General Dynamics GD

Reference table

1. What the business is

General Dynamics is a Delaware corporation headquartered in Reston, Virginia, NYSE-listed, with approximately 117,000 employees, led by Chair and Chief Executive Phebe N. Novakovic, who has held the role since 2013. The fiscal year ends on the Sunday nearest 31 December; the second quarter of 2026 ended 5 July 2026.

Four reportable segments, from the 10-Q filed 2026-07-29 ($M):

First half 2026 ($M):

Three observations that shape everything else.

The growth is concentrated in the lowest-margin segment. Marine Systems grew 15.3% in the first half and earns 7.3%; Aerospace grew 11.8% and earns 14.7%. Because Marine is a third of revenue and Aerospace a quarter, consolidated margin faces a structural mix headwind even when both segments improve — which is exactly what happened: Marine margin rose 40 basis points and Aerospace 130, and consolidated margin rose only 30.

Combat Systems has stopped growing. +0.3% in the quarter and +2.6% in the half, with margin down 30 basis points to 13.9%. This is the armoured-vehicle business, and it is the segment most directly exposed to the "cheap drones make armour obsolete" argument that appears twice in the knowledge base. (Against which: the Q2 backlog increase was partly driven by awards "to produce armored combat support vehicles" in this very segment — so the orders are arriving even as the revenue is flat, which is a timing statement rather than a demand statement.)

Technologies is the odd one out. 25.7% of revenue, 4.1% growth, a 9.4% margin that is declining. It is a government-IT services business sitting inside a platform manufacturer, and it earns neither Aerospace's margin nor Marine's growth.

2. The backlog — quoted, because it is the thesis

From the 10-Q, verbatim:

> "Our total backlog, including funded and unfunded portions, was $136.5 billion at the end of the second quarter of 2026 compared with $130.8 billion at the end of the first quarter... Our total estimated contract value, which combines total backlog with estimated potential contract value, was $186.9 billion on July 5, 2026."

> "Total backlog in our defense segments was $112.5 billion on July 5, 2026. The defense segments achieved a book-to-bill ratio of 1.4-to-1 in the second quarter of 2026 and 1.8-to-1 in the first six months. The increase in backlog was primarily driven by awards in the Marine Systems segment for continued construction of Virginia-class submarines and to produce armored combat support vehicles in the Combat Systems segment. Estimated potential contract value in our defense segments was $49.2 billion on July 5, 2026."

> "The Aerospace segment ended the second quarter of 2026 with backlog of $24 billion... On July 5, 2026, estimated potential contract value in the Aerospace segment was $1.2 billion."

What a 1.8-to-1 book-to-bill actually means, stated without inflation. For every dollar of defense revenue recognised in the first six months of 2026, $1.80 of new orders arrived. Backlog grew $5.7 billion in a quarter in which the company recognised $14.1 billion of revenue. At the current pace the pile of contracted future work is growing faster than the company can consume it, which is the single most reliable leading indicator this industry produces.

Three honest qualifications.

Backlog is not revenue and its conversion rate is not disclosed here. The 10-Q states GD expects to recognise a proportion of remaining performance obligations over time, but the specific percentage falls outside the extracted text. We do not estimate it.

"Unfunded" backlog carries an explicit warning from the company itself: "The unfunded portion of total backlog includes the amounts we believe are likely to be funded, but there is no guarantee that future budgets and appropriations will provide the same funding level currently anticipated for a given program." That is a real caveat on a defense contractor whose customer's budget is set annually by Congress.

A 1.8-to-1 half-year ratio is a step-change, not a run rate. It was produced by specific large awards — Virginia-class submarine construction and combat support vehicles — and multi-billion-dollar programme awards are lumpy. A reading near 1.0 next quarter would be normal, not a deterioration.

3. Earnings, cash and the balance sheet

Five consecutive beats averaging 6.1%. And GD reports no adjusted EPS at all — the earn_cal actuals are GAAP diluted figures and reconcile exactly to net earnings divided by diluted weighted average shares ($1,160M ÷ 273,522 thousand = $4.24). That is the cleanest earnings presentation in this batch: no adjusting items, no non-GAAP bridge, no basis question.

The cash statement is the most improved thing in the file and it is derived from the 10-Q ($M):

(Operating cash flow is derived from the 10-Q's disclosed net change in cash of +$2,000M less investing of −$424M, financing of −$1,605M and discontinued operations of −$6M. The direct line falls outside the extracted text; the derivation is arithmetic and is labelled as such.)

First-half free cash flow of $3,598 million against $1,110 million a year earlier — a 224% increase — is the largest working-capital swing in this batch, and it flows straight through to the balance sheet:

Net debt of $3,183 million against trailing EBITDA of approximately $6,450 million is roughly half a turn. Interest coverage is 22.2x on the vendor's calculation and interest payments were $149 million in the first half against $203 million a year earlier. This is a materially unlevered balance sheet and it is getting less levered.

The capital return is the weak point and deserves to be named as such. First-half repurchases were $319 million — 0.3% of market capitalisation — against $600 million a year earlier. Basic weighted average shares rose from 268,138 thousand to 270,156 thousand year on year, +0.75%. The buyback is not shrinking the share count; it is not even fully offsetting equity-award dilution. The 10-Q describes capital deployment priorities as "investments in our business infrastructure, products and services to drive long-term growth, a predictable dividend, strategic acquisitions and opportunistic share repurchases primarily to address dilution." "Primarily to address dilution" is an accurate description of the intent and an inaccurate description of the result. Dividends of $834 million in the half at a dividendPerShareTTM of $6.18 give a 1.60% yield; the June quarter declared $432 million, or approximately $1.60 a share, $6.40 annualised. Total shareholder yield is therefore roughly 1.6% dividend plus 0.6% gross buyback less 0.75% dilution ≈ 1.5% — the lowest in this batch, and a real negative against names in this same file returning 4-7%.

The event dated 2026-08-04, and it is small but it is today. An 8-K filed 2026-08-04 discloses that General Dynamics "filed a prospectus supplement, dated August 4, 2026... relating to the Company's offer to rescind (the 'Rescission Offer') the acquisition of up to 1,010,925 shares of the Company's common stock, par value $1.00 per share", under an automatic shelf Registration Statement on Form S-3 filed 2026-07-30, accompanied by a legal opinion from Gibson, Dunn & Crutcher LLP as to the validity of the shares. A rescission offer is the standard securities-law remedy where shares were issued without an available registration exemption — typically under an employee benefit plan — and it gives the holders the right to unwind the purchase. 1,010,925 shares is 0.37% of the count. The financial consequence is immaterial; the disclosure is a compliance irregularity and it is the only one in this file. We record it, we do not weight it, and we flag the resolution as something to watch.

4. Where the vendor and the filings disagree on debt

Two separate defects in the same area, and they run in opposite directions.

The lease inclusion. The vendor's FY2025 totalDebt of $9,789M equals the filing's pure debt of $8,013M (short-term and current portion $1,006M + long-term debt $7,007M) plus capitalLeaseObligations of $1,776M. That is a 22.2% overstatement of debt, and it flows into netDebt of $7,456M against a filing-based $5,680M — a 31.3% overstatement. This is the T/TJX lease-inclusion defect class and at 22% it is the largest instance in this batch.

The enterprise-value contradiction, which points the other way. enterpriseValueTTM of $109,516M less marketCap of $104,370M implies net debt of $5,146M. The bal_a block in the same file says $7,456M. The two disagree by $2,310M. Unusually, the enterprise-value field is the closer of the two to the filing's $5,680M — within 9.4%, against the balance block's 31.3% error. This is the ETN/ABT internal-contradiction defect class with an inverted resolution: normally the balance block is right and the EV field wrong; here it is the reverse.

What we use. The filing's figures at 2026-07-05: cash $4,333M, pure debt $7,516M, net debt $3,183M. On that basis true enterprise value is approximately $107,553M and EV/EBITDA on trailing EBITDA of $6,450M is 16.7x, against the vendor's 16.98x — a 1.8% difference. Both netDebtToEBITDATTM of 0.798x and debtToEquityRatioTTM of 0.353x overstate leverage; the corrected figures are approximately 0.49x and 0.29x.

5. Valuation — priced in or room?

At $385.76 (market cap $104.37B, ~270.6M shares):

Estimate coverage is the deepest in this batch and that is worth recording. 16 analysts on FY2027 EPS, 16 on FY2028, and 14 each on FY2029 AND FY2030 — most names in this programme collapse to 2-4 analysts by the third forward year. The FY2027 EPS range is $17.633 to $19.398, a 10.0% spread. (Even so, FY2029 and FY2030 are used only as directional cross-checks and no fair value rests on them.)

One structural observation about the estimates that is easy to miss. Revenue growth decelerates every year — 6.1%, 4.8%, 4.3%, 3.5% — while EPS growth stays near 8-9%. So roughly half of consensus earnings growth from FY2027 onward has to come from margin expansion, and the June quarter delivered exactly 30 basis points of it consolidated (10.7% against 10.4%). That is achievable and it is not automatic, and it is the least-discussed assumption in the price.

est.ebitdaAvg and est.ebitAvg are REJECTED for the fixed-ratio fabrication signature. ebitdaAvg is exactly 12.637% of revenueAvg and ebitAvg exactly 10.687% in every forward year FY2026 through FY2030. They are also wrong in level: FY2025 ebitdaAvg of $6,568.1M against a reported $6,097M, a 7.7% overstatement. All forward valuation uses epsAvg.

Peer context is unusable. The vendor peer set is ADP, HWM, LHX, LMT, MMM, NOC, PH, TDG, TT and UPS — of which LMT, NOC, LHX, HWM and TDG are genuine aerospace-and-defense comparables and the other five are not (a payroll processor, a diversified industrial, a fluid-power manufacturer, an HVAC company and a parcel carrier). The payload supplies only market capitalisations, no multiples, so no peer-multiple comparison is drawn. (Recorded: LMT $136.0B, PH $125.2B, HWM $115.3B, ADP $108.2B, MMM $93.6B, UPS $92.7B, NOC $78.4B, TDG $71.3B, LHX $53.1B. ADP and TT are separately covered in this same batch.)

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

At $385.76 — 22.8x FY2026 consensus, 20.9x FY2027, 16.7x EV/EBITDA — the price embeds:

• Diluted EPS reaches $16.890 in 2026 and $18.476 in 2027. (Consensus; 13 and 16 analysts.) First-half 2026 diluted EPS was $8.34, so FY2026 requires $8.55 in the second half — a 2.5% sequential step, which the Q3 consensus of $4.14 against a Q2 actual of $4.24 does not even ask for. Given five consecutive beats averaging 6.1%, the FY2026 number looks beatable by 2-4%.
• Consolidated operating margin expands from 10.7% toward roughly 11.5% by FY2028. (Our derivation from the consensus revenue and EPS rows.) Revenue growth decelerates to 4.3% while EPS grows 7.9%, and the share count is rising, so the gap must be margin. This is the most fragile assumption in the price, because the fastest-growing segment (Marine, 33.1% of revenue) earns 7.3% and mix therefore works against it — every point of consolidated margin expansion has to be won inside segments, against a mix drag.
• The backlog converts at historical rates and the book-to-bill does not collapse. (Filing-sourced backlog; our derivation of the consequence.) $136.5 billion of backlog and $186.9 billion of total estimated contract value make the revenue line unusually predictable for five years — provided appropriations follow authorisations, which the company itself warns is not guaranteed for the unfunded portion.
• The market keeps paying roughly 21-23x forward earnings. (Our number.) At 18x FY2027E the stock is $333; at 24x it is $443. Defense primes have historically traded on the defense-budget cycle rather than on company-specific news, and this archive contains no budget data, so this band is wider in reality than the estimate dispersion suggests.
• Cheap-drone disruption does not reach GD's programmes of record within the estimate horizon. (Our number; the knowledge base supplies the risk.) The price is paying 20.9x for a platform manufacturer at a moment when two independent voices in our store argue platform manufacturers are structurally impaired and a third argues new entrants are being actively courted.

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

Expected return over the next twelve months decomposes as: diluted EPS growth (+9.4%, from FY2026E $16.890 to FY2027E $18.476) + multiple drift (COMPRESSION of roughly 1%, from today's 22.8x on the forward year toward about 22.7x) + shareholder yield (+1.5%) ≈ +10%.

Our base assumes the multiple essentially holds, and that is a deliberate decision not to underwrite either the disruption bear case or a defense-budget re-rating. At $420 the stock would be 22.7x FY2027 consensus and 21.1x FY2028 — approximately where it trades today on the equivalent forward year. The return is therefore almost entirely earnings growth, with a shareholder yield of only about 1.5% adding very little.

Note the contrast with other names in this batch: GD returns 1.5% a year to shareholders where COF returns 7.4% and BNY 5.5%. A company with a $3.2 billion net debt position, a 22x multiple and $4.3 billion of cash choosing to buy back $319 million in a half-year is making a capital-allocation choice, and at this multiple it is arguably the right one — but it means the entire return has to come from earnings and the multiple.

If the multiple compressed to 18x FY2027E the price would be $333 (−13.7%). If it expanded to 24x, $443 (+14.9%).

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

• We have NO variant perception, and the coincidence is almost exact. Our base of $420 sits 0.2% below the street's $420.67 and at the median target of $420, with 19 of 35 analysts at buy. We are not seeing anything the market is missing, and under this programme's rules that defaults toward Watch or Hold. This is the third name in this batch where the arithmetic and the consensus coincide, and saying so plainly is more useful than manufacturing a difference.
• Where we are marginally more constructive than consensus is the near-term estimate, and it is arithmetic rather than insight. Q3 consensus of $4.14 is BELOW the Q2 actual of $4.24, and the FY2026 consensus of $16.890 requires only a 2.5% sequential step in the second half after five consecutive beats averaging 6.1%. Watchable number: the 2026-10-28 print against $4.14. But a low bar cleared on a 22.8x multiple is worth very little, which is why it does not move our verdict.
• Where we are more cautious than consensus is the margin path. Consensus needs consolidated margin to expand while the lowest-margin segment grows fastest. Watchable number: Marine Systems operating margin, at 7.3%. Every point of margin there is roughly $180 million of annual operating income, about 3.5% of the total — and the segment has moved 40 basis points in a year.
• We think the knowledge base's disruption lane is under-priced at 20.9x, and we say so without claiming to resolve it. jim_bianco (2026-04-22, bearish 66) names General Dynamics directly as one of the primes that are "the past"; luke_gromen (2025-07-29, bearish 78) argues "cheap drones/missiles rendered trillions of carrier-based defense spending obsolete"; no_priors (2026-01-15, bullish 80 — on the competitors) reports that new-entrant primes "once had to sue for contracts" and are now "actively courted." The Bianco claim's testable half is already wrong — GD has matched the index over twelve months and beaten it over three. But a $136.5 billion backlog is a five-year answer to a fifteen-year question, and Combat Systems — the segment most exposed to the argument — grew 0.3% last quarter with margin down.
• Positive surprise that would force a re-rate: a third consecutive quarter of defense book-to-bill above 1.2 with Marine Systems margin above 8%, which would show the submarine ramp is converting to profit rather than only to revenue; or a Gulfstream order cycle that pushes Aerospace backlog materially above $24 billion with margin above 15%.
• Negative surprise that would break the thesis: defense book-to-bill below 1.0 for two consecutive quarters; Marine Systems margin falling below 7.0% on a programme cost overrun, which is the classic shipbuilding failure mode and which the company's own risk language about "cost and schedule overruns" anticipates; Aerospace margin below 13% on a business-jet down-cycle; a new-entrant prime taking a programme of record in a GD category; or an appropriations shortfall against the unfunded portion of backlog, which the 10-Q explicitly warns is not guaranteed.

Synthos fair values

All three anchors are multiples of the FY2027 consensus diluted EPS distribution (mean $18.476, low $17.633, high $19.398, 16 analysts), each cross-checked against FY2028E.

• Bear ~$320 — 18.1x the FY2027 consensus LOW of $17.633, and 16.1x FY2028E. Cross-check: 2.4% above the 52-week low of $312.53; 8.6% below the 200-day average. The scenario: book-to-bill falls below 1.0, a Marine Systems programme takes a cost overrun, Aerospace enters a business-jet down-cycle, and the market applies a disruption discount to platform primes. −17.0%.
• Base ~$420 — 22.7x the FY2027 consensus MEAN of $18.476, and 21.1x FY2028E. Cross-check: 0.2% below the street's $420.67; exactly the median target of $420; 6.8% above the 52-week high. Sensitivity, stated openly: 20x gives $370 — below spot — and 24x gives $443. The whole answer lives in a 20-24x band on one estimate row. The scenario: the backlog converts, consolidated margin grinds up 30-40 basis points a year, EPS compounds at 8-9%, and the multiple holds. +8.9%.
• Bull ~$480 — 24.7x the FY2027 consensus HIGH of $19.398, and 24.1x FY2028E. Cross-check: 3.2% above the highest street target of $465; 22.1% above the 52-week high. The scenario: the naval rebuild the knowledge base's independent lane describes translates into sustained submarine awards above 1.5-to-1 book-to-bill, Marine margin reaches 9%, and the market re-rates the only US company outside one competitor that can build a nuclear submarine. +24.4%.

Base is 8.9% above spot; asymmetry roughly 1.43:1 to the upside (17.0% down, 24.4% up), before a shareholder yield of approximately 1.5%. A base under 9%, a payoff ratio under 1.5:1, a base 0.2% BELOW the street's, an entry 1.9% from the 52-week high with RSI at 69, and a knowledge-base lane that is negative at the name level — that is a Watch. The price at which it becomes a Buy is named in Section 9.

6. Knowledge base — two name-level claims, both bearish, one already wrong

Raw hits: 46 (2 case-sensitive entity, 44 free text). After a case-sensitive entity re-run: 2. Name-level claims on General Dynamics: 2, and BOTH are BEARISH. Discarded: 44.

The case-sensitive entity sweep was run first — "GD" is a collision-prone two-letter ticker and the brief flagged it as such. Entity terms GD, General Dynamics, Gulfstream and Electric Boat across all 51,928 distilled claims returned exactly two.

The two name-level claims, verbatim:

> 2026-04-22 · BEARISH · conviction 66 · horizon: thesis · entities: RTX, GD, BA · channel: jim_bianco · speaker: Jim Bianco · skill 0.8

> "Raytheon, General Dynamics and Boeing are the past; cheap-drone warfare demands new tech-oriented defense, so the primes are correcting and underperforming despite the conflict — a bear's trend."

> 2024-10-23 · BEARISH · conviction 60 · horizon: thesis · entities: Lockheed Martin, General Dynamics · channel: lex_fridman · no named speaker · skill 1.0

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

Grading the Bianco claim, because it is the only one in the store that names GD as a ticker and it contains a falsifiable half. It asserts the primes are "correcting and underperforming despite the conflict." Since 2026-04-22 GD has returned 10.5% over three months against SPY's 7.6% and QQQ's 7.7%, and 24.1% over twelve months against SPY's 24.3%. That is outperformance over the shorter window and parity over the longer one — not underperformance. The claim's testable half is wrong so far. Its structural half — that cheap-drone warfare demands a different kind of defense company — is not testable on a three-month horizon and is the most serious long-run argument against this security that exists anywhere in our sources. We report both halves and grade only the one that can be graded. (Note the speaker is named and speaker_role is null — consistent with the store's confirmed unreliability of that flag; Jim Bianco is an independent strategist, not management, and the content confirms it.)

The lex_fridman claim is not gradeable and we say so. "Massive fraud, waste and cost overruns" in Pentagon procurement is a claim about a system, not a security; it names GD only alongside Lockheed Martin as instances; and it implies nothing testable about GD's earnings, backlog or share price on any horizon. It is recorded, weighted at essentially zero, and is the reason the conviction reads negative-low rather than simply negative.

The 44 free-text discards form a genuinely two-sided defense lane across 23 channels, and unlike most discarded lanes it is worth reading.

Pointing toward General Dynamics — and specifically toward Electric Boat:

> 2025-07-23 · bullish · conviction 75 · channel: all_in

> "The US must reshore submarines, ships, munitions, rare-earth magnets and drones first — China outproduces the US ~200x in shipbuilding."

> 2025-05-16 · bullish · conviction 68 · channel: forward_guidance

> "US naval industrial capacity is stagnant at ~1.3% of defense budget; shipbuilders are interesting as spending rebuilds in-house capacity versus China."

> 2025-09-05 · conviction 65 · channel: dwarkesh

> "China, ringed by narrow seas like Germany, is geographically vulnerable; surface fleets are useless in high-stakes war — submarines/commerce-raiding are the fit."

> 2025-08-28 · bullish · conviction 75 · channel: no_priors

> "Global defense spending hit a record $2.7T; AI and autonomy are reshaping battlefield outcomes, redefining 21st-century hard power."

> 2025-09-21 · bullish · conviction 75 · channel: brent_johnson · skill 0.7

> "Defense and aerospace is a huge beneficiary as nations rearm and boost national-defense spending in a less cooperative world."

Pointing away:

> 2025-07-29 · BEARISH · conviction 78 · channel: luke_gromen

> "Cheap drones/missiles rendered trillions of carrier-based defense spending obsolete; US 'defense-as-a-service' is no longer credible, eroding the dollar-empire structure."

> 2026-01-15 · bullish 80 — ON THE COMPETITORS · entities: SpaceX, Anduril, Palantir · channel: no_priors

> "New-entrant defense primes (SpaceX, Anduril, Palantir) once had to sue for contracts; DoW now actively courting lower-cost, faster, more sophisticated startup options."

> 2026-03-05 · neutral · conviction 55 · channel: luke_gromen

> "Rising defense spending helps those stocks but they can't move faster than China lets them via rare-earth supply."

The synthesis, and it is unusually clean for a discarded lane. The bullish claims converge specifically on naval and submarine capacity — the exact business that drove GD's $5.7 billion of backlog growth last quarter. The bearish claims converge on platform manufacturers being displaced by cheap precision munitions and by faster new entrants — which applies best to Combat Systems, the segment that grew 0.3% with margin declining, and worst to submarines, which no drone replaces. A reader could construct a coherent bull case for GD entirely out of the claims that name no company, and a coherent bear case out of the two that name it. Neither enters the conviction score, and both are in the analysis.

Attribution and hygiene.

• Channel distribution across the free-text lane: lex_fridman 5, niall_ferguson 4, forward_guidance 3, dwarkesh 3, jacob_shapiro 3, luke_gromen 2, no_priors 2, brent_johnson 2, natalie_brunell 2, odd_lots 2, marko_papic 2, mikkel_rosenvold 2, scott_bessent 2, and ten channels with one each. No channel supplies more than 11% of the lane, so no concentration sensitivity test is triggered — the most diversified lane in this batch.
• ONE hit carries the eli_lilly_ceo slug and is EXCLUDED on that ground alone. That slug is the store's confirmed cross-company attribution defect — it is known to carry claims about other registrants — and any claim reaching a defense sweep through it is excluded without further examination.
• One hit carries the scott_bessent slug — a government-official channel — and two carry jensen_huang; none is General Dynamics management. No claim in the entire lane carries a management speaker_role for this registrant.
• Named speakers appear on four claims (Jim Bianco, Jensen Huang, and two others); the remainder carry speaker: null.

Conclusion. Breadth 2, claim count 2, net conviction negative-low. The Synthos knowledge base names General Dynamics twice in 51,928 claims and is bearish both times — once on a thesis whose testable half the price has already falsified, and once on an unfalsifiable claim about Pentagon procurement. Around them sits a two-sided, well-sourced, twenty-three-channel argument about whether the next war is fought with submarines or with drones, in which GD happens to be well positioned for one answer and poorly positioned for the other.

7. Data integrity — what we rejected and why

Six findings. GD's vendor file is among the cleaner ones in this programme — the share count, the segment table and the earnings basis all check out exactly — against two debt-related defects that pull in opposite directions and one geography block that double-counts its own subtotal.

1. totalDebt includes $1,776M of lease obligations — a 22.2% overstatement, the largest in this batch. The vendor's FY2025 totalDebt of $9,789M equals the filing's pure debt of $8,013M plus capitalLeaseObligations of $1,776M. netDebt of $7,456M against a filing-based $5,680M is a 31.3% overstatement. Consequential corrections: netDebtToEBITDATTM of 0.798x (true ≈ 0.49x on the current filing figures), debtToEquityRatioTTM of 0.353x (true ≈ 0.29x), debtToAssetsRatioTTM of 0.158x and interestDebtPerShareTTM of $36.03. This is the T/TJX defect class. All debt figures in this dive use the filing's pure-debt basis at 2026-07-05: cash $4,333M, debt $7,516M, net debt $3,183M.

2. The enterprise-value field and the balance block disagree by $2,310M — and the EV field is the closer of the two to the filing. enterpriseValueTTM of $109,516M less market capitalisation of $104,370M implies net debt of $5,146M; bal_a says $7,456M; the filing says $5,680M at FY2025 and $3,183M at 2026-07-05. This is the ETN/ABT internal-contradiction class with an inverted resolution — normally the balance block is right and the EV field is the error; here the EV field is within 9.4% of the filing and the balance block is 31.3% off. We use the filing throughout: true enterprise value is approximately $107,553M and EV/EBITDA approximately 16.7x, against the vendor's 16.98x.

3. seg_geo FY2025 contains a duplicate subtotal that would inflate a naive total by 85%. The FY2025 entry reads: North America $45,810M, UNITED STATES $43,314M, Europe $3,540M, Other Africa/Middle East $2,015M, Asia Pacific $2,001M, South America $432M. The United States is a subset of North America, so summing all six gives $97,112M against reported revenue of $52,550M. The correct read is North America $45,810M (87.2% of revenue) with the United States alone at $43,314M (82.4%). Note also that the "UNITED STATES" line appears ONLY in FY2025 — every prior year reports North America without a US split — so the series is not comparable year to year. (The remaining lines sum 2.4% above reported revenue in FY2025 and 2.7% above in FY2024, consistent with pre-elimination gross figures.) This is the same defect class found on BNY in this batch.

4. est.ebitdaAvg and est.ebitAvg carry the fixed-ratio fabrication signature — REJECTED. ebitdaAvg is exactly 12.637% of revenueAvg and ebitAvg exactly 10.687% in every forward year FY2026 through FY2030. They are also wrong in level: FY2025 ebitdaAvg of $6,568.1M against a reported $6,097M, a 7.7% overstatement, and FY2025 ebitAvg of $5,554.7M against reported operating income of $5,356M. All forward valuation uses epsAvg.

5. The 8-K carrying second-quarter results contains no financial data. The 8-K filed 2026-07-29 states only that "a copy of the press release is being furnished as Exhibit 99.1" — the release itself is not in the extracted text. All quarterly segment, backlog, cash-flow and balance-sheet figures in this dive therefore come from the 10-Q filed the same day, which is complete and carries preserved [TABLE] statement data. No figure is estimated around the missing exhibit.

6. The filing archive contains duplicate copies of every document. Each of the 10-K, both 10-Qs and all three 8-Ks appears twice — once with an accession suffix and once without (10-K_2026-01-30.txt and 10-K_2026-01-30_000006.txt). The pairs are identical in date and form. This is a pull-script artefact, not a data problem, and it is recorded so that a future reader does not mistake six documents for twelve.

Where vendor and filing AGREED — recorded, and GD passes more checks than any name in this batch:

• The share count matches to 0.15%. Market capitalisation ÷ price implies 270,563,000 shares against the 10-Q's basic weighted average of 270,156,000 for the June quarter.
• seg_prod is EXACT. The FY2025 entry — Aerospace $13,110M, Combat Systems $9,246M, Marine Systems $16,723M, Technologies $13,471M — sums to $52,550M, precisely the reported FY2025 revenue. The four categories are the four reportable segments, correctly named, with a complete sixteen-year history. This is the cleanest seg_prod block in this batch.
• inc_q Q2 2026 revenue of $14,094M matches the 10-Q's segment total exactly (3,525 + 4,660 + 2,290 + 3,619).
• The earnings basis is unambiguous and reconciles. earn_cal epsActual of $4.24 for Q2 2026 equals net earnings of $1,160M ÷ 273,522 thousand diluted shares. GD reports no adjusted EPS, so there is no basis question at all — a rarity in this programme.
• effectiveTaxRateTTM of 17.62% is consistent with FY2025's $893M tax on $5,103M of pre-tax income (17.5%).
• tech.max_dd_from_peak of −1.90% equals pct_from_hi, which is the ADP-class signature — but it was CHECKED and is CORRECT. ret_12m of +24.13% on a last price of $385.735 implies approximately $310.75 twelve months ago, within 0.6% of lo52 of $312.53 — so the low preceded the high and a 1.9% maximum drawdown is arithmetically consistent. Verified, not flagged.
• quote.yearHigh/yearLow of $400.00/$306.77 sit 1.7% and 1.8% from tech.hi52/lo52 ($393.19/$312.53); tech is used throughout for internal consistency.

Vendor composite rating — accepted. The rating block gives B+ / 3 overall, with 4/5 on discounted cash flow, return on equity and return on assets, 3/5 on debt-to-equity and 2/5 on price-to-earnings and price-to-book. The debt-to-equity sub-score is computed on the lease-inflated figure of finding 1 and is therefore one notch too harsh; nothing else here is objectionable, and none of it adds anything this dive has not derived directly.

Non-equity tripwire — checked and passed. GD is common stock, par value $1.00 per share, NYSE-listed, per the 8-K cover pages and the rescission-offer disclosure. Price of $385.76 is not par-like; beta is 0.338; the dividend is a variable declared amount; volume was 1.08M shares (~$418M of turnover); the 52-week band of $312.53 to $393.19 is a 26% range. This is common equity. Note the rescission offer covering up to 1,010,925 shares of this same class, disclosed 2026-08-04 (Section 3).

8. Technicals and insiders

• Price $385.76. −1.9% from the 52-week high of $393.19; +23.4% above the 52-week low of $312.53. Position within the annual range: 91st percentile.
• Both moving averages are well below. +6.7% above a 50-day average of $361.54; +10.2% above a 200-day average of $349.97.
• RSI 69.1 — the highest reading in this batch and at the conventional overbought threshold. MACD +6.89, strongly positive.
• Maximum drawdown from peak over the trailing year: −1.90%, verified consistent with the return series. Combined with a 0.338 beta this is the second-lowest-volatility security in this batch.
• Relative performance is good short-term and exactly average long-term. 3-month +10.5% against SPY +7.6% and QQQ +7.7%; 6-month +11.4% against SPY +11.1%; 12-month +24.1% against SPY +24.3% and QQQ +30.8%. Twelve-month performance is level with the index to within 0.2 points, which is the single most direct refutation of the knowledge base's claim that the primes are "underperforming despite the conflict."
• Sentiment: 19 buy, 15 hold, 1 sell across 35 analysts; consensus target $420.67 (+9.0%), median $420 (+8.9%), high $465 (+20.5%), low $384 — 0.5% BELOW spot. This is the only Buy consensus in this batch. The lowest target sitting essentially at the market price means no covering analyst sees meaningful downside — which we treat as a caution rather than a comfort.

Today's move and the entry

GD closed 2026-08-04 at $385.76, up 0.87% or $3.33 from $382.43, on 1.08M shares. The 8-K filed the same day concerns the rescission offer (Section 3), not results.

The honest read: there is no discount here. 1.9% from the 52-week high, RSI 69, 10.2% above the 200-day average, and a base case 8.9% away that matches the street's to two decimal places. The pre-registered upgrade level in Section 9 is $345 — 20x the FY2027 consensus, roughly the 200-day average, and a level this security traded at within the last six months.

Insiders — eight transactions, and every one is a plan mechanic

The reading: nothing.

Five of the eight belong to one non-executive director on a single day and constitute a classic exercise-and-partial-sell: options struck at $165.47 and $191.71 exercised for 5,480 shares, of which 3,210 were sold at $365.00 for approximately $1.17 million, leaving him with 10,643 shares — a position roughly triple what he sold. Exercising a long-dated option before expiry and selling enough to cover the strike and tax is the least informative transaction type there is.

Three are identical 104-share director awards on the same date at the same price. Board compensation. No signal.

What the file does not contain: any transaction by Phebe Novakovic, by the chief financial officer, or by any executive officer of any kind — and not one open-market purchase. On a stock 1.9% below its 52-week high, the absence of buying is unsurprising and carries no weight; the absence of executive-officer selling is mildly positive and carries very little.

9. Verdict, kill-criteria and flip conditions

Watch.

The operating case is the best in this batch and we want that recorded before the reservations. Total backlog of $136.5 billion, up $5.7 billion in a quarter, on a defense book-to-bill of 1.4 to 1 in the quarter and 1.8 to 1 across the half, driven by Virginia-class submarine and combat-vehicle awards. Total estimated contract value of $186.9 billion, 3.6 times revenue. Aerospace revenue +15.1% with margin +130 basis points to 14.5%. First-half operating cash flow tripling to $4,035 million and net debt falling from $5,680 million to $3,183 million in six months. Five consecutive earnings beats averaging 6.1%, on a GAAP diluted basis with no adjusting items whatsoever. A 0.338 beta, a 1.90% maximum twelve-month drawdown, and the only Buy consensus in this batch.

Why it is a Watch, stated as four specific reservations.

First, we have no variant perception and the coincidence is almost exact. Our base of $420 is 0.2% below the street's $420.67 and equals the median target. We are not seeing anything the market is missing.

Second, the arithmetic does not clear the bar. Base +8.9%, asymmetry 1.43:1, shareholder yield 1.5% — the lowest capital return in this batch, on a company that repurchased $319 million in a half-year while its share count ROSE 0.75%. Total expected return of roughly 10% requires the multiple to hold at 22-23x.

Third, the entry offers nothing. 1.9% below the 52-week high, RSI 69.1, 10.2% above the 200-day average, with the lowest analyst target in the file sitting 0.5% BELOW the market price.

Fourth, the knowledge base is negative at the name level and the long-run question it raises is genuine. Both claims naming General Dynamics are bearish. One of them — jim_bianco, April 2026 — has been falsified on its testable half and remains open on its structural half: whether cheap-drone warfare makes incumbent platform primes "the past." A $136.5 billion backlog answers that for five years. It does not answer it for fifteen, and 20.9x FY2027 earnings is a fifteen-year price.

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

• A price near $345, which is 18.7x the FY2027 consensus of $18.476 and roughly the 200-day average of $349.97. At that level the base case is +21.7%, the asymmetry improves past 2:1, and nothing needs to change at the company. This is the most likely route and it requires a 10.6% pullback — a move this stock made within the last six months.
• Marine Systems operating margin above 8.0% for two consecutive quarters, from 7.3%. This is the single most valuable operational improvement available: Marine is 33.1% of revenue and every point of margin is roughly $180 million of annual operating income, about 3.5% of the total.
• A third consecutive quarter of defense book-to-bill above 1.2, which would establish that the 1.8-to-1 half-year figure is a trend rather than two large awards.
• A buyback that actually shrinks the share count — basic weighted average shares below 268 million, from 270.2 million.
• A name-level knowledge-base claim that is not bearish. At present there are two and both are.

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

• Defense book-to-bill below 1.0 for two consecutive quarters, meaning the backlog has begun to shrink.
• Marine Systems margin below 7.0% on a disclosed programme cost overrun — the classic shipbuilding failure mode, and one the company's own risk language about "cost and schedule overruns" anticipates.
• Aerospace margin below 13.0%, which would signal a business-jet down-cycle in the segment carrying the margin expansion the consensus needs.
• A new-entrant prime taking a programme of record in a General Dynamics category, which would convert the knowledge base's structural bear claim from a thesis into an event.
• An appropriations shortfall against the unfunded portion of backlog, which the 10-Q explicitly warns is "not guaranteed."
• Any adverse development in the rescission offer disclosed 2026-08-04 beyond the 1,010,925 shares already covered.

Where GD fits in the Synthos Framework Portfolio. No position today. It goes on the watch list at $345 (18.7x FY2027 consensus, ~the 200-day average) for a 2% initial sleeve in the industrials/defense bucket, with a 3% target on Marine Systems margin above 8%. Sizing note: this is a name where the business quality is not in question and the price is doing all the work. The mistake would be paying 22.8x at RSI 69 for a consensus view and then having no capacity to add if the drone-disruption argument produces a 20% de-rating that the backlog says is unjustified. Logged as a tracked Synthos call (Watch) as of 2026-08-04 at $385.76, with the fair-value anchors, the kill criteria and the $345 upgrade trigger all gradeable.

Single biggest risk: that the incumbent-prime model is the wrong asset class for the next decade. General Dynamics has $136.5 billion of contracted work, a book-to-bill of 1.8 to 1, and one of the two shipyards in the United States capable of building a nuclear submarine. Against that, two independent voices in our store argue the category is impaired — luke_gromen that "cheap drones/missiles rendered trillions of carrier-based defense spending obsolete", and jim_bianco, naming General Dynamics explicitly, that "the primes are the past" — while a third reports that the Department is now "actively courting" new-entrant competitors. The evidence inside the company is mixed in exactly the way those claims predict: Marine Systems, building submarines that no drone replaces, grew 15.3% in the half; Combat Systems, building armoured vehicles, grew 0.3% with margin declining. The backlog is a five-year answer to a fifteen-year question, and at 20.9 times forward earnings the security is not being paid to bear the difference.

Provenance & disclosures

• Traceability: 2 name-level knowledge-base claims name General Dynamics out of 51,928 distilled claims, and BOTH are BEARISH (raw hits 46, surviving a case-sensitive entity re-run 2, entity matches 2, name-level 2, discarded 44; breadth 2, net conviction negative-low). The case-sensitive entity sweep on GD, General Dynamics, Gulfstream and Electric Boat was run FIRST because "GD" is a collision-prone two-letter ticker. The claims: jim_bianco, 2026-04-22, bearish, conviction 66, skill 0.8, speaker Jim Bianco, entities RTX/GD/BA — "Raytheon, General Dynamics and Boeing are the past; cheap-drone warfare demands new tech-oriented defense, so the primes are correcting and underperforming despite the conflict — a bear's trend" — whose testable half is falsified (GD +10.5% over three months against SPY +7.6%, +24.1% over twelve against SPY +24.3%) and whose structural half is the most serious long-run argument against this security in our possession; and lex_fridman, 2024-10-23, bearish, conviction 60, entities Lockheed Martin and General Dynamics — "Trillion-dollar Pentagon budget with revolving-door contractors and un-auditable spending implies massive fraud, waste and cost overruns" — a claim about a procurement system rather than a security, ungradeable on any horizon, weighted at essentially zero. The 44 free-text discards form a genuinely two-sided defense lane across 23 channels — bullish and pointing at Electric Boat: all_in 2025-07-23 (75, "The US must reshore submarines, ships, munitions... China outproduces the US ~200x in shipbuilding"), forward_guidance 2025-05-16 (68, "US naval industrial capacity is stagnant at ~1.3% of defense budget"), dwarkesh 2025-09-05 (65, "surface fleets are useless in high-stakes war — submarines/commerce-raiding are the fit"), no_priors 2025-08-28 (75, record $2.7T global defense spending), brent_johnson 2025-09-21 (75); bearish and pointing away: luke_gromen 2025-07-29 (78, "Cheap drones/missiles rendered trillions of carrier-based defense spending obsolete"), no_priors 2026-01-15 (80, on the new-entrant primes being "actively courted"), luke_gromen 2026-03-05 (55, rare-earth constraint). Channel distribution: lex_fridman 5, niall_ferguson 4, forward_guidance 3, dwarkesh 3, jacob_shapiro 3 and eighteen others — no channel exceeds 11% of the lane, the most diversified in this batch, so no concentration sensitivity test is triggered. ONE hit carries the eli_lilly_ceo slug and is EXCLUDED on that ground alone, that slug being the store's confirmed cross-company attribution defect. No claim in the entire lane carries a management speaker_role for this registrant, and four carry named independent speakers. All quotes are verbatim from the stored claim text.
• Data as-of: all backlog, segment, cash-flow, balance-sheet and share-count figures through 2026-07-05, from the 10-Q filed 2026-07-29 · the rescission offer from the 8-K filed 2026-08-04 · annual figures for FY2025 and earlier from the vendor payload · estimates 2026-08-04 · prices 2026-08-04, quote timestamp 1785873603 = 2026-08-04T20:00:03Z ($385.76, +0.87%; 50-DMA $361.54; 200-DMA $349.97; RSI 69.1; MACD +6.89) · knowledge-base claims 2026-08-04. General Dynamics' fiscal year ends on the Sunday nearest 31 December; the second quarter of 2026 ended 5 July 2026 and the second quarter of 2025 ended 29 June 2025, so year-on-year quarterly comparisons span periods of slightly different length. All figures come from the Synthos vendor data file for GD or from the SEC filings in the GD archive; no figure comes from memory, recall or external retrieval.
• Filing archive contents: 10-K filed 2026-01-30 (fiscal 2025); 10-Q filed 2026-04-29 (March 2026 quarter); 10-Q filed 2026-07-29 (June 2026 quarter — the source of the backlog, book-to-bill, segment, cash-flow and balance-sheet figures used throughout); 8-K filed 2026-05-11 (annual meeting voting results, Item 5.07); 8-K filed 2026-07-29 (Item 2.02, second-quarter results — the press release is furnished as Exhibit 99.1 and is NOT in the extracted text, so no figure in this dive comes from it); 8-K filed 2026-08-04 (Item 8.01 — the rescission offer covering up to 1,010,925 shares, under an automatic shelf Form S-3 filed 2026-07-30, with a Gibson Dunn validity opinion). All carry preserved [TABLE] statement data. Every document appears twice in the archive under two filenames — a pull-script artefact, recorded so six documents are not mistaken for twelve. The 8-K corporate-action scan was run: NO acquisition, divestiture, merger or spin-off appears anywhere. The 10-Q's goodwill roll-forward shows acquisitions of $1 million. Despite the brief's flag that GD is M&A-active, this archive shows a company that has made no material acquisition.
• Where the filings contradicted or corrected the vendor (detailed in Section 7): totalDebt of $9,789M, which includes $1,776M of capitalLeaseObligations against the filing's pure debt of $8,013M — a 22.2% overstatement, the largest lease-inclusion instance in this batch, carrying netDebt to $7,456M against a filing-based $5,680M (31.3% over) and invalidating netDebtToEBITDATTM (0.798x against ~0.49x on current figures), debtToEquityRatioTTM (0.353x against ~0.29x) and interestDebtPerShareTTM; the $2,310M disagreement between enterpriseValueTTM (implying $5,146M of net debt) and bal_a (saying $7,456M) — an ETN/ABT-class internal contradiction with an inverted resolution, in which the EV field is within 9.4% of the filing and the balance block is 31.3% off; seg_geo FY2025, which reports both "North America" $45,810M and "UNITED STATES" $43,314M — the second being a subset of the first — so that a naive sum of all six lines gives $97,112M against $52,550M of reported revenue, an 85% inflation, with the US split appearing in FY2025 only and therefore not comparable to any prior year; and est.ebitdaAvg/ebitAvg, rejected for a fixed-ratio signature (exactly 12.637% and 10.687% of revenue in every forward year FY2026–FY2030) and for overstating FY2025 EBITDA by 7.7% against the reported figure. Where vendor and filing AGREED — recorded, and GD passes more checks than any name in this batch: seg_prod is EXACT — the FY2025 entry (Aerospace $13,110M, Combat Systems $9,246M, Marine Systems $16,723M, Technologies $13,471M) sums to $52,550M, precisely the reported revenue, with the four correctly-named reportable segments and a sixteen-year history; the share count matches to 0.15% (market capitalisation ÷ price implies 270,563,000 against the 10-Q's 270,156,000 basic weighted average); inc_q Q2 2026 revenue of $14,094M equals the 10-Q's segment total exactly; earn_cal epsActual of $4.24 equals net earnings of $1,160M ÷ 273,522 thousand diluted shares — and GD reports NO adjusted EPS at all, so there is no basis question; effectiveTaxRateTTM of 17.62% reconciles to FY2025's $893M on $5,103M; and tech.max_dd_from_peak of −1.90%, which carries the ADP-class signature of equalling pct_from_hi, was CHECKED against the return series and found CORRECT — ret_12m of +24.13% implies approximately $310.75 a year ago, within 0.6% of lo52 of $312.53, so the low preceded the high. (quote.yearHigh/yearLow of $400.00/$306.77 sit 1.7% and 1.8% from tech.hi52/lo52; tech is used throughout.)
• Basis note: General Dynamics reports GAAP results only — there is no adjusted EPS, no non-GAAP bridge and no adjusting items in the June quarter or the first half. All estimates and all actuals in this dive are on GAAP diluted EPS, and earn_cal reconciles exactly to net earnings divided by diluted weighted average shares. Trailing diluted EPS of $16.39 is the sum of the four quarterly earn_cal actuals ($3.88 + $4.17 + $4.10 + $4.24); the vendor's inc_a EPS of $15.64 for FY2025 is the BASIC figure and is not used for any multiple. Operating cash flow for the first half of 2026 ($4,035M) and 2025 ($1,450M) is DERIVED from the 10-Q's disclosed net change in cash less investing, financing and discontinued-operations flows, because the direct line falls outside the extracted text; the derivation is arithmetic and is labelled as such at every use.
• Estimate coverage: 13 analysts on FY2026 EPS, 16 on FY2027 — the anchor for all three fair values — with an FY2027 range of $17.633 to $19.398, a 10.0% spread; 16 on FY2028 and 14 each on FY2029 and FY2030 — the deepest forward coverage in this batch by a wide margin. FY2029 and FY2030 are used only as directional cross-checks and no fair value rests on them. Revenue coverage: 12, 17, 19, 10 and 16 analysts respectively.
• Peer note: the vendor peer set — ADP, HWM, LHX, LMT, MMM, NOC, PH, TDG, TT and UPS — contains five genuine aerospace-and-defense comparables (LMT, NOC, LHX, HWM, TDG) and five that are not (a payroll processor, a diversified industrial, a fluid-power manufacturer, an HVAC company and a parcel carrier). The payload supplies only market capitalisations and no multiples, so no peer-multiple comparison is drawn. ADP and TT are separately covered in this same batch.
• Fair-value caveat: the $320 / $420 / $480 anchors are multiples of the FY2027 consensus diluted EPS distribution — 18.1x the low of $17.633, 22.7x the mean of $18.476, and 24.7x the high of $19.398 — each cross-checked against FY2028E (16.1x, 21.1x, 24.1x). Stated arithmetic, not a discounted cash flow. The base is sensitivity-disclosed: 20x gives $370, below spot; 24x gives $443. The base case assumes the forward multiple essentially HOLDS, which is a deliberate refusal to underwrite either the cheap-drone disruption case or a defense-budget re-rating; the return is therefore almost entirely earnings growth, with a shareholder yield of only about 1.5% adding very little.
• Capital-return note: first-half 2026 share repurchases were $319 million against $600 million a year earlier, and basic weighted average shares ROSE from 268,138 thousand to 270,156 thousand year on year. The 10-Q describes buybacks as "opportunistic share repurchases primarily to address dilution" — which is an accurate description of the intent and an inaccurate description of the result. Dividends of $834 million in the half at a trailing $6.18 per share give a 1.60% yield, with the June quarter declaring $432 million (approximately $1.60 a share, $6.40 annualised). Total shareholder yield of roughly 1.5% net of dilution is the lowest in this batch.
• Half-weight management framing: the capital-deployment language and the characterisation of unfunded backlog as "amounts we believe are likely to be funded" are company framing and are half-weighted under house rules. The backlog quantities, book-to-bill ratios, segment revenue and margins, and the balance-sheet and cash-flow figures are quantified filing disclosures and are treated as facts.
• Timing: second-quarter 2026 results were released 2026-07-29, six days before this dive, and beat consensus diluted EPS by 7.1% ($4.24 against $3.96) — the fifth consecutive beat, averaging 6.1%. The next print is 2026-10-28, 85 days away (consensus diluted EPS $4.14, revenue $13,680M). An 8-K dated 2026-08-04 — the date of this dive — discloses the rescission offer covering up to 1,010,925 shares. The most recent insider filings are dated 2026-06-18 and 2026-06-22.
• 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.