Lockheed Martin LMT

Reference table

1. What the business is, and the disclosure that carries the thesis

Chairman, president and chief executive James D. Taiclet Jr.; approximately 123,000 employees. Four reportable segments, and the vendor's segment table is one of the cleanest in this batch — it sums to $75,048M against reported revenue of $75,057M, a $9M difference.

Missiles and Fire Control is the growth engine and it is not close — up 28.4% in two years against 10.1% for Aeronautics and 3.4% for Space. That is munitions replenishment and air-and-missile defence, the two categories every allied government is buying.

Geography, and this is where the rearmament shows up:

International revenue rose from 26.1% to 28.4% of the total in a single year, and it did so while US revenue grew 2.2%. Europe and Asia Pacific together added $2,663M of the $4,005M total increase — two-thirds of Lockheed's FY2025 growth came from outside the United States.

The backlog disclosure

From the 10-Q filed 2026-07-23:

> "As of June 28, 2026, our ending backlog was $230.4 billion. The increase in backlog of $36.8 billion during the six months ended June 28, 2026 was primarily due to an undefinitized contractual action (UCA) awarded for the THAAD program at our MFC business segment. We expect to recognize approximately 30% of our backlog over the next 12 months and a total of approximately 50% over the next 24 months as revenue with the remainder recognized thereafter."

$230.4 billion is 3.07x trailing revenue of $77.0 billion. Thirty percent within twelve months is roughly $69 billion, against consensus FY2026 revenue of $80.8 billion — so the backlog covers roughly 86% of the next year's expected sales before a single new order is booked.

One caveat the company itself supplies, and it matters: "For our cost-reimbursable and fixed-priced-incentive contracts, the estimated consideration we expect to receive... may exceed the contractual award amount... As the risks on such contracts are successfully retired, the estimated consideration from customers may be reduced, resulting in a reduction of backlog without a corresponding recognition of sales." Backlog can fall without revenue rising. And an undefinitized contractual action — the source of $36.8 billion of the increase — is by definition a contract whose terms are not yet agreed. The number is real; its precision is not.

The F-35 franchise, from the same filing: 19 aircraft delivered in the June quarter; 1,344 production aircraft delivered since programme inception (956 F-35A, 250 F-35B, 138 F-35C); backlog of 317 aircraft at 2026-06-28.

2. The execution problem, in the company's own numbers

This is the section that determines the verdict, and every figure in it is quoted from the 10-Q.

Aeronautics classified programme:

> "We have experienced significant performance issues on an existing classified program at our Aeronautics business segment. The initial phase is on a fixed-price incentive fee contract with fixed-price incentive fee options for additional phases... during the second quarter of 2025, we recognized additional reach-forward losses of $950 million across both phases of the program."

The drivers the company lists: "(1) software development performance degradation and integration findings observed over a continued period; (2) learnings in recent software and build experience on another program... (3) significant changes in test plan...; (4) safety-critical and other necessary design and engineering changes in response to certain observed performance degradation and a discrete event; and (5) complete schedule realignment."

> "As of June 28, 2026, cumulative losses recognized to date on this program remained at approximately $1.8 billion across both phases. As of June 28, 2026, $427 million of the losses remained accrued... we may need to record additional losses in future periods."

Missiles and Fire Control classified programme:

> "...a competitively bid classified contract, which includes a cost-reimbursable base contract for the initial phase and multiple fixed-price options for additional phases. We previously disclosed that the options may be exercised over the next several years and, if performed, we expect they would each be at a loss. Certain options have been exercised, and... we continue to believe it is probable that the remaining unexercised options will be exercised. As of June 28, 2026, cumulative losses recognized on the program remained at approximately $1.5 billion in total."

Read that carefully. The company expects future contract options to be exercised, and expects each of them to be at a loss, and regards their exercise as probable. That is a disclosed, quantified, forward-looking commitment to lose money — and it is not a hypothetical risk factor.

Canadian Maritime Helicopter Program: the final aircraft was delivered in 2025; the programme "has experienced performance issues and we have previously recorded losses" including $570 million in the June 2025 quarter; restructuring discussions "may be prolonged or unsuccessful, and could result in a contract termination."

United Launch Alliance: the 50/50 joint venture with Boeing. The 10-Q warns that Lockheed and Boeing "expect to provide additional financial support to ULA to support its liquidity or ongoing operations and could incur impairment and operating losses if the Vulcan Centaur rocket does not perform consistent with ULA's assumptions."

Securities litigation:

> "On July 28, 2025, a putative class action was filed—and subsequently amended on January 12, 2026—in United States District Court for the Southern District of New York against us and certain current and former members of our senior management... on behalf of persons and entities that purchased or otherwise acquired our securities between January 23, 2024 and July 21, 2025. Plaintiffs seek unspecified losses allegedly caused by alleged misstatements about certain classified programs in the Aeronautics and MFC business segments and F-35 program, which were allegedly revealed to be false when we announced estimated losses relating to certain of those programs."

What this does to the reported numbers, quarter by quarter:

Two of the last eight quarters had operating margins of 3.7% and 4.1% against a normal 11.5-13.2%. Those are not cyclical fluctuations; they are single-programme charges landing without warning. A business that can lose two-thirds of a quarter's operating income to a disclosure it makes on the day is not a business that supports a premium multiple, however good the backlog is.

The June 2026 quarter is clean, and that is the bull case in one line: revenue $20,063M (+10.5%), operating income $2,479M (+231%), net income $1,836M, EPS $7.94 against a $7.22 estimate — a 10.0% beat. The 10-Q attributes the product-sales increase of $1.6 billion to higher F-35 production volume plus "the sales impact of the 2025 reach-forward loss on a classified program" at Aeronautics, PAC-3 and THAAD ramp-up at MFC (+$635M), undersea combat systems and River Class Destroyer at RMS, and Fleet Ballistic Missile and Next Generation Interceptor at Space. Note that part of the year-on-year improvement is the absence of last year's charge, not incremental performance.

3. The acquisition the vendor payload does not know about

From Note 12, Subsequent Events, in the 10-Q filed 2026-07-23:

> "On July 6, 2026, we announced that we entered into a definitive agreement to acquire Ultra Maritime Solutions (Ultra Maritime), a global defense company specializing in advanced undersea warfare and anti-submarine capabilities for allied naval forces, for $3.45 billion. We expect to fund the acquisition with cash on hand and additional financing arrangements. The transaction is subject to regulatory reviews and approvals and customary closing conditions, and is expected to close in the fourth quarter of 2026."

Nothing in the vendor payload reflects it. bal_a ends at 2025-12-31, cf_a ends at FY2025 with acquisitionsNet of $0 for each of the last four years, and there is no FY2026 row anywhere.

Three consequences. First, $3.45 billion against $4.1 billion of year-end cash and $6.7 billion of equity is a large transaction for this balance sheet, and "additional financing arrangements" means net debt rises from $17.6 billion. Second, the strategic logic is coherent — undersea warfare and anti-submarine capability sits directly alongside the Rotary and Mission Systems undersea combat systems business the 10-Q cites as a June-quarter growth driver. Third, it is the first meaningful acquisition in this file's four-year cash-flow history, in which acquisitionsNet reads zero every year, and it therefore represents a genuine change in capital allocation from a company that has spent that period buying back stock.

4. Financials, and the balance sheet nobody looks at

The annual series:

Read the operating-margin column and the damage is obvious: 13.2% in FY2020, 10.3% in FY2025. Revenue rose 14.8% across those five years while operating income fell 10.6%. Net income in FY2025 was lower than in every year shown, and GAAP EPS of $21.56 was below FY2020's $24.40 — despite a share count 17.5% smaller. Trailing margins have recovered to 11.9% as the FY2025 charges roll out of the comparison, and that recovery is the entire earnings-growth story consensus is underwriting.

Note Lockheed's income-statement presentation and do not misread it. grossProfitMarginTTM reads 11.80% against operatingProfitMarginTTM of 11.88% — the two are essentially identical because Lockheed's "cost of sales" line absorbs most general and administrative expense. This is a presentation convention, not a defect, and it means gross margin carries no information for this company. Research and development at 0.86% of revenue is likewise not comparable to a commercial company's, because most development work is customer-funded and sits inside cost of sales.

The balance sheet at 2025-12-31 (the most recent in the payload):

Equity of $6,721M supports $59,840M of assets — an 11.2% equity ratio and a debt-to-equity ratio of 2.34x. Tangible book value is NEGATIVE $18.75 per share, correctly reported by the vendor as distinct from book value of $38.09, because $15,618M of goodwill and intangibles exceeds $6,721M of equity by $8.9 billion.

This is a consequence of capital return, not distress, and the distinction should be stated clearly. Lockheed has repurchased $7,900M (FY2022), $6,000M (FY2023), $3,700M (FY2024) and $3,000M (FY2025) of stock — $20.6 billion over four years against a current market capitalisation of $136 billion — while paying $3.0-3.1 billion of dividends annually. Equity is a residual that buybacks have consumed. The business generates $8.8 billion of trailing free cash flow and covers interest 8.2x; it is not fragile.

But it does mean two things. priceToBookRatioTTM of 15.47x and returnOnEquityTTM of 86.2% are arithmetically correct and economically meaningless, and we reject both as lenses (returnOnInvestedCapitalTTM of 20.1% is the honest return measure). And there is no equity cushion against a large adverse event — a further multi-billion reach-forward loss, an adverse litigation outcome, or a ULA impairment would come straight out of a $6.7 billion equity base while a $3.45 billion acquisition is being financed.

Cash flow and capital returns:

Lockheed returned more than 100% of free cash flow in each of FY2022, FY2023 and FY2024, funding the excess with debt — total debt rose from roughly $12 billion to $21.7 billion across the period. FY2025 was the first year of restraint at 89%, and the buyback has fallen from $7.9 billion to $3.0 billion. Trailing free cash flow is $8,752M (a freeCashFlowPerShareTTM of $37.92), the highest in the file.

Total shareholder yield is approximately 4.5% — a 2.32% dividend at a 50.5% payout, plus roughly 2.2% of buyback. That is the second-highest yield in this batch and a genuine component of return. It is also precisely what the Treasury Secretary's claim in Section 7 proposes to curb.

5. Valuation — priced in or room?

At $589.13 (market cap $135.97B, enterprise value approximately $152.7B):

Estimate coverage is the best in this batch — 13 to 17 analysts on every year through FY2029 — and the FY2027 range is tight at $31.600 to $34.338, an 8.7% spread. The FY2029 revenue line of $91.32B against FY2028's $90.37B implies +1.0% growth, a sharp deceleration that is either a genuine view on the appropriations cycle or a coverage artefact; either way it caps the multiple that can be justified on the outer years.

est.ebitdaAvg and est.ebitAvg are rejected. In every forward year ebitdaAvg is exactly 13.107% of revenueAvg and ebitAvg is exactly 11.222%, to four decimals — the fixed-ratio fabrication signature. In Lockheed's case the ratios happen to sit close to reality (trailing operating margin is 11.9%), which makes them more dangerous, not less: a mechanically generated number that looks plausible is harder to catch. All forward valuation uses epsAvg and revenueAvg.

Peer context. The vendor peer set — ADP, Boeing, Deere, General Dynamics, Honeywell, L3Harris, Northrop Grumman, Parker Hannifin, TransDigm, Trane — contains four genuine defence comparables (GD, LHX, NOC, and BA in part) and six industrials that are not. No peer multiples are supplied in the file, so no peer-multiple comparison is drawn.

5a. What today's price assumes (the inversion)

At $589.13 — 19.4x FY2026 consensus and 18.1x FY2027 — the price embeds:

• Revenue reaches $80.81B in 2026 and $85.42B in 2027. (Consensus; 14 and 16 analysts.) Backlog of $230.4 billion with ~30% recognised within twelve months implies roughly $69 billion of the FY2026 figure is already contracted, so this is the least fragile assumption in the price.
• EPS reaches $32.63 by 2027 — 19.6% above the trailing $27.31 — driven by margin recovery, not volume. (Consensus; our decomposition.) Revenue grows 13.5% across the same span, so roughly a third of the EPS growth must come from operating margin moving back toward 12.5-13%.
• No further reach-forward loss of consequence is recognised. (Our number, from the 10-Q's disclosures.) This is the most fragile assumption in the price. Two classified programmes carry ~$3.3 billion of cumulative losses, $427 million remains accrued on one, the company states it "may need to record additional losses", and on the other it expects future option exercises to be "at a loss" and regards them as probable. A repeat of the June 2025 quarter is roughly $1.6 billion pretax — about 22% of a year's operating income — and it would arrive without warning on the day of a print.
• The dividend and buyback continue at approximately $6.1 billion a year. (Our number, from FY2025 actuals.) That is 4.5% of the return. The sitting US Treasury Secretary is on record proposing to curb exactly this (Section 7).
• The market keeps paying 18x or better on the forward year. (Our number.) At 15x FY2027E the stock is $489; at 21x it is $685. Lockheed has traded across that whole band within the last two years — the 52-week range of $423.70 to $676.70 is 60% wide on a 0.113-beta stock.

5b. The return bridge (why the multiple moves)

Expected return over the next twelve months decomposes as: EPS growth (+7.5%, from FY2026E $30.363 to FY2027E $32.632) + multiple drift (COMPRESSION, from 19.4x on the forward year to roughly 18x on the then-forward year, about −7%) + shareholder yield (+4.5%) ≈ +3% to +7%.

Our base assumes mild compression, and the reason is the outer years rather than the near ones. Consensus revenue growth runs +7.7%, +5.7%, +5.8%, then +1.0% in FY2029. A multiple appropriate to 6% growth is not appropriate to 1% growth, and the market normally starts discounting a deceleration a year ahead of it. We are also unwilling to underwrite multiple expansion into an RSI of 81 after a 40% run.

Almost all of the expected return is earnings growth plus the dividend, and it is not enough to cover the starting multiple. If the multiple simply HELD at 19.4x on FY2027E the price would be $633 (+7.4%) — which is the street's target almost exactly. If it compressed to 16x the price would be $522 (−11.4%). The street's $634.38 is, arithmetically, "today's multiple, one year forward" — a no-re-rating target. Our $585 differs from it only in assuming a 7% de-rating, and that single assumption is the entire disagreement.

5c. Variant perception (where we differ, what would surprise)

• We differ from the street by 7.8%, downward, and the disagreement is one variable: the multiple. The street's $634.38 holds today's forward multiple constant; we compress it 7% on the FY2029 deceleration and on positioning. On the earnings we do not differ at all. That is a thin basis for a differentiated view and we say so.
• We think the reach-forward loss risk is systematically under-priced because it is unforecastable. No analyst model contains a $950 million charge; the consensus FY2027 EPS range of $31.60 to $34.34 spans 8.7%, which is far narrower than the distribution the last eight quarters actually produced. Two of those eight quarters had operating margins of 3.7% and 4.1%. Watchable event: any reach-forward loss disclosure in the 2026-10-20 print. The company has told us where to look — $427 million accrued on the Aeronautics programme, probable loss-making option exercises on the MFC programme, and a Canadian Maritime Helicopter restructuring that "could result in a contract termination".
• We think the policy risk to the capital return is genuinely novel and not in any model. The knowledge base carries a claim from Scott Bessent, role policymaker, dated 2026-01-20, that defence primes "have failed, 5-7 years behind on deliveries while buying back stock instead of building facilities" and that it is "a good idea to bring them to heel by curbing dividends/buybacks." Lockheed returned 89% of free cash flow that way in FY2025 and more than 100% in each of the three years before. This is not a forecast — it is one policymaker's stated view — but it is a specific, sourced threat to 4.5 percentage points of expected return, and no sell-side model carries it.
• Positive surprise that would force a re-rate: a clean third quarter with operating margin above 12.5% and no new charge, which would begin to establish that the classified programmes are contained; a definitised THAAD contract converting the undefinitized action into firm revenue at a disclosed margin; or an F-35 multi-year procurement award at scale.
• Negative surprise that would break the thesis: a new reach-forward loss of $500 million or more; an adverse development in the securities class action; a ULA impairment; the Ultra Maritime financing arriving at a cost that pushes leverage materially higher; or an F-35 procurement reduction in a defence budget. Any of these hits a stock at RSI 81 that has risen 40% in twelve months.

Synthos fair values

All three anchors are multiples of the FY2027 consensus EPS distribution (mean $32.632, low $31.600, high $34.338, 16 analysts), cross-checked against FY2026E of $30.363 and FY2028E of $34.994.

• Bear ~$442 — 14.0x the FY2027 consensus LOW of $31.600. Cross-check: 14.6x FY2026E; 4.3% above the 52-week low of $423.70. The scenario: a new reach-forward loss, a margin recovery that does not arrive, the FY2029 deceleration pulled forward, and a de-rating to the multiple Lockheed carried before the rearmament narrative. −25.0%.
• Base ~$585 — 17.9x the FY2027 consensus MEAN of $32.632. Cross-check: 19.3x FY2026E; 16.7x FY2028E; enterprise value of $152.0B on FY2027E revenue of $85.42B is 1.78x sales. Sensitivity, stated openly: 17x gives $555 — 5.8% BELOW spot — and 19x gives $620, 5.2% above. The whole answer lives in a 17-19x band and the market at 18.1x is already inside it. The scenario: backlog converts on schedule, margin recovers modestly, no new charge lands, and the multiple drifts down with the outer-year growth rate. −0.7%.
• Bull ~$721 — 21.0x the FY2027 consensus HIGH of $34.338. Cross-check: 20.6x FY2028E; 6.5% above the 52-week high of $676.70; 13.7% above the street's high target of $700. The scenario: the classified programmes are contained, margin returns to 13%, the international mix keeps compounding at double digits, and the market pays a rearmament multiple. +22.4%.

Base is 0.7% BELOW spot; asymmetry is roughly 0.90:1 — unfavourable. This is the only name in this batch where our base fair value sits below the closing price, and that is the verdict. A 25% downside against a 22% upside, with a base case at par, is not a position — it is a watchlist entry with a price on it.

6. Knowledge base — seventeen hits, nine entity claims, and the bears are newer

Raw hits: 17. Entity matches: 9. Text-only: 8. Used as name-level conviction: 7. Excluded as suspected management voice: 2. Discarded as homographs: 0.

The sweep ran entity terms LMT, Lockheed and Lockheed Martin, plus free text on Lockheed, F-35, defense primes and defence primes, across all 51,928 distilled claims. "Lockheed" is a clean token and no homograph collision occurred — a contrast with APP and COF elsewhere in this batch.

Channel distribution: lex_fridman 4, natalie_brunell 2, mikkel_rosenvold 2, luke_gromen 2, and eight channels with one each. The maximum single-channel share is 23.5%, below the one-third threshold, so no concentration sensitivity test is triggered. Stated because the test is mandatory when the threshold is crossed and its absence should be explicit.

The entity-matched claims — five bullish, three bearish, one neutral

> 2026-03-16 · BULLISH · conviction 75 · horizon: FACT · entity: Lockheed Martin · channel: odd_lots · no named speaker

> "Lockheed ramping PAC-3 Patriot from ~600 to 2000/year and Tomahawk toward 1000/year (from 57 requested) — large announced production increases."

> The best claim in the lane — a fact-horizon, quantified capacity assertion, and the 10-Q corroborates the direction independently, citing "production ramp-up on Patriot Advanced Capability-3 (PAC-3) and Terminal High Altitude Area Defense (THAAD) programs" adding $635 million of MFC product sales in the June quarter.

> 2026-07-27 · bullish · conviction 55 · entities: CAT, LMT · speaker: Dan Ives · channel: anthony_pompliano

> "AI's second-third-fourth derivatives extend to names like Caterpillar and, on the defense side, Lockheed Martin as defense-tech beneficiaries of the AI buildout."

> A basket claim, named speaker, low conviction.

> 2026-03-02 · bullish · conviction 40 · entity: LMT · speaker: Mikkel Rosenvold · role: independent

> "UAE's successful drone/missile intercepts mean a lot of restocking at Lockheed Martin, if you're into those kinds of stocks."

> Conviction 40 is the lowest band we record, and the phrasing ("if you're into those kinds of stocks") signals the speaker's own detachment.

> 2026-03-20 · BEARISH · conviction 68 · entities: Lockheed Martin, Northrop Grumman · channel: a16z · no named speaker

> "Post-Last-Supper consolidation bred conformity and financialization (dividends/buybacks over growth); primes can't harbor founder innovation."

> This claim is directly corroborated by the filings: $20.6 billion of buybacks over four years, capital returns exceeding 100% of free cash flow for three consecutive years, and negative tangible book value.

> 2025-08-24 · BEARISH · conviction 80 · entity: Lockheed · channel: lex_fridman · no named speaker

> "Defense contractors extract private dividends from wars that deliver no net benefit to society; military-industrial complex is a value-destroying rent machine."

> An ethical rather than financial claim, and we weight it accordingly — it is a statement about social value, not about return on capital. It is recorded because suppressing it would be selective.

> 2024-10-23 · BEARISH · conviction 60 · entities: Lockheed Martin, General Dynamics · channel: lex_fridman · no named speaker

> "Trillion-dollar Pentagon budget with revolving-door contractors and un-auditable spending implies massive fraud, waste and cost overruns."

> 2025-06-28 · NEUTRAL · conviction 60 · entities: Raytheon, Lockheed Martin · channel: natalie_brunell · no named speaker

> "Defense primes face demand to replenish depleted missile stocks but can't produce rocket motors domestically — a manufacturing/supply-chain constraint."

> The most operationally useful bearish-leaning claim, and consistent with the 10-Q's acknowledgement that "there can be no assurance that production capacity will increase at the rate or to the extent we or our customers expect due to supply chain constraints, workforce limitations, government funding decisions."

Two claims excluded as suspected management voice

> 2019-08-19 · conviction 85 · entities: Lockheed Martin, LMT · channel: lex_fridman · no named speaker

> "Conviction that AI-enabled autonomy will be retrofit into existing systems and designed into every future Lockheed Martin system."

> 2019-08-19 · conviction 70 · entity: Lockheed Martin · channel: lex_fridman · no named speaker

> "Sustained presence beyond low-earth orbit and a real space economy hinge on driving launch cost down; deep-space human missions (Orion) advancing."

Both are dated 2019-08-19 — seven years old — carry no named speaker, and read unmistakably as a Lockheed executive describing his own company's roadmap ("every future Lockheed Martin system", Orion programme detail) on a host-named interview channel. This is precisely the failure mode the 4-lane attribution policy exists to catch: a channel slug named for the interviewer carrying the interviewee's — here, management's — voice. Both are EXCLUDED from the conviction pool. They are also the two highest-conviction bullish claims in the entity set, and removing them changes the lane's balance materially: from 5 bullish / 3 bearish / 1 neutral to 3 bullish / 3 bearish / 1 neutral. That reweighting is the reason net conviction is recorded as mixed-slightly-negative rather than positive.

Two text-only claims that matter more than most of the entity set

Neither names Lockheed in its entity list, so neither enters the conviction pool. Both are more decision-relevant than several claims that do, and suppressing them for a technicality would be dishonest.

> 2026-01-20 · conviction 66 · horizon: thesis · speaker: Scott Bessent · role: POLICYMAKER · channel: scott_bessent

> "Defense primes have failed, 5-7 years behind on deliveries while buying back stock instead of building facilities; good idea to bring them to heel by curbing dividends/buybacks."

> This is the sitting United States Treasury Secretary, and the speaker_role field records it as policymaker rather than management or independent. Lockheed returned 89% of free cash flow via dividends and buybacks in FY2025 and more than 100% in each of the three preceding years. A policy move against that mechanism would remove roughly 4.5 percentage points of expected annual return. It is one person's stated view, not a policy — but it is a sourced, dated, attributable statement from the official with the most influence over it.

> 2026-07-30 · conviction 82 · horizon: PRINCIPLE · speaker: jim_bianco · role: independent · skill 0.8 — the MOST RECENT and HIGHEST-CONVICTION claim in the entire sweep, dated five days before this dive

> "21st-century war is cheap, mass, iterative unmanned drones beating expensive legacy systems; carriers, F-35s, cruise missiles increasingly only create war crimes, not decisive results."

> This is the structural bear case on Lockheed's core product, stated at the highest conviction in the lane by a named, independent speaker, five days ago. It names the F-35 explicitly. We cannot adjudicate it and we do not try — but a company deriving 40.3% of revenue from Aeronautics, with 317 F-35s in backlog, faces a claim that its principal product category is being obsoleted, and the knowledge base's most recent word on the subject is that one.

Two further text-only claims for completeness: no_priors 2026-01-15 (conviction 80) records that new-entrant primes (SpaceX, Anduril, Palantir) "once had to sue for contracts" and are now being actively courted — a competitive-displacement claim; and luke_gromen 2026-01-27 (two claims, conviction 63-65) puts a 60-70% base case on the United States being unable to produce weapons such as the F-35 within seven to eight months because of critical-mineral dependence on China. That is an extraordinarily specific and alarming assertion which we could not corroborate in any Lockheed filing, and we record it as an unverified third-party claim rather than as evidence.

Conclusion. Breadth 11, claim count 9 (7 after the management-voice exclusions), net conviction mixed, slightly negative. This is the most genuinely two-sided knowledge-base lane in this batch, and the balance tilts negative on recency: the three most recent substantive claims (2026-03-20 bearish 68, 2026-07-27 bullish 55, 2026-07-30 bearish 82) are two bearish to one bullish, and the highest-conviction claim in the whole sweep is a structural argument against the product. That is a direct input to the Watch verdict.

7. Data integrity — what we rejected and why

Eight findings. Lockheed's file is unusually good on segments and geography and unusually bad on depreciation, with one entire transaction missing.

1. The $3.45 billion Ultra Maritime Solutions acquisition is absent from every vendor field. Announced 2026-07-06, expected to close in the fourth quarter of 2026, to be funded with "cash on hand and additional financing arrangements." bal_a ends at 2025-12-31, cf_a ends at FY2025, and acquisitionsNet reads exactly $0 in each of the four years shown. The transaction is material against $4.1 billion of year-end cash and $6.7 billion of equity. Supplied from Note 12 of the 10-Q.

2. inc_a depreciation and amortisation for FY2025 contradicts the cash-flow statement by a factor of 2.4 — REJECTED. The income-statement block reports FY2025 depreciationAndAmortization of $714M; the cash-flow block for the same year reports $1,687M. FY2024 is internally consistent ($1,560M against $1,559M), so the defect is confined to FY2025. Consequence: inc_a.ebitda for FY2025 reads $7,754M against operating income of $7,731M — implying D&A of $23M for a company with $59.8 billion of assets. We use the cash-flow figure.

3. Trailing EBITDA is internally inconsistent — recomputed. ratios_ttm.ebitdaMarginTTM of 12.796% implies trailing EBITDA of $9,856M against trailing operating income of $9,153M — D&A of $703M. But the vendor's own capexToDepreciationTTM of 0.99053 against trailing capex of $1,673M implies $1,689M, consistent with the FY2025 cash-flow statement. We compute EBITDA as $9,153M + $1,689M = $10,842M (14.1% margin), moving EV/EBITDA from the vendor's implied 15.5x to ~14.1x and net-debt/EBITDA from 1.699x to ~1.62x.

4. est.ebitdaAvg and est.ebitAvg carry the fixed-ratio fabrication signature — REJECTED, and the case is instructive. In every forward year ebitdaAvg is exactly 13.107% of revenueAvg and ebitAvg is exactly 11.222%. Unusually, those ratios sit close to Lockheed's actual margins (trailing operating margin 11.9%), which makes the fabrication harder to spot and no less disqualifying — a mechanically generated number that happens to look plausible is more dangerous than one that does not. All forward valuation uses epsAvg and revenueAvg.

5. epsdiluted is null in every row of inc_a and inc_q — worked around. The vendor supplies eps and weightedAverageShsOutDil but no diluted EPS. All per-share figures here are computed and labelled, and cross-checked against the earn_cal actuals where available.

6. quote.price and tech.last disagree, and quote.yearHigh/yearLow disagree with tech.hi52/lo52 — we use tech for the technical series. The quote block prints $589.13 and the technical block $589.33, a $0.20 difference; the quote block reports a 52-week high of $692.00 and low of $421.86 against the technical block's $676.70 and $423.70, discrepancies of 2.3% and 0.4%. We use quote.price of $589.13 as the strike price (it is the field the dive is struck against per the data contract) and tech for all range, drawdown and relative-return figures, and state the mixing explicitly.

7. Gross margin carries no information for this company — flagged, not rejected. grossProfitMarginTTM of 11.80% sits essentially on top of operatingProfitMarginTTM of 11.88%, because Lockheed's cost-of-sales line absorbs most general and administrative expense. Likewise researchAndDevelopementToRevenueTTM of 0.86% materially understates development activity, most of which is customer-funded and recorded in cost of sales. Neither is a defect; both would mislead a cross-sector screen.

8. Book-equity metrics are economically meaningless — REJECTED as lenses. priceToBookRatioTTM and priceToFairValueTTM (both 15.47x), returnOnEquityTTM (86.2%) and debtToEquityRatioTTM (2.34x) are arithmetically correct and all driven by the same thing: $20.6 billion of buybacks over four years against $6,721M of remaining equity. returnOnInvestedCapitalTTM of 20.1% is the honest return measure and is what we use. The vendor's own composite rating scores Lockheed 1 out of 5 on both price-to-book and debt-to-equity, dragging the overall grade to B+ / 3 — a buyback artefact, not a judgement about the business, and rejected on the same mechanism documented elsewhere in this batch.

Correct and worth recording — Lockheed's file is clean where it matters most: seg_prod sums to $75,048M against reported revenue of $75,057M, a $9M difference, with all four segments present and correctly named across four years — the cleanest segment table in this batch; seg_geo sums to the same figure with five regions and no omission (contrast DHR, where 40.6% of revenue is missing from the geographic block); bookValuePerShareTTM of $38.089 is correctly distinct from tangibleBookValuePerShareTTM of −$18.753 — a genuinely negative tangible book properly reported; the share count (8-K record-date count of 230,462,118 against 230.8M implied by market capitalisation ÷ price); dividendPerShareTTM of $13.65 and dividendPayoutRatioTTM of 50.5%; and interestCoverageRatioTTM of 8.24x.

Filing-archive limitation, stated plainly. The LMT archive is flagged tables: false in the manifest — the filings were extracted as prose without preserved statement tables. In practice the 10-Q's narrative disclosures carried everything this dive needed (backlog, reach-forward losses, F-35 deliveries, litigation, the acquisition), but figures appear on separate lines from their labels and no primary income statement, balance sheet or cash-flow table was available for direct reconciliation. The three 8-K filings in the archive are cover documents only — Exhibit 99.1, carrying the earnings releases and any guidance, was not extracted, so no management guidance figures appear anywhere in this dive. All financial-statement figures therefore come from the vendor payload, cross-checked against the earn_cal actuals and against narrative amounts in the 10-Q. This is stated so no reader mistakes the annual and quarterly statements here for filing-verified tables.

Non-equity tripwire — checked and passed. LMT is common stock, NYSE-listed. Price of $589.13 is not par-like; the dividend is variable and has been raised over many years; volume was 0.95M shares (~$559M of turnover); the 52-week band of $423.70 to $676.70 is a 60% range. Beta of 0.113 is unusually low and warranted the check — a sub-0.3 beta is characteristic of fixed-income-like instruments — but the 60% annual range, the variable dividend and the equity capital structure settle it. This is common equity with a genuinely low market correlation.

8. Technicals

• Price $589.13. −12.9% from the 52-week high of $676.70; +39.1% above the 52-week low of $423.70. Position within the annual range: 65th percentile.
• Above both moving averages: +10.9% above a 50-day average of $531.29; +7.9% above a 200-day average of $545.94, with the 50-day below but rising sharply toward the 200-day.
• RSI 81.0 — the most overbought reading in this batch, and comfortably above the conventional 70 threshold. MACD +17.28.
• Maximum drawdown from peak over the trailing year: −12.9%, the shallowest in this batch after COF.
• Relative performance, and the beta makes it remarkable: 3-month +13.7% against SPY +7.6%; 6-month −7.3% against SPY +11.1%; 12-month +40.0% against SPY +24.3% and QQQ +30.8%. A stock with a beta of 0.113 outperformed the S&P 500 by 16 points over twelve months — which means essentially all of the return is idiosyncratic re-rating, not market exposure. That cuts both ways: it can un-rate the same way.
• Sentiment: 21 buy, 15 hold, 1 sell across 37 analysts; consensus target $634.38 (+7.7%), median $627, high $700 (+18.8%), low $575 — 2.4% BELOW spot. Forty-three percent of the coverage is not at buy, and the street's implied upside of 7.7% is the second-lowest in this batch.

Today's move and what it does to the entry

LMT closed 2026-08-04 at $589.13, up 0.48% or $2.84. It opened at $587.19, traded $578.30 to $592.50, and closed near the high on 0.95M shares. No company-specific filing is dated 2026-08-04; the last events were the 2026-07-23 earnings release and 10-Q, twelve days earlier.

The honest read: the entry is the problem, not the business. RSI at 81.0 after a 40% twelve-month run, on a stock whose beta is 0.113, describes a name that has been bid up on a narrative rather than dragged up by the market. Our base fair value of $585 is below the price and the street's own low target of $575 is below it too.

What would change this, and these are the falsifiers:

• A price near $500 — roughly 15.3x FY2027E, close to the 200-day average of $546 overshot to the downside — which would put our base $585 at a 17% premium and restore a genuine margin of safety. That is 15% below today's close and well inside the range this stock has traversed in a year.
• Or the 2026-10-20 print with operating margin above 12.5% and no new reach-forward loss, which would begin to establish that the classified programmes are contained and would justify a higher base multiple. At that point paying up is defensible, because the uncertainty being paid for has been resolved.

9. Insiders — eight transactions, all mechanical, no signal

There is nothing here and the table is reported for completeness. Two director restricted-stock awards at zero cost; two director unit conversions coded M-Exempt at zero price; a charitable or personal gift (G-Gift) of 368 shares by the president of Rotary and Mission Systems; and three untyped entries for the president of Aeronautics with blank transaction types and zero prices, which the vendor has not classified.

Zero open-market purchases. Zero open-market sales. Zero tax-withholding dispositions. This is the emptiest insider file in this batch — and notably, not a single transaction by the chief executive or chief financial officer. With the stock up 40% over twelve months and at RSI 81, the complete absence of any discretionary sale is a very mild positive; the absence of any purchase is a very mild negative; and together they are neutral and uninformative.

One small data note: the three Orlando Sanchez entries carry an empty transactionType field, so they cannot be classified as acquisitions or dispositions. Two of them show holdings falling to zero and one shows 460.226 held after, which is consistent with an internal transfer between direct and indirect ownership on his appointment as President of Aeronautics. We do not read a signal into an unclassified field.

10. Verdict, kill-criteria and flip conditions

Watch.

What is genuinely strong, and none of it is in dispute: a backlog of $230.4 billion, 3.07x trailing revenue, up $36.8 billion in six months, with roughly 30% expected to convert within twelve months; international revenue rising from 26.1% to 28.4% of the total with Europe +14.1% and Asia Pacific +25.2%; Missiles and Fire Control up 28.4% in two years; a June quarter with revenue +10.5%, operating margin back to 12.4% and an EPS beat of 10.0%; trailing free cash flow of $8.75 billion, a 15.6x multiple and a 6.4% yield; a 2.32% dividend at a 50.5% payout plus roughly 2.2% of buyback; beta of 0.113; and a customer that is the United States government.

What we are declining to pay for: 21.6x trailing and 18.1x FY2027 consensus after a 40.0% twelve-month return into an RSI of 81.0; a base fair value of $585 that is 0.7% BELOW the closing price and 7.8% below the street's $634.38, in a sensitivity band where 17x gives $555; approximately $3.3 billion of cumulative reach-forward losses across two classified programmes, with $427 million still accrued, a stated expectation that future option exercises "would each be at a loss", and an explicit warning that the company "may need to record additional losses"; two of the last eight quarters with operating margins of 3.7% and 4.1%; a securities class action alleging misstatements about exactly those programmes; negative tangible book value of $18.75 per share and an 11.2% equity ratio; a $3.45 billion acquisition to be part-financed against that balance sheet; a knowledge-base lane whose newest and highest-conviction claim argues the core product is being obsoleted; and a sitting Treasury Secretary on record proposing to curb the capital return that supplies 4.5 points of expected annual return.

The distinction that matters. This is not a call that Lockheed is a bad business or that the rearmament cycle is fake — the backlog is real, the international mix shift is real, and the June quarter was genuinely good. It is a call that the price has already moved 40% to reflect all of that, that our arithmetic lands below it, and that the specific risk this company carries — an unforecastable multi-hundred-million charge appearing on the morning of a print — is not compensated at 18x with a 0.90:1 payoff ratio. In the Synthos frame, a name whose base fair value sits below the price is not a Watch by hesitation; it is a Watch with a number attached, and the number is roughly $500.

Pre-registered KILL criteria — what would take this to Avoid:

• Any new reach-forward loss of $500 million or more, on either classified programme, on the Canadian Maritime Helicopter Program, or on a programme not yet disclosed.
• Operating margin below 10.5% in a quarter without a disclosed one-off, which would say the margin recovery consensus needs is not happening.
• An adverse ruling or large settlement in the securities class action filed 2025-07-28.
• A ULA impairment or a required capital injection beyond the support already contemplated in the 10-Q.
• The Ultra Maritime financing arriving on terms that push net debt above roughly $22 billion without a corresponding earnings contribution.
• An F-35 procurement reduction in a US or major allied budget, or a definitive shift of programme funding toward autonomous collaborative aircraft.
• Policy action curbing defence-prime dividends or buybacks, which would remove roughly 4.5 percentage points of annual return at a stroke.

Pre-registered FLIP conditions — what would take this to Buy — Tactical:

• A price near $500 — roughly 15.3x FY2027E — with no deterioration in backlog or disclosure. This is the single most likely path to a Buy and it is 15% below today's close, well inside the range this stock traverses in a year.
• Two consecutive clean quarters with operating margin above 12.5% and no new reach-forward loss, which would materially narrow the earnings distribution and justify a higher base multiple.
• Definitisation of the THAAD undefinitized contractual action at a disclosed margin, converting $36.8 billion of backlog growth from an estimate into a contract.
• A resolution of the Canadian Maritime Helicopter Program that does not involve termination.
• Knowledge-base conviction turning decisively positive on the platform question — at present the newest and highest-conviction claim in the lane argues the opposite, and that alone is worth roughly a tier.

Where LMT fits in the Synthos Framework Portfolio. The defence and national-security sleeve, at 0% today with a 2-3% target on a fill near $500 or two clean quarters. On sector overlap: the vendor peer set contains General Dynamics, L3Harris and Northrop Grumman, none of which is in this batch, so no direct comparison is drawn — but the knowledge base's structural claims about primes (financialisation, delivery delays, new-entrant competition, the drone argument) apply to all of them equally, and a defence sleeve built on multiple primes is one trade with several expressions, not diversification. Logged as a tracked Synthos call (Watch) as of 2026-08-04 at $589.13.

Single biggest risk: another reach-forward loss. Two classified programmes have already absorbed approximately $1.8 billion and $1.5 billion; $427 million remains accrued on one; on the other the company states it expects the remaining unexercised options "would each be at a loss" and considers their exercise probable; and the Canadian Maritime Helicopter restructuring "could result in a contract termination". The June 2025 quarter shows what this looks like: $1,615 million of disclosed charges cutting operating income from a normal $2.3 billion to $748 million, and net income to $342 million. No sell-side model contains such a charge — the FY2027 consensus EPS range across 16 analysts spans just 8.7%, far narrower than the distribution the last eight quarters actually produced. At 18.1x forward earnings with a base fair value below the current price, there is no valuation cushion to absorb one, and the disclosure will arrive on the morning of a print with no warning. That is the $442 bear case, and it requires no change in the demand environment at all.

Provenance & disclosures

• Traceability: 9 tagged knowledge-base claims name Lockheed Martin out of 51,928 distilled claims, of which 7 enter the conviction pool and 2 are EXCLUDED as suspected management voice (raw hits 17, entity matches 9, text-only 8, discarded as homographs 0; breadth 11, net conviction mixed-slightly-negative). The entity set is 5 bullish / 3 bearish / 1 neutral before exclusions and 3 bullish / 3 bearish / 1 neutral after — the most balanced lane in this batch. Bullish: odd_lots 2026-03-16 (conviction 75, horizon FACT, PAC-3 ramping ~600→2,000/year and Tomahawk toward 1,000 from 57 — independently corroborated in direction by the 10-Q's MFC production commentary), Dan Ives on anthony_pompliano 2026-07-27 (conviction 55, AI second-derivative basket) and mikkel_rosenvold 2026-03-02 (conviction 40, UAE restocking). Bearish: a16z 2026-03-20 (conviction 68, financialisation over growth — corroborated by $20.6B of buybacks in four years and negative tangible book), lex_fridman 2025-08-24 (conviction 80, an ethical rather than financial claim, weighted accordingly) and lex_fridman 2024-10-23 (conviction 60). Neutral: natalie_brunell 2025-06-28 (conviction 60, domestic rocket-motor supply constraint). EXCLUSIONS: two lex_fridman claims dated 2019-08-19 (conviction 85 and 70), seven years old, no named speaker, reading unmistakably as a Lockheed executive's own roadmap on a host-named channel — the exact failure mode the 4-lane attribution policy exists to catch. Removing them flips the lane from 5-3-1 to 3-3-1 and is why net conviction is recorded as slightly negative rather than positive. Two text-only claims are reported in full despite failing the entity test because they are more decision-relevant than several that pass: scott_bessent 2026-01-20 (role POLICYMAKER, conviction 66) on curbing primes' dividends and buybacks, which bears directly on 4.5 points of expected return; and jim_bianco 2026-07-30 (conviction 82, horizon PRINCIPLE, the most recent and highest-conviction claim in the entire sweep, dated five days before this dive) on cheap mass drones beating expensive legacy systems, naming the F-35 explicitly. Also recorded: no_priors 2026-01-15 (conviction 80) on new-entrant primes being courted, and luke_gromen 2026-01-27 (two claims, conviction 63-65) putting a 60-70% base case on an eight-month US inability to produce weapons such as the F-35 on critical-mineral dependence — an assertion we could not corroborate in any filing and record as unverified. No homograph collisions occurred; "Lockheed" is a clean token. Maximum single-channel share is 23.5% (lex_fridman, 4 of 17), below the one-third threshold, so no concentration sensitivity test is triggered — stated explicitly because the test is mandatory when it is. All quotes are verbatim from the stored claim text.
• Data as-of: fundamentals — the most recent filing-verified narrative disclosures are as of 2026-06-28 from the 10-Q filed 2026-07-23 (backlog $230.4B, reach-forward loss balances, F-35 deliveries, litigation, and the Ultra Maritime agreement in Note 12); financial-statement figures come from the vendor payload for the reason set out below · estimates 2026-08-04 · prices 2026-08-04, quote timestamp 1785873599 = 2026-08-04T19:59:59Z ($589.13, +0.48%; 50-DMA $531.29; 200-DMA $545.94; RSI 81.0; MACD +17.28) · knowledge-base claims 2026-08-04. Lockheed's fiscal year is the calendar year; quarters end on the Sunday nearest quarter-end (Q2 2026 ended 2026-06-28). All figures come from the Synthos vendor data file for LMT or from the SEC filings in the LMT archive; no figure comes from memory, recall or external retrieval.
• Filing archive contents and its limitation: 10-K filed 2026-01-29 (fiscal 2025); 10-Q filed 2026-04-23 (March 2026 quarter); 10-Q filed 2026-07-23 (June 2026 quarter — the source of every filing-verified figure in this dive); 8-K filed 2026-04-23 (Item 2.02, first-quarter results — cover document only); 8-K filed 2026-05-13 (annual meeting voting results, and the source of the 230,462,118 share count at the 2026-03-02 record date); 8-K filed 2026-07-23 (Item 2.02, second-quarter results — cover document only). The archive is flagged tables: false in the manifest: the filings were extracted as prose with no preserved statement tables, and figures appear on separate lines from their labels. No primary income statement, balance sheet or cash-flow table was available for direct reconciliation, and both earnings 8-Ks are cover documents with Exhibit 99.1 unextracted — so NO management guidance figure appears anywhere in this dive. All financial-statement figures come from the vendor payload, cross-checked against earn_cal actuals and against narrative amounts in the 10-Q. This is stated so no reader mistakes the annual and quarterly tables here for filing-verified data.
• Where the filings supplied what the vendor could not, and where they contradicted it (detailed in Section 7): the $3.45 billion Ultra Maritime Solutions acquisition announced 2026-07-06 and expected to close in Q4 2026, absent from every vendor field, with acquisitionsNet reading exactly $0 in all four cash-flow years; backlog of $230.4 billion at 2026-06-28, up $36.8 billion in six months, which no vendor field carries; approximately $1.8 billion and $1.5 billion of cumulative reach-forward losses on two classified programmes with $427 million still accrued, and the disclosure that unexercised options on one "would each be at a loss"; the securities class action filed 2025-07-28; the F-35 delivery and backlog counts; and the ULA support warning. Vendor figures rejected: inc_a FY2025 depreciation and amortisation of $714M against the cash-flow statement's $1,687M (a 2.4x internal contradiction confined to that year, which makes inc_a.ebitda of $7,754M imply $23M of D&A); trailing EBITDA, recomputed from the vendor's implied $9,856M to $10,842M using the vendor's own capexToDepreciationTTM; est.ebitdaAvg/ebitAvg, rejected for a fixed-ratio signature (exactly 13.107% and 11.222% of revenue in every forward year — unusually close to Lockheed's real margins, which makes it harder to catch and no less disqualifying); and quote.yearHigh/yearLow ($692.00/$421.86) rejected for tech.hi52/lo52 ($676.70/$423.70). Where vendor and filing AGREED — recorded because these defect classes usually bite: seg_prod sums to $75,048M against reported revenue of $75,057M with all four segments present — the cleanest segment table in this batch; seg_geo sums to the same figure across five regions with no omission; bookValuePerShareTTM of $38.089 correctly distinct from tangibleBookValuePerShareTTM of −$18.753; and the share count (8-K record-date 230,462,118 against 230.8M implied).
• Basis note: the earn_cal "actual" EPS figures appear to be on an adjusted basis close to GAAP — the June 2026 actual of $7.94 against GAAP net income of $1,836M on 231.1M diluted shares gives $7.94 exactly, so for this company the two coincide in the current quarter. In quarters carrying reach-forward losses they will not. epsdiluted is null in every row of the vendor's income statements, so all trailing per-share figures here are computed and labelled. Forward multiples use est.epsAvg.
• Estimate coverage: 14-15 analysts on FY2026, 16 on FY2027 — the anchor for all three fair values, with a range of $31.600 to $34.338 (8.7%) — 16-17 on FY2028 and 13-15 on FY2029. This is the best-covered estimate block in the batch. Note that the FY2029 revenue line of $91.32B against FY2028's $90.37B implies +1.0% growth, a deceleration that caps the multiple justifiable on the outer years and is a direct input to the compression assumed in the return bridge.
• Peer note: the vendor peer set — ADP, Boeing, Deere, General Dynamics, Honeywell, L3Harris, Northrop Grumman, Parker Hannifin, TransDigm, Trane — contains four genuine defence comparables and six unrelated industrials. No peer multiples are supplied in the file, so no peer-multiple comparison is drawn.
• Fair-value caveat: the $442 / $585 / $721 anchors are multiples of the FY2027 consensus EPS distribution — 14.0x the low of $31.600, 17.9x the mean of $32.632, and 21.0x the high of $34.338. Stated arithmetic, not a discounted cash flow. The base is sensitivity-disclosed: 17x gives $555, 5.8% BELOW spot; 19x gives $620. This is the only name in this batch whose base fair value sits below the closing price, and the asymmetry of 0.90:1 is unfavourable — both facts are the verdict rather than caveats to it.
• Timing: second-quarter 2026 results were released 2026-07-23, twelve days before this dive, beating consensus EPS by 10.0% ($7.94 against $7.22). The next print is 2026-10-20, 77 days away. The Ultra Maritime acquisition is expected to close in the fourth quarter of 2026. 2026-08-04 carried no company-specific filing.
• Accessibility note: no information in this dive is conveyed by colour. All emphasis is carried by bold text, table structure and explicit labelling.
• Not investment advice. Independent research, educational and informational only, never personalised. No recommendation to buy, sell or hold any security is made to any person.
• Version: 2026-08-04-full.