is still ahead

Capital expenditure over the last five and a quarter years totalled $50.5M. The company has barely spent. Q1-2026 marked the turn: $23.2M of capex plus $22.0M of non-capex expansion spend, against $35.1M of operating cash burned in the quarter, with the CFO warning that “the pace of our CapEx and non-CapEx spend [will] accelerate throughout the year.”

So the capital-allocation verdict must be split. Financing execution: A−. Deployment execution: entirely unproven, on a first-of-a-kind nuclear industrial build, by a management team that has never operated at commercial scale, at a company whose predecessor abandoned the American Centrifuge Project after spending billions. And the structure of the deployment is adverse, as the DOE-contract discussion above establishes: firm-fixed-price revenue against time-and-materials cost.

There is also an unflattering second-order effect worth naming. The cash raised now generates $44.7M a year of interest income that flatters EPS — and that income disappears precisely as the cash is spent on capex, which is the entire plan. The market is capitalising a treasury yield that the business model requires to be extinguished.

7.3 Insider behaviour — the loudest datum in the file

A review of the Form 4 corpus covering the move from roughly $125 through the $436 peak and down to $156 produced an unambiguous result:

• Zero code-P open-market purchases. Nobody bought a 64% drawdown.
• Two discretionary sales, neither under a 10b5-1 plan: the former CFO on 2025-05-30 at approximately $125.86, and — more pointedly — the sitting CFO on 2026-05-11, selling his entire 306-share position at $203.55, to zero.
• The June-2026 Form 4 clusters are routine RSU grants and tax withholding, not open-market activity.

Aggregate insider ownership is 0.75% of Class A shares (142,099 shares) across all directors and executive officers. CEO Amir Vexler owns 4,392 shares — approximately $685,000 — against $2.17M of 2025 compensation. Much of the nominal director “ownership” consists of vested-but-unsettled RSUs.

There is no activist. The three 5% holders are D.E. Shaw (5.1%, of which 22% is call options), BlackRock (5.0%, on a 13G that is twenty-seven months stale) and the Global X Uranium ETF (5.0%, passive thematic).

7.4 Compensation and governance

The incentive design is poor and poorly disclosed. The long-term incentive plan turns on a single three-year cumulative NET INCOME goal — a metric that includes the $44.7M-a-year of money-market interest and the valuation-allowance releases. The annual bonus “Goal Three” is explicitly “revenue, operating income and cash balances.” There is no ROIC, ROE, per-share, free-cash-flow or relative-TSR metric anywhere in the design. Every target and achievement threshold is withheld as competitively sensitive; payouts ran 121% (2025) and 165% (2024). The CEO’s LTI target is only 33% of base salary against a 100% bonus target, so 82% of his 2025 pay was cash.

Governance reinforces the picture: a Delaware exclusive-forum bylaw adopted March 2026, Rule 14a-19 nomination hurdles, and a 4.99% poison pill extended for a seventh time on 2026-06-18 (which resolves the otherwise-mysterious 8-A12B/A filed that day). Combined with 0.75% insider ownership, this is an insulation package rather than an alignment package.

7.5 Verdict — Capital Allocation

Split, and the split matters. Financing has been opportunistic, well-timed and shareholder-friendly — selling equity at $269 and issuing zero-coupon converts into a mania are genuinely value-creating acts, and they are why the company has $689.6M of net cash and the ability to attempt this at all. But every dollar of deployment is ahead, the structure transfers overrun risk onto Centrus, the incentive plan pays on a net-income figure that treasury interest inflates, and not one insider bought a single share through a 64% decline. Credit the treasurer; withhold judgement on the operator.

8. Changes and Headwinds — Last Two Years

8.1 The 2028 convergence

The single most important structural fact in this report is that everything expires at once:

• TENEX Supply Contract: ends 2028
• DOE import-ban waiver authority: terminates end-2027; from 2028-01-01 the prohibition is absolute with no waiver mechanism
• Russian export decree licensing TENEX shipments: runs through 2027-12-31
• Russian Suspension Agreement quota set aside for Centrus: through 2028
• Canadian ocean-shipper’s sanctions permit: expires March 2027
• Orano Supply Agreement: deliveries end 2030
• Centrus’s first new cascade: online 2029

And the fragility is not merely contractual. The Q4-2025 call disclosed that a single scheduled and permitted shipment from Russia that failed to leave on time moved the full-year gross margin and net income. The 10-K catalogues further chokepoints: TENEX’s financial institutions have had difficulty accepting US-dollar payments since the January-2025 OFAC determination, and payments have been mutually delayed; the shipper uses Chinese-built vessels currently exempt from USTR Section 301 port fees that Centrus says “would be cost prohibitive” if withdrawn; and one processing facility has already refused to receive natural uranium destined for TENEX.

Management has been asked directly about the resulting gap, four times across three calls. The CEO’s answer: “You’re asking me to speculate. I would not be able to answer that. I do think… there is going to be an extremely tight market in the years that you’re mentioning… these are the years that I’ve put a question mark on as well.”

INTERPRETATION. The CEO frames the coming tightness as bullish for pricing — which it is — while never addressing that Centrus’s own cost of goods disappears in the same window. A tight market helps a producer. Centrus in 2028 is not yet a producer.

8.2 Verdict — Changes and Headwinds

Net-weakening on economics; net-strengthening on optionality. The genuine positives are real: the August-2025 waiver de-risked 2026–27, the June-2026 DOE contract is firm and confers title, NNSA sole-source intent is a uniquely Centrus opportunity, and ~$1.9B of cash was raised at excellent prices. The negatives are structural: gross profit flat for five years, Technical Solutions working fee-free since November 2024, Q1-2026 operating income down 96%, guidance flat, a promotional new non-GAAP measure introduced exactly when GAAP collapsed, total project cost undisclosed, the largest competitor expanding aggressively, and the 2028 convergence unaddressed.

9. Risk Analysis

Concentration of risk. Risks 1, 2 and 3 are not independent — they are one risk expressed three ways: Centrus’s current profits expire before its future profits begin, and the bridge requires capital it does not have. Risk 6 compounds it, because the capacity arriving in 2029–2036 lands precisely when Centrus needs pricing to be strong. An investor in this security is, in substance, underwriting a timing problem with a financing problem attached.

10. Valuation Discussion — Embedded Expectations

10.1 Rebuilding enterprise value at the live price

Third-party enterprise values for this name are computed off stale fiscal-year-end price snapshots and should not be used. On convertible face value of $1,207.5M rather than carrying value, EV is approximately $2,412M; both readings round to ~$2.4B.

10.2 The multiples

10.3 Own-history context — the central inversion

The P/B percentile must not be read as cheapness. Shareholders’ equity was negative for eight consecutive years — 2015 through 2022, ranging from −$74.1M to −$336.9M — and only marginally positive in 2023–24 ($32.3M and $161.4M). A price-to-book ratio computed against a negative or de-minimis denominator is undefined or absurd, so the ten-year percentile series is ranking today’s 4.52x against a history of meaningless numbers. Book equity exists today only because the 2025 converts and ATM created it.

The load-bearing readings are P/E at the 95th percentile and P/S at the 90th percentile of the stock’s own history. This is the inversion at the heart of the report: after a 64% drawdown, LEU still trades near the richest multiples it has ever commanded. The drawdown de-rated the price without de-rating the valuation, because earnings fell faster than the stock did — TTM EPS of ~$2.75 against $3.90 in FY2025 and $5.44 in FY2023.

10.4 What the market must be underwriting

Reverse-engineering $2,380M of enterprise value: assume steady state in 2032 (first capacity 2029 plus a three-year ramp), a required return of 10–15% (small-cap, single-asset, policy-dependent, 68% idiosyncratic vol — 12% is the reasonable centre), and a terminal EV/EBITDA of 8–12x (Urenco is a private, contracted industrial, not a compounder).

The centre case requires roughly $470M of EBITDA by 2032 — 7.8x the FY2025 figure of $60.1M — with a plausible range of $350–690M.

Translating that into physical capacity is the step that makes the expectation legible. Urenco’s audited FY2025 results (revenue €2,096.2M, EBITDA €804.2M, a 38.4% margin on roughly 18M SWU of capacity) imply approximately €45 (~$50) of EBITDA per SWU at world-class scale with largely amortised plant. Meanwhile, enriching to 19.75% at 0.25% tails requires 41.0 SWU per kilogram of HALEU product, so Centrus’s funded 12 MT/yr HALEU target equates to roughly 492,000 SWU/yr.

This is the finding. The $900M DOE contract buys 492,000 SWU of capacity — a contracted, contemporaneous capital-intensity benchmark of $1,835 per SWU installed. To support today’s enterprise value, Centrus must reach roughly 9 million SWU per year — nineteen times the funded expansion, and nearly twice the entire Urenco USA plant — at a capital cost of roughly $11–17 billion, against $689.6M of net cash and one $900M award. The funded 12 MT HALEU line, run flat out at premium pricing and a generous 45% margin on DOE-funded plant, generates perhaps $50–90M of EBITDA — around 15% of what is embedded in the price.

10.5 The either/or the bull case cannot escape

There is an unavoidable contradiction in the optimistic construction:

• If the government funds the $11–17B, Centrus is a cost-plus contractor earning a single-digit-to-low-double-digit fee, not an owner earning $50/SWU. Contractors are worth 8–12x EBITDA, not 14–20x.
• If Centrus funds it as an owner, it must raise $11–17B against a $3.07B market capitalization — four-to-six-fold dilution unless customer prepayments and non-dilutive federal capital carry nearly all of it.

Management’s own language concedes the scale of the gap: the build must be “matched several times over with billions of dollars in capital, including other non-dilutive, non-debt funding as well as customer contracts.” The bull case cannot simultaneously assume owner-level margins and government-funded capex. Today’s price appears to assume both.

Note too that the ~$1.87B of cash cannot be treated as a free rebate against the market cap without also stripping the $44.7M of interest income from earnings. Several constructions net the cash but keep the income. Doing both consistently is what produces the 80x figure.

10.6 Scenarios

INTERPRETATION. The current enterprise value is roughly twice the base case and is justified only by the bull case in full — and the bull enterprise figure is before the dilution required to fund it, so per-share bull outcomes are materially worse than enterprise bull outcomes. The distribution is not favourably skewed: the bear case costs 80–90% of enterprise value, while the bull case is approximately par.

All 2032 figures are estimates derived from Urenco’s disclosed per-SWU economics and Centrus’s own contracted capex rate; none is a company forecast.

10.7 Comparable companies

There is no clean listed enrichment comparable. Urenco is owned by the UK and Dutch governments and two German utilities; Orano is roughly 90% French state-owned; TENEX is Russian state-owned. Centrus is the only listed Western enrichment-exposed equity — which is simultaneously the bull’s scarcity argument and the reason no market-clearing multiple exists.

INTERPRETATION. LEU at 39.6x EV/EBITDA trades at roughly 2.8x Constellation’s ~14x and above BWXT’s ~29x forward — while owning less plant, earning a run-off margin, and possessing a weaker moat than either. The wider nuclear cohort is likewise trading at or near its richest-ever own-history multiples; LEU is the most extreme member on EV/EBIT. The factor-model “related stocks” list — five uranium ETFs in the top six — is factor noise rather than a comp set, but its message is informative: the market trades LEU as nuclear-theme beta, not as a company.

11. Variant Perception

11.1 Consensus

Most of a fourteen-to-seventeen analyst panel rates the stock Buy, but the dispersion is the signal, not the average. Recent targets: Needham Buy $264; Truist Buy $215 (initiated 2026-07-14); BofA Neutral $205; Roth Neutral $195 (cut from $230); UBS Neutral $170 — against a $156.05 tape. The newest initiations and the Neutral cohort cluster at $170–215, well below the older $264. FY2027 EPS estimates span $0.51 to $9.04 — an eighteen-fold spread on a two-year-out estimate.

INTERPRETATION. Revisions are falling and the distribution is bimodal. The sell side does not have an estimate problem; it has an option-pricing problem, and is publishing point estimates for a binary. The consensus belief is roughly: the Russian ban plus AI-driven nuclear demand hands the only listed Western enricher a policy-protected monopoly; the DOE contract validates it; execution is a detail.

11.2 The strongest bull case

Enrichment is a genuine Western choke point. Russia holds roughly 44% of global enrichment capacity and supplied 27% of US enrichment services as recently as 2023. The ban is law, the waivers die on 2028-01-01, and Urenco and Orano cannot close the gap alone before the mid-2030s. Centrus is the only listed vehicle, holds the only US-owned centrifuge technology base, is the only entity eligible for NNSA national-security enrichment work, has a $2.4B contingent backlog, a signed $900M federal contract that confers title, and $1.87B of cash to move fast. If Washington comes to treat enrichment the way it treats shipbuilding — sole-source, cost-plus, multi-decade, strategically indispensable — Centrus becomes a sovereign-backed franchise and today’s multiple on a run-off trading book is an irrelevance.

11.3 The strongest bear case

Centrus today is a reseller of Russian and French SWU whose entire gross margin derives from a 2018 price reset expiring with TENEX in 2028 and Orano in 2030. It owns essentially no plant. Its $2.4B backlog is contingent on funding it does not have. The current enterprise value requires roughly 9M SWU/yr and $11–17B of capex against a funded 0.49M SWU. Gross profit has been flat for five years while revenue grew 53%; two-thirds of last year’s pretax income was money-market interest on borrowed cash; the effective tax rate was 9.4% against a $355.4M valuation allowance that is management’s own statement it expects not to earn enough to use its NOLs. Meanwhile Urenco is adding 2.8M SWU with its own balance sheet, Orano is building a $5B greenfield with an identical $900M subsidy, and a start-up has DOE land at Paducah. The HALEU customers — Oklo, NuScale, X-energy — have collectively never operated a commercial reactor. Strip the interest income, normalize the tax rate, and the stock trades at 80x.

11.4 The assumptions that actually matter

1. Does owned production replace the run-off arbitrage before 2028–2030? This is a question of timing, not existence — and the disclosed timeline says no.
2. Who funds $11–17B, and at what dilution or what contractor margin? The either/or set out in the valuation section.
3. Does HALEU demand arrive on the 2029–2032 schedule, or slip with its customers?
4. Does SWU pricing hold $180–200, or mean-revert as Urenco and Orano capacity lands?
5. Is the terminal multiple a contracted-industrial 8–12x or a strategic-monopoly 14–20x?

11.5 Falsification tests

The bull case is falsified by: a DOE award structured as cost-plus with a single-digit fee (proving contractor rather than owner economics); TENEX/Orano roll-off arriving before owned production with no margin bridge; first new capacity slipping past 2029; SWU spot printing below ~$160; or a large equity raise at depressed prices to fund capex.

The bear case is falsified by: a second, materially larger federal award (>$3B) or a customer-prepaid expansion funding >2M SWU; a binding, priced, multi-year HALEU offtake from a developer that has broken ground; TENEX or Orano extensions on 2018-reset economics; or SWU term prices ratcheting above ~$220 with Centrus’s realised-price gap closing.

11.6 The positioning read

LEU is a falling knife, not an abandoned value name — and the factor data makes this precise. The stock carries a positive momentum loading (base-model beta 1.18; Base+Sector Market 1.37, SmallSize 1.15) into a market where momentum is working (+15.5% over 252 days, z-score 0.67). It is nonetheless down 47.6% over six months, with m3 and m6 Sharpe ratios of −0.71 and −0.84 against a y3 Sharpe of +0.79. A high momentum beta with deeply negative realised momentum in a friendly momentum regime is the signature of a crowded thematic trade losing its crowd — the y3 and y1 numbers are the same stock on either side of the break.

It is emphatically not a value name. Value has been the best-performing factor over 252 days (z-score 1.70), LEU carries no meaningful Value loading, and it sits at the 95th percentile of its own P/E history. With R² of 0.31–0.35 and specific volatility of 68.2%, roughly two-thirds of the variance is idiosyncratic — this is a stock-specific unwind, not a factor casualty.

Positioning and flows. Short interest is 22.3% of float (2026-07-02) — approximately 4.0M shares against an 18.14M float and 874k average daily volume, or 4.6 days to cover. That is high enough to fuel violent squeezes in both directions and explains the +31% and +37% counter-rallies inside a −64% drawdown. On index flows, LEU was removed from multiple Russell indices in June 2026 and added to the S&P SmallCap 600 effective 2026-07-14. The tape’s verdict on the latter was unambiguous: on the rebalance date the stock traded 3,061,035 shares — 3.5x normal — and fell 8.8%. S&P inclusion should not be read as a positive technical catalyst here.

12. Fact vs. Interpretation

13. Open Questions

1. What exact percentage of LEU-segment cost of sales is Russian-origin? The company discloses only “>>50% of delivered volume through 2027.” Without it, the 2028 gross-profit cliff cannot be sized precisely. This is the single most valuable undisclosed number in the filings.
2. What is the total capital cost of the build-out? Management explicitly refuses: “We are not disclosing full costs.” The gap between “about $2.8 billion currently” and “multi-billion-dollar” is unquantified.
3. What replaces TENEX gross profit in 2028–2029, before the first cascade produces at scale? Asked four times across three calls; never answered.
4. What is the fee or margin structure inside the $900M fixed-price contract? This single disclosure determines whether Centrus is an owner or a contractor and swings the terminal multiple from ~8x to ~14x.
5. What is Centrus’s realised $/SWU on the legacy book versus the ~$200 spot, and what is the roll-off schedule of the 2018-reset economics? The 10-K says the lag is “several years” without quantifying it.
6. NNSA sole-source: what scope, what value, what timing? Management has declined to disclose. Could be material or immaterial.
7. Does Option 1b of the HALEU Operation Contract get exercised, and on what economics? DOE has already stated the estimated cost is “insufficient to support full performance.”
8. When and how does the undefinitized Phase 2 fee get resolved? Centrus has performed this work fee-free since November 2024.
9. Will Urenco’s 2.8M SWU, Orano’s Project IKE and General Matter’s Paducah plant all land? If so, the US market is likely oversupplied by the mid-2030s.
10. What happens to SWU pricing on a Ukraine settlement restoring Rosatom’s 27M SWU to Western markets? Neither management nor the sell side models this. It is the bear case’s detonator.

14. What Must Be True

14.1 For the bull case

14.2 For the bear case

The shared clock. Both cases resolve on the same calendar. FY2027 guidance, issued in early 2027, is the first disclosure that must confront the 2028 cliff directly. Before then, the questions to watch are the FY2026 capex run-rate (does the build curve validate the timeline?), any second federal award, and whether the Oklo LoI converts into a priced, binding contract.

15. Source Appendix

See the accompanying source appendix document. Primary sources comprise the FY2021–FY2025 Forms 10-K, the Q1-2026 and prior Forms 10-Q, the 2021–2026 Form 8-K corpus (88 filings), the Form 3/4/5 insider corpus (107 filings), DEF 14A proxy statements, and Schedule 13D/G filings — all retrieved from SEC EDGAR (CIK 0001065059) and mirrored locally. Management commentary is drawn from the Q2-2025, Q3-2025, Q4-2025 and Q1-2026 earnings-call transcripts. Market and factor data are as of 2026-07-16/17. Peer framing draws on the public filings and disclosures of Cameco, NuScale, BWXT, Constellation and Oklo, attributed inline.

APPENDIX A — Standard Diligence Questionnaire

Centrus Energy Corp. (NYSE: LEU) — 2026-07-18

Supplemental to the analysis above. Fact / Interpretation / Assumption labels are applied where the distinction matters.

General

What thoughtful questions have other investors asked about this company?

The best questions asked on the calls have been about the 2028 gap, and they have gone unanswered. On the Q3-2025 call an analyst asked directly: “Has there been any shift in political commentary out of Washington about the Jan 1, 2028 deadline for Russian imports?” and followed up on what the business looks like in 2028–2030 if the deadline stands. The CEO: “You’re asking me to speculate. I would not be able to answer that… these are the years that I’ve put a question mark on as well.” (FACT.)

The second-best question came from H.C. Wainwright on the Q1-2026 call, catching that SWU selling price rose 52% while SWU cost rose 45% — and asking whether gross margin should therefore keep expanding. The CFO deflected to “look at this on an average basis over several quarters.” (FACT.) That exchange exposes the reseller economics more clearly than any disclosure.

Questions that are conspicuously not being asked: what the total build-out costs (management refuses to say); what percentage of cost of sales is Russian-origin; how the $2.4B contingent backlog converts; and what happens to the $44.7M of interest income when the cash is spent on capex. (INTERPRETATION.)

Cyclicality & Earnings Nature

Are earnings at a cyclical high or low? Neither, precisely — they are at an artificial level. (INTERPRETATION.) Operating income of $50.2M (FY2025) sits below FY2021’s $68.3M despite 50% more revenue. But reported net income of $77.8M is elevated by $44.7M of money-market interest and a 9.4% tax rate. The gross margin, meanwhile, is past its cyclical peak: the LEU segment earned 55.4% in FY2022 and 32.2% in FY2025, because FY2022 was the arbitrage at its widest. (FACT.) The right characterisation is a run-off asset near the end of its life, temporarily supplemented by treasury income.

Driven by the external environment or internal actions? Overwhelmingly external. (INTERPRETATION.) The margin derives from a contractual price reset that took effect in 2019 — a mechanism Centrus did not engineer — combined with a war-driven repricing of SWU. The interest income derives from interest rates. Internal action accounts for the capital raising (genuinely good) and cost control (modest).

How stable are revenues? Lumpy at the quarter, steadier at the year. The average individual SWU order fulfilled in 2025 was ~$10.2M, so a handful of shipment timings swings a quarter: Q1-2026 SWU volume fell 47%. (FACT.) Management repeatedly insists “annual results are more indicative.” Annual revenue has ranged $247M–$449M over six years with one discontinuity (the FY2024 one-time uranium sale).

Outlook for products/services? The product — enrichment services — has an excellent long-term outlook. Centrus’s position in supplying it is the question. FY2026 revenue guidance of $450–500M has a midpoint flat against FY2025’s $448.7M. (FACT.) The company is guiding to zero growth in the year it describes a demand super-cycle.

How big will this market be — growing, shrinking, domestic or international? World enrichment demand is roughly 50M SWU/yr against non-Russian capacity of roughly 36M SWU. (FACT, per World Nuclear Association data cited in the 10-K.) If Russia genuinely exits, that is a ~14M SWU structural deficit, of which management sizes the near-term domestic piece at ~6.5M SWU. The market is growing on reactor restarts, life extensions, uprates and datacenter-driven procurement. Centrus is roughly two-thirds domestic; international customers are about a third of LEU-segment revenue. The market is genuinely attractive; the question is who captures it — and announced Western additions of 6–8M SWU landing 2029–2036 suggest the answer is the incumbents. (INTERPRETATION.)

Business Quality & Competitive Moat

Is the industry getting more or less competitive? More. (FACT, then INTERPRETATION.) DOE awarded $900M each to three parties — Centrus, Orano and the start-up General Matter — plus $28M to Global Laser Enrichment, a substitute technology. Urenco USA is expanding from 4.3M SWU to over 7M. Orano is building a ~$5B greenfield at Oak Ridge. Two years ago Centrus was the only US-owned enrichment story; by 2032 it will be one of four domestic suppliers, and the smallest.

How profitable is the business (ROIC, ROE)? Conventional ROIC is not meaningful here and is not reported below. (INTERPRETATION.) Invested capital net of cash is approximately $85M (equity $775.2M + debt $1,178.6M − cash $1,868.2M), so ROIC computed conventionally whipsaws between 3% and 117% across recent years — a broken denominator, not a volatile business. The honest measures: gross profit has been flat at $110–120M for five years; operating margin is 11.2%; retained earnings stand at $11.5M after twenty-eight years. (FACT.)

How profitable is the industry — how many competitors, what barriers to entry? Barriers are genuinely enormous: capital intensity, decade-long lead times, NRC licensing, non-proliferation treaty constraints. Four suppliers hold >95% of world capacity. (FACT.) But the industry’s profit pool is largely captured by sovereigns — Rosatom, Urenco’s government shareholders, the French state — which do not require commercial returns and depreciate publicly funded assets accordingly. High barriers plus state ownership equals a structurally unattractive industry for a small private participant. (INTERPRETATION.)

Can the business be easily understood? The current business, yes, once you see through the framing: buy SWU wholesale under legacy contracts, resell at market, book the spread. The future business — a multi-billion-dollar first-of-a-kind centrifuge build with undisclosed total cost, contingent backlog and government dependency — is genuinely hard to underwrite, which is why FY2027 consensus EPS ranges from $0.51 to $9.04. (FACT.)

Can it be undermined by foreign low-cost labour? Not by labour — enrichment is capital- and energy-intensive, not labour-intensive. But it is directly undermined by foreign state-subsidised capital, which is the same threat in a different form. Rosatom’s 27M SWU and Urenco’s 17M SWU were built with public money. (FACT/INTERPRETATION.)

Do brands matter? No. The 10-K is explicit: suppliers “compete on the basis of price and reliability of supply.” (FACT.) A SWU is defined by physics and is identical regardless of origin. What matters instead is provenance as a legal category — US-origin technology is required for national-security work, which is a regulatory attribute, not a brand.

What is the nature of competition? Competitive RFP-based procurement by utilities on price and supply reliability, layered over a geopolitical allocation of who is permitted to sell into which market. (FACT.)

Customers’ switching costs? Effectively nil. There are no technical switching costs for a fungible commodity. The only stickiness is contract duration — a legal artifact rather than captivity. (INTERPRETATION.) Centrus fails Greenwald’s market-share-stability test comprehensively: its share is under 5%, and even that is a function of a purchased supply contract.

Financial Condition & Balance Sheet

Assets not fully recognised on the balance sheet? Yes, and they are the most valuable things Centrus owns. (INTERPRETATION.) The NRC licence to construct and operate a commercial enrichment plant (granted 2007, running to 2037, amended 2021 for HALEU) carries no meaningful balance-sheet value but is genuinely scarce — the only such licence held by a US-owned entity. The $636.6M of federal NOL carryforwards are partially recognised, with a $355.4M valuation allowance still standing. The AC100M technology base and the cleared, trained workforce are likewise unrecognised.

Off-balance-sheet liabilities? Modest but real. Pension liabilities of $73.3M remain, though the projected benefit obligation was reduced from $527.3M (2022) to $31.0M (2025) via two group-annuity buyouts. (FACT.) The more material exposure is contractual rather than accounting: the Geiger Brothers construction contract runs on time-and-materials pricing against a firm-fixed-price DOE receivable, so cost overruns above $900M land on Centrus without appearing anywhere today. The Piketon lease from DOE currently runs only to 2027-06-30.

How conservative is the accounting? The accounting is unremarkable; the presentation is not. (INTERPRETATION.) In Q1-2026 — the exact quarter GAAP operating income collapsed 96% to $0.8M — management introduced “adjusted net income” and “adjusted EPS,” converting $0.45 of GAAP EPS into $1.05, a 135% uplift, by adding back $17M of “growth expenses” that are recurring cash operating costs. The measure appears five times in the Q1-2026 10-Q and zero times in the FY2025 10-K, the FY2024 10-K or the Q1-2025 10-Q. Separately, the headline $3.9B backlog sits against ~$0.6B of ASC 606 remaining performance obligations. Both are legitimate under the rules and both flatter.

How CapEx-hungry is the business? Historically, barely at all — total capex over five and a quarter years was $50.5M, and net fixed assets are $59.5M, because DOE owns the Piketon centrifuges. (FACT.) Prospectively, enormously: FY2026 capital spend is guided to $350–500M, roughly equal to full-year revenue, and the CFO warns the first year “is not indicative of the linear spend that I anticipate seeing in the twenty-seven through twenty-nine period.” The business is transitioning from asset-light broker to capital-intensive industrial, which is precisely the transition the market is capitalising and the company has not yet demonstrated it can execute.

Capital Allocation & Management

How much FCF does the business generate, how does management use it, what is the philosophy? Cumulative five-year operating cash flow was $167.7M against $462.6M of cumulative net income — 36% conversion. (FACT.) The gap is the deferred-revenue drawdown ($216.4M at Q1-2025 to $112.8M at Q1-2026) and working capital. Q1-2026 burned $35.1M of operating cash plus $23.2M of capex. The philosophy is explicit and coherent: retain everything, raise more, and spend it on the build. There are no dividends and no buybacks, which is correct for this situation.

Significant acquisitions recently? None. (FACT.) No M&A history of consequence, which is a mark in management’s favour.

Buying back shares? No. The company is a net issuer, and thoughtfully so.

Issuing large amounts of new shares to insiders? Not disproportionately. Stock-based compensation is small ($0.4M added back in the Q1-2026 non-GAAP reconciliation). The dilution that matters is external: two convertible issues totalling $1,207.5M face — $402.5M at 2.25% due 2030 (conversion $97.50, 4,128,201 shares) and $805.0M at 0% due 2032 (conversion $229.62, 3,505,856 shares) — plus 2,866,261 ATM shares in FY2025. (FACT.) Full conversion would add 38.8% to the count, though principal settles in cash, limiting economic dilution at $156 to roughly 7.9%. No capped calls were found on either issue.

Compensation policy of directors/management? Poor and poorly disclosed. (INTERPRETATION, on FACTs.) The long-term incentive plan turns on a single three-year cumulative net income goal — a metric inflated by the $44.7M/yr of money-market interest and by valuation-allowance releases. The annual bonus explicitly references “revenue, operating income and cash balances.” There is no ROIC, ROE, per-share, FCF or relative-TSR metric anywhere. All targets are withheld as competitively sensitive; payouts ran 121% (2025) and 165% (2024). The CEO’s LTI target is only 33% of base against a 100% bonus target, so 82% of his 2025 pay was cash. A plan that pays for holding cash, at a company whose thesis requires spending it, is misaligned.

Motivations of management? Mixed, and the insider data is the most honest signal available. (FACT.) Across the move from ~$125 through the $436 peak to $156 there were zero code-P open-market purchases and two discretionary (non-10b5-1) sales — the former CFO at ~$125.86 in May 2025, and the sitting CFO selling his entire 306-share position to zero at $203.55 on 2026-05-11. Directors and all executive officers together own 0.75% of Class A (142,099 shares); the CEO owns 4,392 shares (~$685,000) against $2.17M of 2025 pay. Governance adds a Delaware exclusive-forum bylaw (March 2026), Rule 14a-19 nomination hurdles, and a 4.99% poison pill extended for a seventh time on 2026-06-18. Nobody bought a 64% drawdown. Against that, management’s financing decisions — selling equity at $269.21 and declining to issue in a down market — genuinely served shareholders.

Valuation & Market Data

Is the stock an ADR, MLP, or K-1 issuer? No. Centrus is a Delaware corporation filing standard US forms and issuing Form 1099. Class A common stock, NYSE-listed (uplisted from NYSE American in December 2025). Note the certificate of incorporation contains foreign-ownership restrictions tied to the NRC licence.

Dividend policy? No common dividend; trailing yield is zero. (FACT.) The Series B preferred was retired in 2021 for $43.3M. Given the capital programme, retaining everything is correct.

How profitable is the business? On GAAP, apparently very: 17.3% net margin and $3.90 of diluted EPS in FY2025. On operations, much less: an 11.2% operating margin, and roughly 41% of cumulative five-year pretax income was non-operating, rising to over 100% in Q1-2026. Normalized operating EPS is approximately $1.06–1.94 against $3.90 reported. (FACT/INTERPRETATION.)

Is net income diverging from cash from operations? Yes, materially and persistently. Cumulative five-year OCF of $167.7M against net income of $462.6M is a 36% conversion rate. (FACT.) The drivers are the reversing customer-advance float and working capital. This is one of the clearest quality-of-earnings signals in the file.

Risks & Downside

What factors would cause the stock to decline? In rough order of probability-weighted impact: (1) the 2028 convergence arriving without a disclosed bridge — TENEX contract end, waiver-authority termination, Russian export-decree expiry and RSA quota expiry all landing before the 2029 first cascade; (2) an equity raise at depressed prices to fund capex; (3) first-cascade slippage past 2029; (4) SWU price mean-reversion as Urenco’s and Orano’s capacity lands; (5) cost overruns on a fixed-price contract with time-and-materials subcontracting; (6) HALEU demand slipping with its pre-revenue customers; (7) normalisation of the tax rate and loss of interest income as cash is deployed; (8) continued thematic de-rating in a name where five of six factor-nearest peers are uranium ETFs.

Risk of a catastrophic loss? Real, and empirically demonstrated. (FACT.) The predecessor entity, USEC, filed Chapter 11 in 2014 and wiped out the equity; the lifetime maximum drawdown on this security is −99.98%. Shareholders’ equity was negative for eight consecutive years (2015–2022). The current balance sheet is far stronger — $689.6M of net cash — but it is a borrowed strength: $1,207.5M of convertible principal settles in cash at maturity in 2030 and 2032, straddling the plant startup. A scenario in which the arbitrage expires, the build overruns, and the notes come due is not exotic.

Chance of a total loss? Low over the next three to five years, given the net cash position and the strategic importance of the licence — a distressed Centrus would likely be acquired or recapitalised rather than liquidated, because the NRC licence and the US-origin technology have national-security value independent of the equity. (INTERPRETATION.) But “low” is not “negligible,” and the bear scenario in the memo implies an 80–90% loss of enterprise value without requiring bankruptcy at all.

Recent News & Events

Has the business environment changed recently? Substantially, and in both directions. (FACT.) Positive: DOE granted a waiver on 2025-08-04 covering all committed US deliveries in 2026 and 2027, materially de-risking the near term; NNSA notified an intent to sole-source certain enrichment activities; the definitive $900M firm-fixed-price DOE contract was signed 2026-06-30 with title to the capacity passing on completion; and the company raised ~$1.33B of converts plus ~$533.6M of ATM equity at excellent prices. Negative: Russian Decree No. 1516 extends TENEX’s per-shipment licensing regime only through 2027-12-31; Urenco USA announced a further +2.1M SWU expansion on 2026-06-02; DOE funded two direct competitors at $900M each; Q1-2026 operating income fell 96%; the stock was deleted from Russell indices and its S&P SmallCap 600 addition was sold into on 3.5x volume.

Significant acquisitions? None. The relevant transactions are contractual: the DOE task order, the Geiger Brothers construction contract (2026-04-16), Fluor as EPC, and the Palantir Foundry arrangement (with $5M prepaid).

Change in accounting policies? No change in accounting policy, but a change in presentation: “adjusted net income” and “adjusted EPS” were introduced in Q1-2026, and trailing-twelve-month presentation was introduced in Q4-2025. Both were introduced as GAAP results deteriorated. (FACT.)

Recent changes — new markets, facilities, management? Facilities: the Piketon expansion (announced 2025-09-25, “multi-billion-dollar,” contingent on federal funding) and a $560M+ Oak Ridge centrifuge manufacturing plant (announced 2026-01-23), with the first new machines reaching Ohio in 2029. Markets: a non-binding Oklo LoI (2026-06-18, deliveries from 2029, no volume or price), an exploratory Oklo deconversion JV, and a KHNP/POSCO International MOU — none of which carries contractual weight. Management: CEO Amir Vexler (successor to Daniel Poneman); CFO transition from Kevin Harrill to Kevin Tinelli in mid-2025. Listing: uplisted from NYSE American to NYSE in December 2025.

APPENDIX B — Source Appendix

Centrus Energy Corp. (NYSE: LEU) — 2026-07-18

All sources accessed 2026-07-17 / 2026-07-18 unless otherwise stated. SEC filings retrieved from EDGAR under CIK 0001065059.

1. SEC Filings — Primary

The trailing 60-month corpus was enumerated and downloaded: 313 filings since 2021-07-01, comprising 107 Form 4, 88 Form 8-K, 15 Form 10-Q, 5 Form 10-K, 12 Form 3, 18 Schedule 13D/G, 13 Schedule 13G, 6 DEFA14A, 5 DEF 14A, 4 424B5, 1 S-3ASR, 1 S-3, 1 S-8, plus ARS and administrative filings. Structured-note noise (424B*, FWP, 144) was excluded from analysis.

EDGAR XBRL company facts — https://data.sec.gov/api/xbrl/companyfacts/CIK0001065059.json and companyconcept endpoints — used to verify share counts, revenue and EPS tags against the aggregated financial data.

2. Earnings-Call Transcripts

Retrieved via the ROIC.ai transcript service and read in full. All management quotations in the memo are verbatim from these bodies and are labelled as management hypothesis, not evidence (per the framework’s rule that management commentary requires external validation).

3. Market, Pricing and Factor Data

4. News, Analyst and Industry Sources

5. Analytical Frameworks Applied

• Competition Demystified (Greenwald & Kahn) — barriers-to-entry taxonomy applied to the competitive position: the supply/cost, demand-captivity and scale-plus-captivity tests, the market-share-stability test, and the ROIC test (the last found inapplicable given a ~$85M net-invested-capital denominator).
• Capital Returns (Marathon Asset Management) — supply-side capital-cycle analysis applied to the industry: high prices plus subsidy attracting capital, the announced 6–8M SWU of Western additions landing 2029–2036, and the asset-growth caution applied to Centrus’s own build.

6. Data-Reliability Notes

1. Enterprise value was rebuilt by hand at the live price. Third-party enterprise values and market capitalizations for LEU are computed off stale fiscal-year-end price snapshots and materially misstate the current figure. The EV of $2,380.2M used throughout is computed from the Q1-2026 10-Q balance sheet at the 2026-07-17 close.
2. The P/B own-history percentile is not usable. Shareholders’ equity was negative for eight consecutive years (2015–2022), so the historical distribution against which today’s 4.52x is ranked is meaningless. Only the P/E and P/S percentiles are load-bearing.
3. Conventional ROIC is not reported. Invested capital net of cash is approximately $85M, producing values between 3% and 117% across recent years. Any ROIC figure for this company should be treated as an artifact.
4. FY2021 EPS discrepancy. Third-party data reports diluted EPS of $12.46 for FY2021; this is the company’s non-GAAP figure adding back $37.6M of distributed earnings allocable to retired preferred shares. GAAP diluted EPS was $9.75, itself inflated by a $39.1M deferred-tax valuation-allowance release.
5. TTM EPS. Third-party sources showing $3.02 do not reconcile to the filings; the correct trailing figure is approximately $2.75.
6. Management commentary is treated as hypothesis throughout, and is validated against filings and financial data wherever a claim is load-bearing. Points where management framing conflicts with the filings are flagged explicitly above (backlog quality, adjusted EPS, cost savings, margin trajectory).
7. Derived estimates are labelled. The ~$50/SWU steady-state EBITDA benchmark (from Urenco’s audited FY2025), the 41.0 SWU/kg HALEU separative-work conversion (from the standard separative value function at 19.75% product / 0.25% tails), and the $1,835/SWU capital-intensity figure (derived from the $900M DOE contract against 12 MT/yr of HALEU capacity) are analytical derivations, not company or third-party disclosures.