← All companies

MRVLMarvell Technology, Inc.

Industry AI semis Updated 24 Sep 2026

The idea

The base case is below the price even on management's 60%/60% data-centre growth guides; the bull ($259) is roughly today's price. SBC and the NVIDIA preferred (21.8m shares at ~$92) dilute.

Events to watch

  • Data-centre growth vs the 60% FY27/FY28 guides
  • Q3 FY27 non-GAAP EPS $1.05-1.15 guide
  • custom-silicon program wins
How we got the target

Dossier: base 30x FY28 model EPS $6.76, built from management's own guides. Time frame: FY2028 (Jan 2028).

Target: from the research notes. Prices: market data get_latest_stock_price, close of 2026-09-22 (fetched 2026-09-23)

What would prove it wrong

6 warning signs
  1. Distributor A's share of revenue keeps climbing.

    Details

    34% → 44% of net revenue in a year, 45% for the first half. Above ~50% in a single quarter and Marvell is a single-channel company whatever the end-customer count behind it; below ~35% and the concentration thesis is wrong.

  2. Data-center revenue growth undershoots the 60% management guided for FY28.

    Details

    Management guided 60% twice — FY27 and FY28. A cut to that FY28 figure is the loudest available signal, as it was at NVIDIA.

  3. Non-GAAP gross margin breaks below the 57.5% floor of the Q3 FY27 guide.

    Details

    Custom-silicon mix dilutes margin structurally; the guide already steps down from FY26's ~59-60% run rate.

  4. Capacity prepayments rise beyond the ~$1bn named for FY27 without matching revenue.

    Details

    Cash out now against capacity later is the same exposure the cluster pass flagged at NVIDIA.

  5. NVIDIA converts, sells, or adds to the Series A preferred.

    Details

    Any of the three re-prices the partnership; conversion alone adds ~21.8m shares (~2.5% of the common) at ~$91.84.

  6. The XPU-program question gets answered by a filing, or it gets worse.

    Details

    Marvell's position, exclusivity and continuity in its XPU programs was dodged in six quarters (dodge log). Kill if the 10-Q significant-customer table drops Customer A (16% of revenue, flat for a year) below 10%, or if the 2026-10-06 Investor Day's custom breakdown cuts the FY28 "more than double" custom guide or shows the new tier-1 program shared with a second vendor.

How far along

7 of 10 steps done · 3 done in a group review

Next step

price the NVIDIA preferred as a governance fact, not a funding one. NVIDIA paid $2.0bn on 2026-03-31 for stock convertible into ~21.8m Marvell shares at ~$91.84 while the stock trades at $244.25 — an ~$3.3bn unrealised gain for the counterparty, sitting alongside a partnership that routes Marvell's XPUs into NVIDIA's ecosystem and a 10-K that lists NVIDIA among the companies that "compete directly with our businesses". Whether that shareholding constrains Marvell's freedom to sell against NVIDIA is the question the filings do not answer and the next transcript read might.

Needs your decision

S6 product test / expert call, S7 conviction. Open (routine): the efts.sec.gov "restatement" full-text sweep; the proxy's bonus metrics; re-run S7's table after the 2026-10-06 Investor Day resets the long-term model, with the earn-out and warrant shares in the count.

Failed checks

none. S4/S7 are not closed by design (cluster stages, carried forward). S8 passes as reliable, but read its two ⚠️ lines: the XPU-program exclusivity/continuity question has been dodged in six of eight calls (dodge log below) — a thread to pull, with a kill criterion — and the CFO left two weeks after a GAAP EPS miss. S3 carries 3 earnings-quality flags (SBC, add-backs, DSO).

Who runs it

The Front Office

Step by step

  1. S0How it does as AI gets cheapDone in a group review

    carry-forward (cluster: AI-capex complex — cluster pass, 2026-09-20)

    cluster verdict is.

    ⚠️ SPLIT; MRVL sits on the "seller of the shortage" side that passes. Owner checks dropped 2026-09-21 (README §Stage 0 override).

  2. S1Where the idea came fromDone

    watchlist (owner, Apple Stocks, 2026-09-20); no Tier 0 promotion.

  3. S2Quick deal-breaker checkDone

    FY26 rev $8,195m +42%, GAAP GM 51.0%, GAAP op margin 16.1%, OCF $1,750m, capex $354m, FCF ~$1,396m. Net debt ~$1.1bn against $3.9bn cash. Survives.

    ⚠️ FY26 net income $2,670m is not earnings power — $1,901m of it landed in one quarter (Q3 FY26) on the automotive-ethernet divestiture. · capital allocation: neutral (dossier-2026-09-24.md). ROIC −2.4%, 1.1%, −3.0%, −4.2%, then 7.3% in FY26 (12.5% before acquired-intangible amortisation); FY22→FY26 NOPAT −$348m → +$1,191m while invested capital fell $19.7bn → $16.2bn, so the turn came from using the Inphi-era capital harder; newest increment (FY26 → H1 FY27 annualised) +$248m on +$3.4bn, 7.2%. Buybacks 42.9m shares at ~$80 average vs $260.90 today; they offset SBC only in FY25–FY26 (count flat 847.8m → 848.1m, 2022-03 → 2025-11), then acquisitions took it to 876.9m. No goodwill impairment; $711.8m of FY25 restructuring/intangible impairment. Automotive Ethernet sold for a $1,830m gain.

    ⚠️ ~105m potential shares contracted (~12% of the common): NVIDIA preferred ~21.8m, Celestial earn-out 24.4m, customer (Google) warrant 59.0m at $206.58 vesting on revenue milestones to FY33.

  4. S3Reading the company's reportsDone

    (gate-minimum)

    FY26 10-K risk factors + MD&A, Q2 FY27 10-Q MD&A and notes, DEF 14A ownership, EDGAR form census. all eight archived transcripts (FY25 Q3 → FY27 Q2) read Q&A-first on 2026-09-24, nine earnings releases, the Item 5.02 set via efts.sec.gov.

    Full notes

    No 4.01/4.02 8-K since 2024-08 (index read; the efts "restatement" full-text sweep is still not run). · EQ: 3 flags — add-backs +65% vs revenue +32% in H1 FY27, with a new earn-out mark-to-market add-back ($302.7m) and "restructuring" in seven years of seven. SBC back to 11.9% of revenue in Q2 FY27 and 55% of H1 FCF (FCF after SBC 44% of FCF). DSO 51.5 → 73.7 days in six quarters despite $412m of receivables factored in H1. Capacity prepayments $278.8m → $487.0m are guided cash, not a flag. Cash above GAAP earnings (negative accruals).

  5. S4Rivals and suppliersDone in a group review
  6. S5Who owns the stockDone

    proxy 2026-01-31 table, 13F Q2-26 by manager, 13D/G, Form 4 and Form 144 cadence, FINRA short interest, buyback authority and the live S-3ASR shelf.

    ⚠️ the Q2-26 13F quarter is incompletely ingested — FMR, both Vanguard entities, Geode, Franklin and Morgan Stanley have no filing in it yet, so no "exit" may be read from their absence.

  7. S6What workers and customers sayDone

    carry-forward (cluster: AI-capex complex — cluster pass 2: S6 for 41 names, S0/S4/S7 for three, 2026-09-21)

    closed there for all 41 members. Product test / expert call open OPEN (user)

  8. S7The case for it, and pricesDone in a group review

    carry-forward (cluster: AI-capex complex — cluster pass, 2026-09-20)

    conviction open OPEN (user) · valuation (bear/base/bull; moved from S8 2026-09-23): ~59x model FY27 and ~36x model FY28 non-GAAP EPS at $244.25 (2026-09-18), both built from management's own guides.

    Full notes

    S8 = reliable (2026-09-24), so no haircut; FY28 guide is now ~$18bn vs the model's ~$17.9bn. The model's ~915m shares omit the earn-out and warrant (~83m if fully vested) — for the next S7 re-run.

  9. S8Do the bosses keep their word?Done

    **20 rows: 3 beat, 12 met, 4 missed, 1 withdrawn

    reliable; current team made 18 of 20 (CFO changed 2026-06-15: Meintjes → Durn.

    Full notes

    Meintjes's own 2 rows departed).** All eight quarterly revenue / non-GAAP GM / non-GAAP EPS guides landed inside the range just above the midpoint FY27 revenue raised four times (~$9.5bn → ~$12bn) Misses: two GAAP EPS (Q3 FY25 restructuring, Q1 FY27 earn-out mark), the 38–40% op-margin target a year late, the 2025 Investor Day postponed Withdrawn: the FY26 $2.5bn AI-revenue target, never reported against.

    ⚠️ CFO resigned 2026-06-10, two weeks after the Q1 FY27 GAAP miss (8-K: no disagreement; successor was the audit chair).

    ⚠️ Dodge log: XPU-program position/exclusivity dodged in six quarters, once on a date management itself had given.

  10. S9What to watch nextDone

    trigger written. Tier 0 EDGAR sweep covers the feed (cik 1835632)

Sources and notes

Sources 7 documents
Short nameDocumentWhere it came fromRead on
10K-FY26Marvell 10-K for FY ended 2026-01-31, filed 2026-03-11https://www.sec.gov/Archives/edgar/data/1835632/000183563226000011/mrvl-20260131.htm2026-09-21
10Q-Q2FY27Marvell 10-Q for the quarter ended 2026-08-01, filed 2026-08-28https://www.sec.gov/Archives/edgar/data/1835632/000183563226000025/mrvl-20260801.htm2026-09-21
DEF14A-26Marvell DEF 14A filed 2026-05-13 (ownership table as of 2026-01-31)https://www.sec.gov/Archives/edgar/data/1835632/000110465926060253/tm261486-1_def14a.htm2026-09-21
13F-Q2-26Institutional holders by manager, 13F report date 2026-06-30 (ingestion incomplete — see note)python ledger/research/tools/threespread.py holders MRVL2026-09-21
GUIDE-Q2FY27Marvell Q2 FY2027 results 8-K (2026-08-27) and the earnings call of the same date, as guidance rowsEquibles GetGuidance MRVL; the 8-K rows link https://www.sec.gov/Archives/edgar/data/0001835632/0001835632-26-000022.txt2026-09-21
8K-2026-06-11Marvell 8-K filed 2026-06-11, Item 5.02 — CFO resignation and appointment (acc 0001193125-26-267688)https://www.sec.gov/Archives/edgar/data/1835632/000119312526267688/d151562d8k.htm2026-09-24
EDGAR-INDEXMarvell EDGAR filing index, forms filed since 2023-01-01 (CIK 1835632)https://data.sec.gov/submissions/CIK0001835632.json2026-09-21
Links to other companies 16 links
holds 3issuer of 3officer of 3acquired 2technology partner 1channel concentration 1customer of 1supplier dependency 1context 1
FromLinkToDateWhere it says so
NVIDIA CorporationholdsMRVL2026-03-3110Q-Q2FY27 — Note 10 – Stockholders' Equity, 'Series A Convertible Preferred Stock'
NVIDIA Corporationtechnology partnerMRVL2026-08-0110Q-Q2FY27 — Item 2, MD&A, 'Overview' — paragraph on the first quarter of fiscal 2027
MRVLissuer ofMarvell Series A Convertible Preferred Stock, 2.0m shares, $2.0bn, issued to NVIDIA 2026-03-312026-08-0110Q-Q2FY27 — Part II, Item 1A, 'The rights, preferences and privileges of our preferred stock may adversely affect holders of our common stock...'
Distributor at 44% of Q2 FY27 net revenue (unnamed in the filings)channel concentrationMRVL2026-08-0110Q-Q2FY27 — Item 2, MD&A, significant-customer table: 'Distributor: Distributor A 44% 34% 45% 35%'
Direct customer at 16% of Q2 FY27 net revenue (unnamed in the filings)customer ofMRVL2026-08-0110Q-Q2FY27 — Item 2, MD&A, significant-customer table: 'Direct Customer: Customer A 16% 16% 16% 16%'
Distributor at 44% of Q2 FY27 net revenue (unnamed in the filings)supplier dependencyMRVL2026-01-3110K-FY26 — Note 2 – Significant Accounting Policies, 'Concentration of Credit Risk and Significant Customers'
MRVLacquiredCelestial AI, Inc.2026-02-0210Q-Q2FY27 — Note 4 – Business Combinations, 'Celestial AI, Inc.'
MRVLacquiredXConn Technologies Holdings, Ltd.2026-02-1010Q-Q2FY27 — Note 4 – Business Combinations, 'XConn Technologies Holdings, Ltd.'
MRVLissuer ofMarvell $1.0bn 5.300% senior notes due 2036, issued 2026-04-152026-08-0110Q-Q2FY27 — Item 2, MD&A, 'Liquidity and Capital Resources' — debt paragraph
MRVLissuer ofMarvell automatic shelf registration statement on Form S-3ASR, filed 2025-03-122025-03-12EDGAR-INDEX — filings.recent, form 'S-3ASR', filingDate 2025-03-12, primaryDocument d893565ds3asr.htm
FMR LLCholdsMRVL2026-01-31DEF14A-26 — Security Ownership of Certain Beneficial Owners and Management, 5% Stockholders, row 'FMR LLC'
BlackRock, Inc.holdsMRVL2026-06-3013F-Q2-26 — row 'BlackRock, Inc.', filing https://www.sec.gov/Archives/edgar/data/2012383/000201238326003238/
Matthew J. Murphyofficer ofMRVL2026-01-31DEF14A-26 — Security Ownership table, rows 'Matthew J. Murphy, Chairman and Chief Executive Officer (7)' and 'All current directors and executive officers as a group (13 persons) (9)'
MRVLcontextDistributor at 44% of Q2 FY27 net revenue (unnamed in the filings)2026-08-27GUIDE-Q2FY27 — rows 'overall Marvell revenue / fiscal 2027 / $12B', 'data center revenue growth / fiscal 2027 / 60%', 'data center revenue growth / fiscal 2028 / 60%', 'capacity prepayments to suppliers / fiscal 2027 / $1B'
Willem Meintjesofficer ofMRVL2026-06-158K-2026-06-11 — Item 5.02, 'Resignation of Chief Financial Officer'
Daniel Durnofficer ofMRVL2026-06-158K-2026-06-11 — Item 5.02, 'Appointment of New Chief Financial Officer and Departure of Director'
History 3 entries
  • 2026-09-21 — first pass, slot names-ai-capex-2, model-drafted unattended by the cloud research routine. S1, S2, S3 (gate-minimum), S5 and S9 closed; S0/S4/S7 carried forward from the 2026-09-20 cluster file and S6 from the 2026-09-21 one; S8 modelled and left open. Evidence and access limits in dossier-2026-09-21.md. Nothing here is human-verified.
  • 2026-09-23 — stage redefinition, owner decision (WATCHLIST_PLAN.md §25): S8 is now Management credibility and is not yet run for this name; the old S8 valuation figures moved to the valuation: line under S7, and any proposed buy price or position size was removed (the pipeline no longer drafts either). No research was redone. S2 capital allocation and S3 earnings quality are also pending, in the same backfill slot.
  • 2026-09-24 — backfill slot backfill-2026-09-23-2, model-drafted in a research session; nothing here is human-verified. Added only the three checks of 2026-09-23 (dossier-2026-09-24.md): S8 written (20 rows, reliable, ✅; CFO change 2026-06-15 recorded with officer_of edges in relationships.yaml); S2 capital allocation (neutral; glyph unchanged, ⚠️ contracted-share line added); S3 earnings quality (3 flags; glyph ✅ (gate-minimum) — the "partial" came from no transcript read, and all eight archived calls are now read). Added the ## Dodge log, one kill criterion (XPU-program thread), updated the ⚠️/⏳ lines. S7 valuation figures unchanged (no haircut: S8 is reliable). Corrected the buyback-authority figure in the dossier ($5.1bn remaining, not $9.7bn). Zero Equibles calls; EDGAR, the transcript archive and market data (price, one ratio call).
Full notes, 24 Sep 2026 25 min read

MRVL — dossier, 2026-09-24 (backfill)

Backfill slot backfill-2026-09-23-2. Model-drafted by a research session; nothing here is human-verified. This file holds only the three checks added to the pipeline on 2026-09-23 (WATCHLIST_PLAN.md §25): S2 capital allocation, S3 earnings quality (with the eight archived transcripts read Q&A-first under §3.1) and S8 management credibility. Every other stage stands as written in dossier-2026-09-21.md, which is not edited.

Quote discipline: every string in quotation marks below was located by exact substring match in the source named beside it — the transcript archive's content string, or a whitespace-collapsed text extraction of the EDGAR document — and copied from there. EDGAR's HTML puts a space after $ ($ 412.0 million); the spacing is the filing's. Transcript speaker names are the transcript's own Name: labels at the quoted position (the archive spells Murphy three ways across the eight calls: Matt Murphy, Matthew J. Murphy, Matthew Murphy). market data transcripts carry names only; roles are from EDGAR (8-K Item 5.02 and the officer_of edges in relationships.yaml) as of each call date. Anything from background knowledge is marked [background].

Price reference for this file: $260.90 close, 2026-09-23 (market data get_latest_stock_price NASDAQ:MRVL; volume 13,638,012).

Sources

KeyDocumentLocator
CALL-FY25Q3 … CALL-FY27Q2Eight earnings calls, 2024-12-03, 2025-03-05, 2025-05-29, 2025-08-28, 2025-12-02, 2026-03-05, 2026-05-27, 2026-08-27ledger/research/_transcripts/MRVL/FY<y>-Q<q> (market data; sha256 in _transcripts/index.json)
PR-Q2FY25 … PR-Q2FY27Ex. 99.1 earnings releases (the written outlook and the non-GAAP reconciliation), nine quarterse.g. https://www.sec.gov/Archives/edgar/data/1835632/000183563226000022/q227_8kx812026ex-991.htm; accessions 0001835632-24-000140, -24-000197, -25-000051, -25-000115, -25-000187, -25-000193, -26-000006, -26-000014, -26-000022
8K-2025-05-068-K Ex. 99.1 — Investor Day postponed; Q1 FY26 revenue range narrowed to ±2%https://www.sec.gov/Archives/edgar/data/1835632/000119312525113881/d920821dex991.htm
8K-2026-06-118-K Item 5.02 — CFO Willem Meintjes resigns effective 2026-06-15; director Dan Durn appointed CFOhttps://www.sec.gov/Archives/edgar/data/1835632/000119312526267688/d151562d8k.htm
8K-5.02-setThe other Item 5.02 8-Ks in the window: 2024-09-19 (director Ford Tamer), 2025-04-14 (Raghib Hussain, President Products & Technologies, resigns), 2025-06-20 (Chief Accounting Officer change), 2025-07-17 (Koopmans, Bharathi promoted), 2025-07-23 (director appointed)efts.sec.gov full-text search, "Item 5.02", cik 0001835632, 2024-08-01 → 2026-09-24: 7 hits, same set as the submissions items field
10Q-Q2FY2710-Q for the quarter ended 2026-08-01https://www.sec.gov/Archives/edgar/data/1835632/000183563226000025/mrvl-20260801.htm
10K-FY2610-K for FY ended 2026-01-31 (Item 5 issuer purchases)https://www.sec.gov/Archives/edgar/data/1835632/000183563226000011/mrvl-20260131.htm
XBRLCompany facts, CIK 0001835632 — cash-flow, balance-sheet and share tags named inlinehttps://data.sec.gov/api/xbrl/companyfacts/CIK0001835632.json
MKT-PXClose $260.90, 2026-09-23market data get_latest_stock_price NASDAQ:MRVL

Not used: Equibles (zero calls — every fact below came from EDGAR or the archive); Alpha Vantage, Bigdata.com, FMP (not called). market data get_profitability_ratios returned only FY25–FY26 for MRVL, so the five-year return series below is computed from XBRL instead.


S2 — Capital allocation and returns

Fiscal years end late January/early February. $ millions. All from XBRL (tags: NetCashProvidedByUsedInOperatingActivities, PaymentsToAcquirePropertyPlantAndEquipment, PaymentsToAcquireBusinessesNetOfCashAcquired, PaymentsForRepurchaseOfCommonStock, PaymentsOfDividends, ProceedsFromIssuanceOfLongTermDebt/ProceedsFromIssuanceOfDebt, RepaymentsOfLongTermDebt, ProceedsFromDivestitureOfBusinessesNetOfCashDivested, ProceedsFromIssuanceOfPreferredStockAndPreferenceStock, ShareBasedCompensation, PaymentsRelatedToTaxWithholdingForShareBasedCompensation, StockRepurchasedDuringPeriodShares, TreasuryStockAcquiredAverageCostPerShare).

FY22FY23FY24FY25FY26H1 FY27
Operating cash flow8191,2891,3711,6811,7511,244
Capex(169)(206)(336)(285)(354)(282)
FCF6501,0831,0341,3971,396962
Acquisitions, cash net(3,555) (Inphi, Innovium [background])(112)0(10)0(1,271) (Celestial AI, XConn)
Divestiture proceeds——00+2,479 (automotive Ethernet)—
Buybacks0(115)(150)(725)(2,040) incl. $1.0bn ASR(400)
Dividends(191)(204)(207)(208)(205)(108)
Debt raised / (repaid)+3,896 / (527)+200 / (266)+1,295 / (1,623)0 / (109)+1,199 / (791)+999 / (500)
Preferred issued—————+2,000 (NVIDIA)
SBC461552610597591534
SBC tax withholding (net-settlement)(306)(228)(224)(275)(241)n/a
Shares repurchased (m)02.32.59.026.62.5
Average price paid—$49.93$61.51$80.47$76.46~$160 (cash ÷ shares)

Did buybacks offset SBC? Only in the last two years. FY22–FY24 repurchases of $265m sat against $1,623m of SBC. FY25–FY26 repurchased $2,765m against $1,188m of SBC, and the cover-page count shows the result: 847.8m shares (2022-03-03) → 848.1m (2025-11-26) — flat across three and a half years (dei EntityCommonStockSharesOutstanding). Then it stepped up again on acquisitions: 874.3m (2026-03-04) → 876.9m (2026-08-21), and the diluted weighted count went 856.2m (Q4 FY26) → 921.2m (Q2 FY27) (PR-Q2FY27 reconciliation) — "The increase from the prior quarter reflects the full impact of shares issued for the Celestial AI and XConn acquisitions as well as the shares issued as part of the NVIDIA investment." (CALL-FY27Q1, Willem A. Meintjes). H1 FY27 repurchased $400m against $534m of SBC — no offset. The CFO's Q1 FY27 line — "In parallel, we plan to continue to repurchase shares to manage dilution." (CALL-FY27Q1, Willem A. Meintjes) — is a direction, not a number, and the count rose 7.6% diluted in two quarters.

More issuance is contracted, not yet counted. Two instruments sit outside every share figure above:

  • Celestial earn-out: "we may be required to issue 24.4 million additional shares of our common stock if Celestial achieves specified cumulative revenue milestones through the end of fiscal 2029" (10Q-Q2FY27, Part II Item 1A).
  • A customer warrant signed after the quarter: "issued a warrant to a customer to purchase an aggregate of up to 59.0 million of the Company’s common stock at an exercise price of $ 206.58 per share" (10Q-Q2FY27, Note 15). The filing does not name the customer; on the call two analysts called it Google (Tom O'Malley opened with "the warrants with Google") and Matthew Murphy answered without correcting them, later calling Marvell "a part of the ecosystem, the TPU ecosystem" (CALL-FY27Q2).

Preferred ~21.8m + earn-out 24.4m + warrant 59.0m = ~105m potential shares, ~12% of the 876.9m common, all tied to revenue that management is guiding up. That is the price of the growth, and it has to be in the per-share model (see the last section of this file).

Returns on invested capital. Invested capital = stockholders' equity + debt carrying amount − cash (XBRL StockholdersEquity, DebtInstrumentCarryingAmount, CashAndCashEquivalentsAtCarryingValue, year-end). NOPAT = GAAP operating income × (1 − 10%), the non-GAAP tax rate management used for FY26 (PR-Q4FY26 note d); losses untaxed. Average of opening and closing capital. The second column adds back amortisation of acquired intangibles and restructuring/impairment — what the acquired franchise earns before paying for its own purchase accounting.

Invested capital (end)NOPATROICNOPAT ex-amortisationROIC ex-amortisation
FY2219,677(348)−2.4%5974.2%
FY2319,2492141.1%1,2126.2%
FY2418,080(568)−3.0%5953.2%
FY2516,569(720)−4.2%9405.4%
FY2616,1701,1917.3%2,05312.5%
H1 FY27, annualised19,5991,4388.0%2,24112.5%
  • Four of the five years earned below any cost of capital on the ~$19bn Marvell paid to become a data-center company (goodwill $5.3bn → $11.5bn in FY22). FY26 is the first year the GAAP return clears 7%.
  • Incremental ROIC, FY22 → FY26: NOPAT −$348m → +$1,191m (+$1,539m) on invested capital $19,677m → $16,170m (−$3,507m). The ratio is not meaningful — capital shrank (amortisation ran down equity, the automotive-Ethernet sale returned $2.5bn) while profit rose. Read it as: the FY26 turn came from using the Inphi-era capital harder, not from new capital.
  • The newest increment, FY26 → H1 FY27 annualised: NOPAT +$248m on invested capital +$3,429m (Celestial and XConn stock, the NVIDIA preferred, $0.5bn net new debt) — 7.2% (5.5% on the ex-amortisation basis). Early and unfair to Celestial, whose revenue management places in FY28; it is the number to re-run in a year.

What the acquisitions were worth. Goodwill impairment: zero in every year FY22–FY26 (GoodwillImpairmentLoss = 0). But FY25 carried $711.8m of restructuring and impairment (RestructuringCostsAndAssetImpairmentCharges) — impairments of acquired intangibles and purchased technology licences in the non-data-center businesses, taken when management redirected investment to data center (CALL-FY25Q3, Willem Meintjes, CFO section). The automotive-Ethernet sale booked a $1,830.4m gain on $2,478.6m of proceeds (PR-Q4FY26 reconciliation) — a business built organically and sold well.

Buybacks at what price. 42.9m shares bought FY23–H1 FY27 for $3,430m — ~$80 average against $260.90 today. The FY26 program (26.6m shares at $76.46, including the $1.0bn ASR executed in Q3 FY26) is the best-timed capital decision in the window. The pace then fell as the price rose: 1.1m shares at $176.24 in May 2026 and none in June–July (10Q-Q2FY27, Part II Item 2 table); "Our existing stock repurchase program had approximately $5.1 billion of repurchase authority remaining as of August 1, 2026." (10Q-Q2FY27, same table's note 2). Correction to dossier-2026-09-21.md, which described $9.7bn of authority against $400m used: $9.7bn was the total after the 2025-09-24 addition; $5.1bn is what remains. What the bonus is paid on is still unread (proxy compensation section not extracted — same gap as 2026-09-21).

Capital-allocation verdict: neutral. Well-timed buybacks (~$80 against $261), a good divestiture and FCF that funded both. Against that, five years of low returns on the acquired base, and a growth plan now being paid for in shares: preferred, earn-out and warrant together ~12% of the common. Not a destroyer: FY26 turned the return positive and the latest increment is too young to judge. No kill criterion added on S2.


S3 — Earnings quality

Checks in the order of SKILL.md §3. Revenue: FY24 $5,508m, FY25 $5,767m, FY26 $8,195m, H1 FY27 $5,157m (H1 FY26 $3,901m, +32%).

  1. The non-GAAP bridge. ⚠️ Flag 1.
  2. Total pre-tax add-backs (GAAP → non-GAAP net income): FY25 $2,375.7m (41% of revenue, swollen by $711.8m restructuring) → FY26 $1,523m excluding the $1,830.4m divestiture gain (19%) → H1 FY27 $1,318.1m against $801.0m in H1 FY26: +65%, twice revenue's +32% (PR-Q4FY26, PR-Q2FY27 reconciliations).
  3. A new add-back arrived in FY27: "Change in fair value of contingent consideration liability, net of forward stock purchase contract" — $302.7m in H1 FY27 ($250.7m in Q1 alone), zero before (PR-Q2FY27). It is the Celestial earn-out marked to Marvell's own share price: the higher the stock, the bigger the GAAP charge and the bigger the add-back.
  4. "Restructuring" recurs every year — FY21 $180.4m, FY22 $31.6m, FY23 $21.6m, FY24 $131.1m, FY25 $711.8m, FY26 $16.0m, H1 FY27 $5.7m (RestructuringCostsAndAssetImpairmentCharges; H1 from PR-Q2FY27). Small since FY25, but a charge present in seven years of seven is not a one-off.
  5. Amortisation of acquired intangibles is falling ($1,052.6m → $942.0m → $440.1m in the half) and is a genuine non-cash item; it is not the problem.
  6. Stock-based compensation. ⚠️ Flag 2. SBC/revenue: FY22 10.3%, FY23 9.3%, FY24 11.1%, FY25 10.4%, FY26 7.2%, H1 FY27 10.4%, Q2 FY27 11.9% ($326.2m against $153.6m a year earlier; PR-Q2FY27). Part of the jump is acquisition accounting — of the Celestial replacement awards, "$ 190.1 million, representing post-acquisition stock-based compensation expense that will be recognized as these employees provide service over the remaining vesting periods" (10Q-Q2FY27, Note 10 – Stockholders' Equity, Celestial replacement awards) — the rest is not explained in the sections read, and the Q3 FY27 outlook carries $285m of SBC in opex alone (PR-Q2FY27 outlook). Against FCF: FY26 SBC was 42% of FCF (FCF after SBC 58% of FCF, above the skill's half-line); H1 FY27 SBC $534m is 55% of FCF $962m — FCF after SBC is 44% of reported FCF, below half. This reverses the FY26 improvement management promised in December 2024 (S8 row 14).
  7. Cash conversion and accruals. GAAP net income is a poor denominator here (losses FY22–FY25; FY26 includes the $1,830m gain), so both are shown. Accruals, (NI − OCF)/average assets: FY24 −10.5%, FY25 −12.4%, FY26 +4.3% (−4.3% excluding the gain), H1 FY27 −3.6% — cash above GAAP earnings throughout, because amortisation is non-cash. Against non-GAAP net income, OCF conversion fell: 1.22× (FY25) → 0.71× (FY26) → 0.79× (H1 FY27), and FCF was 57% of FY26 non-GAAP net income. Management's explanation is working capital and capacity prepayments (next two items). Watch, not a flag on its own.
  8. Working capital. ⚠️ Flag 3. DSO (quarter-end receivables ÷ quarterly revenue × 91; XBRL AccountsReceivableNetCurrent and revenue): Q4 FY25 51.5 → Q1 FY26 54.9 → Q2 65.9 → Q3 67.8 → Q4 FY26 89.7 → Q1 FY27 70.4 → Q2 FY27 73.7 days. Four consecutive rises through FY26 while revenue grew, and Q2 FY27 is still +7.8 days year on year — measured after factoring: "Total trade accounts receivable sold under the factoring arrangement was $ 111.1 million and $ 412.0 million for the three and six months ended August 1, 2026" (10Q-Q2FY27, Note 14; $526.8m in H1 FY26). Four customers hold 72% of gross receivables and Distributor A is 44% of revenue (10Q-Q2FY27, MD&A). Inventory days (DIO) moved 96–118 with no trend.
  9. Costs moved onto the balance sheet. No capitalised-software or contract-cost line of any size (PP&E "Computer software" $147.3m gross). What is building on the balance sheet is cash paid ahead of supply: "Prepayments on supply capacity reservation agreements" $278.8m → $487.0m in six months (10Q-Q2FY27, Note 14, other non-current assets), inside the ~$1bn FY27 prepayment management guided. It is disclosed, guided and a cash item, not an expense being deferred — not a flag; it is the same exposure as the existing kill criterion on prepayments.
  10. Sector equivalents: n/a (semiconductors).

EQ: 3 flags — add-backs +65% vs revenue +32% in H1 FY27, with a new earn-out mark-to-market add-back ($302.7m) and "restructuring" in seven years of seven; SBC back to 11.9% of revenue in Q2 FY27 and 55% of H1 FCF (FCF after SBC 44% of FCF); DSO 51.5 → 73.7 days in six quarters despite $412m of receivables factored in H1.

Red-flag sweep additions (this pass): the Item 5.02 set is in 8K-5.02-set. No Item 4.01 (auditor change) or 4.02 (non-reliance) 8-K since 2024-08-01, from the EDGAR submissions items field — an index read, not the efts.sec.gov full-text sweep for "restatement", which is still not run.


Transcripts read (§3.1)

All eight archived calls, FY25 Q3 (2024-12-03) → FY27 Q2 (2026-08-27), read in §3.1 order: safe harbour skipped, CFO section skimmed for guidance, Q&A read in full. Format: a live queue, one question plus a follow-up, 9–12 analysts a call; the CEO takes nearly every question, and the new CFO got one question on his first call (Murphy: "you guys are sparing him too much on this call", CALL-FY27Q2, Matthew Murphy).

Guidance changes (step 2)

Quarterly outlook, first guide vs result (the only guide — Marvell does not revise within the quarter, except the Q1 FY26 range narrowing):

QuarterRevenue guideActualNon-GAAP GM guideActualNon-GAAP EPS guideActualGAAP EPS guideActual
Q3 FY25$1,450m ±5%1,516~61%60.5%$0.40 ±0.050.43$(0.09) ±0.05$(0.78)
Q4 FY25$1,800m ±5%1,817~60%60.1%$0.59 ±0.050.60$0.16 ±0.050.23
Q1 FY26$1,875m ±5%, narrowed to ±2% on 2025-05-061,895~60%59.8%$0.61 ±0.050.62$0.19 ±0.050.20
Q2 FY26$2,000m ±5%2,006~59.5%59.4%$0.67 ±0.050.67$0.21 ±0.050.22
Q3 FY26$2,060m ±5%2,07559.5–60.0%59.7%$0.74 ±0.050.76$2.03 ±0.052.20
Q4 FY26$2,200m ±5%2,21958.5–59.5%59.0%$0.79 ±0.050.80$0.36 ±0.050.46
Q1 FY27$2,400m ±5%2,41858.25–59.25%58.9%$0.79 ±0.050.80$0.31 ±0.05$0.04
Q2 FY27$2,700m ±5%2,73958.25–59.25%58.9%$0.93 ±0.050.94$0.37 ±0.050.33
Q3 FY27$3,150m ±5%Dec 202657.5–58.5%$1.10 ±0.05$0.53 ±0.05

Guides: each quarter's prior PR- outlook table; results: each quarter's own PR- headline. Revenue landed $6m–$66m above the midpoint in all eight quarters, inside the ±5% band every time; non-GAAP EPS landed $0.00–$0.03 above the midpoint every time. This is a management that guides to a number it will hit, not one it will beat.

The annual path, raised every quarter since September 2025:

MetricSep 2025Dec 2025Mar 2026May 2026Aug 2026
FY27 revenue~$9.5bn (JPMorgan-hosted investor call; figure as Murphy restated it on 2026-03-05)~$10bnapproaching $11bnnearly $11.5bn~$12bn (+45%)
FY27 data-center growthindexed to cloud capex (18%)>25%40%~50%~60%
FY27 custom growth—≥20%>20%>20%"ramp significantly in the second half"
FY28 revenue—~$13bn (DC ~40%)~$15bn, non-GAAP EPS "well over $5"$16.5bn (+45%)~$18bn (+50%)
FY28 data-center growth—~40%close to 50%~55%>60%
FY27 non-GAAP opex—half the rate of revenue growth, ex-acquisitionsQ1 $575m, flat Q2, low-mid single digits Q3/Q4~$2.45bn~$2.55bn

Capex/buyback/margin changes: the ~$1bn of FY27 capacity prepayments first appears on 2026-05-27 (Meintjes) and is reaffirmed 2026-08-27 (Durn); buybacks ran at $200m a quarter from Q4 FY25 except the Q1 FY26 $340m and the Q3 FY26 $1.0bn ASR; the non-GAAP operating-margin target (38–40%) moved from approaching the bottom of the range by the end of FY26 (Meintjes, 2024-12-03) to entering the range in Q4 FY27 (Durn, 2026-08-27) — S8 row 13. Durn's first call adds that FY28 gross margin should sit in the same range as H2 FY27 and that the long-term model will be reset at the 2026-10-06 Investor Day.

What analysts asked about (step 3), by topic

  • The XPU programs — continuity, exclusivity and second-sourcing at the lead customer and the new tier-1 customer. Asked on every one of the eight calls (Seymore FY25 Q4; Harlan Sur FY25 Q3, FY26 Q2 and FY26 Q3; Vivek Arya FY26 Q1, FY26 Q4, FY27 Q1, FY27 Q2; Quinn Bolton and Josh Bacalter FY26 Q1). The single most persistent question in the record.
  • The long-range revenue bridge: FY27/FY28 outlooks, the custom $10bn+ FY29 target, the Google warrant's $120bn-over-six-years framing (Seymore, Reitzes, Sur, Kumar, Schneider).
  • Optics and interconnect growth vs cloud capex, 800G vs 1.6T mix, LPO/CPO (Rakers, Svanberg, O'Malley, Pajjuri, Bolton, Arya).
  • Scale-up switching, UALink/ESUN, Celestial/CPO timing and customers (Rolland, Bolton, Rakers, Moore, Caso, Svanberg).
  • Supply constraints and capacity (Caso, Moore, Kwan).
  • Gross margin under custom mix (Curtis FY26 Q1, Bolton FY27 Q2).
  • Capital allocation of the divestiture proceeds (Schneider FY26 Q2).
  • Management commitment — Seymore asked Murphy directly in December 2024 whether he was looking elsewhere; answer: "I'm 100% focused on Marvell" (CALL-FY25Q3, Matt Murphy).

Dodges (step 4)

Scored as dodges — the answer addressed a different question:

  1. FY25 Q3 (2024-12-03) · next-generation XPU vendor · Harlan Sur (JPMorgan). Asked "is it fair to assume that you will be the ASIC vendor supporting your customer on this next-gen 3-nanometer training ASIC targeted to ramp late next year?" Matt Murphy answered with the $40bn custom TAM, the 20% share target and the AWS agreement; the vendor question was not addressed. CALL-FY25Q3.
  2. FY26 Q1 (2025-05-29) · second-sourcing protection · Quinn Bolton (Needham). Asked "Do you guys have guarantees that you don't get blanked at the end of the year if they, you know, take the other path?" Matt Murphy referred back to his earlier answer and to revenue confidence; guarantees were not addressed. CALL-FY26Q1. (To Vivek Arya's exclusivity question on the same call he gave an unusually direct answer: "it's certainly possible and likely that customers and our customer may be pursuing multiple paths to meet their requirements" — logged as an answer, not a dodge.)
  3. FY26 Q2 (2025-08-28) · status of two programs management itself had dated · Harlan Sur. Asked whether the 3nm follow-on at the lead customer — for which Murphy had said in May "we have now secured three-nanometer wafer and advanced packaging capacity and expect to start production in calendar 2026" (CALL-FY26Q1) — and the third XPU customer's back-half-2026 ramp were on track: "Is this program still tracking?" Matthew J. Murphy: "commenting on just the individual sockets at this point is only probably increasing the noise level", then the 20% share target. A dodge on a topic management guided a quarter earlier — the strongest signal S8 produces. CALL-FY26Q2.
  4. FY26 Q2 · growth next year vs a ~50% industry · Vivek Arya (BofA). Matthew J. Murphy: "We don't do an annual guide, and we typically just guide a quarter at a time", then enterprise-networking and carrier recovery. Four weeks later (2025-09-24) the company gave an FY27 revenue outlook on an investor call. CALL-FY26Q2.
  5. FY26 Q4 (2026-03-05) · exclusivity at the new XPU customer · Vivek Arya. Asked "How would you handicap kind of your exclusivity at the large new XPU customer you plan to start at next year?" Matthew Murphy answered on engagement depth and multigenerational planning; exclusivity was not addressed. CALL-FY26Q4.
  6. FY27 Q1 (2026-05-27) · exclusivity at the new XPU program · Vivek Arya. "Do you expect to be exclusive in this?" Matthew J. Murphy answered on milestones and the program's third of custom growth; exclusivity was not addressed. CALL-FY27Q1.
  7. FY27 Q2 (2026-08-27) · size of the new XPU program · Vivek Arya (partial). Matthew Murphy: "without quantifying it", then "probably the most meaningful part of the next year raise is actually from the scale-up optics switching in other areas". CALL-FY27Q2.

Logged, not scored (an explicit refusal or a dated deferral is not an evasion): Ross Seymore FY25 Q4 on a competitor at the lead customer — "it's just something we can't comment on" (CALL-FY25Q4, Matt Murphy); Josh Bacalter FY26 Q1 on multiple tracks (Murphy: he had given "the max color"); the new socket wins FY27 Q1 ("No additional details at this time"); the Google warrant's size, deferred to the 2026-10-06 Investor Day (FY27 Q2, Sur and Reitzes).

The thread to pull: the same topic — Marvell's position, exclusivity and continuity inside its XPU programs — was dodged in six quarters (items 1, 2, 3, 5, 6, 7), three of them to the same analyst, and once on a date management itself had given. It matters because the custom line carries the growth: custom "to more than double" in FY28 with the new tier-1 program a third of it (CALL-FY27Q1). A filing can settle it: the 10-Q significant-customer table and the Investor Day's custom breakdown. Carried to the scorecard's ⚠️ line and to a kill criterion.


S8 — Management credibility

Twenty rows; eighteen scorable now, both locators on each. Speaker names are the transcript labels at the promise; roles from EDGAR as of the call date (8K-2026-06-11, 8K-5.02-set, relationships.yaml officer_of edges): Matt Murphy, Chairman and CEO throughout; Willem Meintjes, CFO from January 2023 to 2026-06-15 (departed); Dan Durn, CFO from 2026-06-15. Written quarterly outlooks are company (written); on each call the CFO read the same figures (Meintjes 2024-12-03 → 2026-05-27; Durn 2026-08-27).

#Promise (metric, period, figure)Where madeWhoPathResultLocatorScore
1Q3 FY25 revenue $1,450m ±5%; non-GAAP EPS $0.40 ±0.05PR-Q2FY25 outlook, 2024-08-29company (written)not revised$1,516m; $0.43PR-Q3FY25met
2Q4 FY25 revenue $1,800m ±5%; EPS $0.59 ±0.05PR-Q3FY25; CALL-FY25Q3company (written); Willem Meintjes, CFOnot revised$1,817m; $0.60PR-Q4FY25met
3Q1 FY26 revenue $1,875m ±5%; EPS $0.61 ±0.05PR-Q4FY25; CALL-FY25Q4company (written); Willem Meintjesrange narrowed to ±2% at the same midpoint (8K-2025-05-06)$1,895m; $0.62PR-Q1FY26met
4Q2 FY26 revenue $2,000m ±5%; EPS $0.67 ±0.05PR-Q1FY26; CALL-FY26Q1company (written); Willem Meintjesnot revised$2,006m; $0.67PR-Q2FY26met
5Q3 FY26 revenue $2,060m ±5%; GM 59.5–60.0%; EPS $0.74 ±0.05PR-Q2FY26; CALL-FY26Q2company (written); Willem A. Meintjesnot revised$2,075m; 59.7%; $0.76PR-Q3FY26met
6Q4 FY26 revenue $2,200m ±5%; GM 58.5–59.5%; EPS $0.79 ±0.05PR-Q3FY26; CALL-FY26Q3company (written); Willem Meintjesnot revised$2,219m; 59.0%; $0.80PR-Q4FY26met
7Q1 FY27 revenue $2,400m ±5%; GM 58.25–59.25%; EPS $0.79 ±0.05PR-Q4FY26; CALL-FY26Q4company (written); Willem Meintjesnot revised$2,418m; 58.9%; $0.80PR-Q1FY27met
8Q2 FY27 revenue $2,700m ±5%; GM 58.25–59.25%; EPS $0.93 ±0.05PR-Q1FY27; CALL-FY27Q1company (written); Willem A. Meintjes — departed before the result; reaffirmed in 8K-2026-06-11 Ex. 99.1not revised$2,739m; 58.9%; $0.94PR-Q2FY27met
9Q3 FY25 GAAP EPS $(0.09) ±0.05PR-Q2FY25company (written)—$(0.78): restructuring and intangible-impairment charges (FY25 total $711.8m) taken in the quarterPR-Q3FY25missed (explained)
10Q1 FY27 GAAP EPS $0.31 ±0.05PR-Q4FY26; CALL-FY26Q4company (written); Willem Meintjesguide issued after Celestial and XConn had closed$0.04 — "lower than our guidance reflecting the impact of purchase accounting for the Celestial AI and XConn acquisitions and the related earn out obligation" (CALL-FY27Q1, Willem A. Meintjes); $250.7m earn-out mark-to-marketPR-Q1FY27missed (explained)
11FY26 custom revenue ~$1bn — "about $500 million from custom going to $1 billion next year"CALL-FY25Q3, 2024-12-03Matt Murphy, CEO"overshooting" in the same answer"we have scaled from zero revenue to $1.5 billion in fiscal 2026"CALL-FY26Q4 (Matthew Murphy)beat
12FY26 non-GAAP gross margin ~60% — "we do see a path to continue to be about 60% through next year"CALL-FY25Q3Willem Meintjes, CFO (departed)—FY26 non-GAAP GM 59.5%PR-Q4FY26 reconciliationmet
13❗ Non-GAAP operating margin "really starting to approach the bottom end of our long-term range towards the end of next year" (38–40%, by end FY26)CALL-FY25Q3Willem Meintjes, CFO (departed)FY26 Q4 margin 35.7%; the range now "likely to enter our 38% to 40% long-term target range in Q4 of this fiscal year" (Daniel Durn, CALL-FY27Q2)~2.3 pts short at the date; pushed a year, not presented as a slipPR-Q4FY26; CALL-FY27Q2missed
14SBC as a share of revenue down — "We are also focused on reducing our stock-based compensation expense as a percent of revenue, and we expect significant improvement in this metric going forward."CALL-FY25Q3Matt Murphy, CEO—FY25 10.4% → FY26 7.2%; then Q2 FY27 11.9%PR-Q4FY26; PR-Q2FY27met for FY26 (reversed since — S3 flag 2)
15Automotive-Ethernet sale to close "within calendar 2025"CALL-FY26Q1, 2025-05-29Matt Murphy, CEO—closed 2025-08-14PR-Q3FY26beat
16Celestial AI close — "The acquisition is expected to close in the first quarter of next year"CALL-FY26Q3, 2025-12-02Matthew Murphy, CEO—closed 2026-02-0210Q-Q2FY27, Note 4met
17Lead-customer 3nm follow-on XPU — "expect to start production in calendar 2026"CALL-FY26Q1Matt Murphy, CEOstatus dodged 2025-08-28 (dodge 3); POs for all of FY27 (2025-12-02)"we have purchased orders covering the entirety of this fiscal year's forecast for this next-generation program and are now ramping production"CALL-FY26Q4 (Matthew Murphy)met
18FY27 revenue ~$9.5bn (first guide, September 2025)JPMorgan-hosted investor call 2025-09-24, not archived — figure as Matthew Murphy restated it: "we provided a fiscal 2027 revenue outlook of approximately $9.5 billion"Matt Murphy, CEO→ ~$10bn → approaching $11bn → nearly $11.5bn → "roughly $12 billion"period ends 2027-01-30; H1 FY27 $5,157m and Q3 guided $3,150mCALL-FY26Q4; CALL-FY27Q2beat (in progress) — raised four times
19Investor Day on 2025-06-10announced in CALL-FY25Q4 (Matt Murphy)Matt Murphy, CEO"postpone its previously scheduled Investor Day from June 10, 2025, to a future date in calendar 2026" (8K-2025-05-06), citing the macro environment; now 2026-10-06postponed 16 months, openly8K-2025-05-06; CALL-FY27Q2missed (disclosed)
20FY26 AI revenue — "we also expect to very significantly exceed our $2.5 billion target in fiscal 2026"CALL-FY25Q4Matt Murphy, CEOdeclined to size it (Arcuri, same call); from Q3 FY26 the company reports data center and communications onlyno FY26 AI-revenue figure was ever reported; custom alone was $1.5bn and data center $6.1bn, so the target was very likely passed — but not against the promised metricPR-Q4FY26; CALL-FY26Q4withdrawn (metric dropped)

Not scored, still open: FY28 revenue ~$18bn (first guide ~$13bn in Dec 2025, raised three times); the ~$1bn FY27 capacity prepayments ($487.0m on the balance sheet at 2026-08-01); the new tier-1 XPU program's calendar-2026 production start (first guided CALL-FY25Q4); Celestial's $500m annualised run rate by Q4 FY28; Q3 FY27 outlook. Context, not scored: on 2025-12-02 Meintjes said Celestial would be funded "through a combination of stock and cash on hand" without additional debt; it was (cash plus 26.8m shares issued across the two acquisitions, 10Q-Q2FY27 equity statement), and debt then rose $0.5bn net with the April 2026 notes, two months after the close.

Tally (20 rows): 3 beat (one in progress), 12 met, 4 missed, 1 withdrawn. Promisers: Murphy made 8 rows and holds the role; the ten outlook rows (1–10) are company-written (read out on the calls by Meintjes); Meintjes personally made 2 of 20 (rows 12–13) and has gone. The current team — Murphy, and the company outlooks now signed off by Durn — carries 18 of 20.

Are the promisers still there? Murphy yes. Meintjes resigned 2026-06-10, effective 2026-06-15 — he "notified the Company on June 10, 2026 of his decision to resign from his position effective as of June 15, 2026", with no disagreement stated and an advisory role to April 2027 (8K-2026-06-11). That falls two weeks after row 10's GAAP miss (reported 2026-05-27) and one quarter after row 13's margin date passed short — the skill's own flag (a CFO leaving within two quarters of a missed guide), recorded as such. Both misses have ordinary explanations (earn-out accounting; a target moved with a larger revenue plan), and his successor came off Marvell's own board (audit chair), which argues continuity rather than rupture. Other departures in the window touched no scored row: Raghib Hussain (President, Products and Technologies, resigned 2025-05-02) and Loi, the Inphi co-founder whose retirement was announced on CALL-FY25Q3 (surname Nguyen [background]), made no call-level promises in the archive.

Dodge log rolled up: 7 scored dodges in 8 calls; one topic — Marvell's position, exclusivity and continuity in its XPU programs — dodged in six quarters, once on a date management had given itself (dodge 3). Four explicit refusals, not scored.

Verdict: ✅ reliable. Every quarterly revenue, non-GAAP gross-margin and non-GAAP EPS guide in eight quarters landed inside the range, just above the midpoint — met, never beat by much, never missed. The annual outlook was raised four times running. The misses are small in number and explained: two GAAP EPS misses on charges (restructuring; the earn-out marked to a rising share price), a margin target a year late, an Investor Day postponed in the open. The one withdrawn row (the FY26 AI-revenue target, never reported against) matters more than its count, because it is the same habit as the dodge log: management is precise about the P&L it guides and vague about the customer programs underneath it. S7's base case takes the guide as given; no haircut.


What the S7 table should know (not changed here)

S8 says reliable, so the valuation line keeps its figures. Two facts from this pass bear on it and are for the next S7 re-run, not applied here: management's FY28 outlook is now ~$18bn against the 2026-09-21 model's ~$17.9bn (the model is on the guide); and the model's ~915m FY28 diluted shares omit the Celestial earn-out (24.4m) and the Google warrant (59.0m at $206.58, vesting on revenue milestones through FY33) — together ~9% more shares if both vest in full.

Full notes, 21 Sep 2026 11 min read

MRVL — dossier, 2026-09-21

Slot names-ai-capex-2. Model-drafted unattended by the cloud research routine; nothing here is human-verified. Sources and locators are in relationships.yaml; figures below carry theirs inline. Anything from background knowledge is marked [background].

Price reference: $244.25 close, 2026-09-18 (market data get_latest_stock_price NASDAQ:MRVL; volume 23,549,789 shares, so roughly $5.8bn traded that day — liquidity is not a constraint). Shares outstanding 876.9m at 2026-08-21 (10-Q cover, dei EntityCommonStockSharesOutstanding), plus up to ~21.8m on conversion of the NVIDIA preferred → ~898.7m as-converted, ~$219bn market capitalisation as-converted.

S1 — Source

watchlist (owner, Apple Stocks, 2026-09-20), cluster ai_semis, coverage sec_domestic, cik

  1. No Tier 0 promotion; queue.json has never promoted MRVL.

S2 — Kill test

All figures from the XBRL company facts of the SEC filings (data.sec.gov/api/xbrl/companyfacts, CIK 0001835632), fiscal years ending late January.

FY (ends)Revenue $mGross profit $mGM %Op income $mOp margin %OCF $mCapex $mFCF $mDiluted WA shares m
FY22 (2022-01-29)4,4622,06446.3-348-7.8819169650797
FY23 (2023-01-28)5,9202,98850.52384.01,2892061,083851
FY24 (2024-02-03)5,5082,29441.6-568-10.31,3703361,034861
FY25 (2025-02-01)5,7672,38241.3-720-12.51,6812851,396866
FY26 (2026-01-31)8,1954,18151.01,32316.11,7503541,396870

Quarterly, most recent six (10-Q XBRL):

Quarter endedRevenue $mGross profit $mGM %Op income $mNet income $m
2025-05-03 (Q1 FY26)1,89595250.2271178
2025-08-02 (Q2 FY26)2,0061,01150.4290195
2025-11-01 (Q3 FY26)2,0741,07051.63581,901
2026-05-02 (Q1 FY27)2,4181,26152.233934
2026-08-01 (Q2 FY27)2,7391,45653.2460308

Survives — but the headline profit number is a trap. FY26 GAAP net income of $2,670m is not earnings power: $1,901m of it landed in Q3 FY26 alone, the quarter of the automotive-ethernet divestiture, against $358m of operating income in the same quarter. The FY26 operating line — $1,323m on $8,195m — is the honest read, and it is the first positive operating year since FY23. GAAP gross margin has climbed five consecutive quarters, 50.2% → 53.2%.

  • Growth. Revenue +42% in FY26 and +37% year-on-year in Q2 FY27, "primarily due to a 46% increase in sales from the data center end market which benefited from strong AI-related demand" (10-Q, Item 2, MD&A). Data center was $6,100.3m, 74% of FY26 revenue (10-K, Note 3, end-market table).
  • Leverage. $5.0bn of borrowings at 2026-08-01, all senior notes, against $3.93bn cash — roughly $1.1bn net debt, comfortably under 1x any reasonable EBITDA measure. The $1.5bn revolver is undrawn and runs to 2030-06-30. The nearest maturity was the $500.0m 2026 notes, repaid at maturity out of the April 2026 issue.
  • Dilution. Diluted weighted-average shares 676m (FY20) → 870m (FY26), +29% in six years, and that is before the ~21.8m as-converted preferred. Buybacks are not offsetting it: $400.0m repurchased in the first half of FY27 against $9.7bn of authority.
  • Cash runway. Not applicable; operating cash flow positive every year in the table.
  • Analyst coverage count. Not obtained — see "Evidence that was thin".

S3 — Filings, gate-minimum

Read: FY26 10-K (Item 1A risk factors, Item 7 MD&A, Notes 2/3); Q2 FY27 10-Q (Item 2 MD&A, Notes 4, 5, 10, plus Part II Item 1A); DEF 14A filed 2026-05-13 (beneficial ownership); the EDGAR form census since 2023-01-01. No transcript was read — guidance came from Equibles' structured rows off the 2026-08-27 8-K and call, not from the call text.

The three biggest risks management itself discloses

  1. Customer and channel concentration, deepening. "during fiscal 2026, there were two customers (one distributor and one direct customer) whose revenues represented 10% or more of total net revenue... net revenue from our ten (10) largest customers... represented 82% of our total net revenue for fiscal 2026" (10-K, Item 1A, "Our sales are concentrated in a few large customers"). The Q2 FY27 table puts Distributor A at 44% of net revenue against 34% a year earlier, and Customer A flat at 16%. Credit follows: "The Company's accounts receivable was concentrated with four customers at January 31, 2026, who represented a total of 73 % of gross accounts receivable" (10-K, Note 2).
  2. The AI capex cycle itself, named as a risk in management's own words. "the current level of capital expenditure (capex) on AI infrastructure may not be sustainable over the long term and a significant reduction in AI-related spending will likely harm our financial results" (10-K, Item 1A, "Unfavorable or uncertain conditions in the Data Center and Communications markets..."). With 74% of revenue in data center, this is the whole company.
  3. Insourcing by the customers themselves. "AI-driven design tools may lower traditional barriers to entry in the semiconductor industry by enabling new market participants... to develop high-performance, custom semiconductor solutions in-house with reduced reliance on third-party chip suppliers" and "Given their dependence on semiconductor products to operate their data centers... some large customers may begin developing and making their own semiconductor solutions which could result in a loss of business for Marvell" (10-K, Item 1A). Marvell's business is custom silicon for hyperscalers; the risk factor describes its customers becoming its competitors.

Alongside these, geographic concentration rose sharply: sales to customers with operations in Asia were 84% of net revenue in Q2 FY27 against 76% a year earlier (10-Q, Item 2).

The quarter's structural change

Two acquisitions closed in the first ten days of FY27 — Celestial AI for $3.5bn (2026-02-02, photonic fabric for scale-up interconnect, with contingent cash and shares payable through fiscal 2029 on revenue milestones) and XConn for $469.0m (2026-02-10, PCIe/CXL switching). Goodwill went $11.1bn → $13.9bn; $2.5bn of the consideration was non-cash. Marvell then entered a cash-settled forward stock purchase contract with a $300.0m notional to hedge its own share price against the Celestial contingent consideration — the earn-out is payable partly in stock.

Red-flag sweep

  • Auditor: Deloitte put to the 2026 AGM for appointment; no 8-K item 4.01 and no NT 10-K in the form census since 2023. No restatement signal found. This is a filing-census read, not the efts.sec.gov full-text sweep the skill specifies — see "Evidence that was thin".
  • Going concern: not applicable on these cash flows.
  • Related party: nothing surfaced in the sections read. The NVIDIA preferred is disclosed as an equity transaction, not a related-party one.
  • Executive changes: none in the period read.

S5 — Ownership

Register (DEF 14A table, as of 2026-01-31, on 847,287,680 shares): FMR LLC 126,698,928 / 14.95%; The Vanguard Group 79,610,185 / 9.40%; BlackRock 60,534,116 / 7.14%.

Insider ownership is negligible. All 13 directors and executive officers as a group held 1,046,798 shares, marked "*" for under one percent — about 0.12%. Matthew Murphy, Chairman and CEO, holds 412,871. Alignment here runs through compensation, not through stock.

13F, report date 2026-06-30 (threespread.py holders MRVL; 1,560 filers, 499.6m shares reported). Largest adds: BlackRock +21,798,190 to 85,641,174; State Street +18,495,866; T. Rowe Price +12,134,409; Cantor Fitzgerald +11,818,810; Susquehanna +9,769,993; Citadel +8,984,768. Sixth Street Partners is new at 5,225,664. Largest cut: Ameriprise −32,577,650 to 9,111,628. ⚠️ The quarter is incompletely ingested. FMR (131.3m shares), both Vanguard entities (78.2m and 33.2m), UBS, Geode, Franklin, FIL, Morgan Stanley and Northern Trust have no 13F in this quarter yet. Their absence is ingestion lag, not an exit, and must not be read as one.

Insider transaction cadence: 143 Form 4 filings in the twelve months to 2026-09-21 and 22 Form 144 proposed-sale notices since 2026-03-25, the most recent 2026-08-17. The tool returns filings, not transaction detail, so this is a cadence reading — heavy and one-directional in the Form 144 sense — not a measured net buy or sell.

Short interest (FINRA via Equibles): 38,175,634 shares at 2026-08-31, 4.4% of shares outstanding, 1.4 days to cover. Nobody is short this in size.

What hangs over it. A live S-3ASR automatic shelf filed 2025-03-12 — the only S-3 on the register since 2023, and the one the April 2026 $1.0bn notes drew on. An S-3ASR carries no stated dollar cap. Against it, $9.7bn of buyback authority (a $5.0bn addition authorised 2025-09-24), of which only $400.0m was used in the first half of FY27 while $2.0bn of preferred and $1.0bn of notes were issued. Capital is coming in, not going out.

And the fact that reframes the register: on 2026-03-31 NVIDIA bought 2.0m shares of Series A Convertible Preferred for $2.0bn cash, convertible into up to ~21.8m common shares at ~$91.8355, voting with the common on an as-converted basis on everything except director elections. At $244.25 that conversion right is worth roughly $3.3bn more than the $2.0bn paid — a gain to NVIDIA larger than Marvell's entire FY26 operating income. Management's own risk factor names the consequence: "because the Series A Preferred Stock was issued to a single significant holder, the concentration of ownership could increase NVIDIA's influence over matters submitted to our stockholders". Marvell's 10-K lists "Nvidia Corporation" among companies that "compete directly with our businesses" (10-K, Item 1, Competition).

S8 — Valuation and sizing ⏳ OPEN (user)

Built from management's own guides (Equibles GetGuidance, rows dated 2026-08-27, sourced to the Q2 FY27 8-K and call): FY27 revenue ~$12B; data-center revenue growth 60% in FY27 and 60% in FY28; FY27 non-GAAP opex $2.55B; Q3 FY27 non-GAAP gross margin 57.5–58.5%; Q3 FY27 non-GAAP EPS $1.05–1.15 against GAAP EPS $0.48–0.58; FY28 non-GAAP tax rate 13%.

Model lineFY27EFY28E
Data center revenue $m9,760 (FY26 $6,100 × 1.60)15,616 (× 1.60 again)
Communications & other $m~2,240~2,240 (held flat)
Revenue $m~12,000 (matches the guide)~17,900
Non-GAAP gross margin58.0% (mid of guide)58.0% held
Non-GAAP gross profit $m6,96010,380
Non-GAAP opex $m2,550 (guided)3,060 (+20%)
Non-GAAP operating income $m4,4107,320
Net interest $m-200-200
Tax11% (guided Q3 rate)13% (guided)
Non-GAAP net income $m~3,740~6,190
Diluted shares m (as-converted)~905~915
Model non-GAAP EPS~$4.13~$6.76
Multiple at $244.25~59x~36x

Bear / base / bull, on the FY28 line: bear — data center grows 25% not 60%, margin holds, EPS ~$3.90, and a 25x multiple on a decelerating custom-silicon franchise gives ~$98; base — the table above, 30x on $6.76 gives ~$203; bull — 60% holds and mix lifts margin to 60%, EPS ~$7.40, 35x gives ~$259. The base case is below today's price. That is the finding, and it says the market is already paying for the FY28 guide plus more.

Two adjustments the owner should make before trusting any of it. First, the GAAP/non-GAAP gap is not noise: Q3 FY27 is guided at $0.48-0.58 GAAP against $1.05-1.15 non-GAAP, a ~$0.55 quarterly wedge, ~$2bn a year of stock compensation and acquired-intangible amortisation — and the share count above does not absorb the continuing issuance that wedge represents. Second, the FY28 revenue line is management's 60%, compounded; it is a guide, not a forecast, and the kill criteria are written to catch it breaking.

Proposed buy price and proposed size are owner-only and are not stated here. Liquidity does not constrain sizing: ~$5.8bn traded on 2026-09-18.

S9 — Monitoring

  • Next earnings: Q3 FY27 results, expected early December 2026 — derived from filing cadence (Q3 FY26 results 8-K filed 2025-12-02, 10-Q 2025-12-03), not from a calendar feed.
  • Filing feed: the Tier 0 EDGAR sweep already covers cik 1835632 (watchlist.yaml carries it), so 8-K, 10-Q, S-3, 13D/G and Form 4 arrive without further setup.
  • Leading indicator between quarters: the Distributor A percentage in each 10-Q's significant-customer table. It is disclosed quarterly, it moved 34% → 44% in a year, and it is the single number that decides whether this is a diversified custom-silicon franchise or one channel.
  • Trigger that converts this to a buy: data-center revenue on track to the guided 60% for FY28 and non-GAAP gross margin holding at or above 58% and the price at or below ~30x the model FY28 EPS — i.e. a FY28 multiple that does not already assume the guide. On today's model that is roughly $203, about 17% below the 2026-09-18 close.

Evidence that was thin, and why

  • No earnings-call transcript was read. Guidance came from Equibles' structured GetGuidance rows (one call), which carry the 8-K source links but not the Q&A. The gate-minimum spec asks for two transcripts and the Q&A; this run traded them for budget across three names.
  • The efts.sec.gov full-text sweep for "changes in accountants" / "restatement" / NT 10-K was not run. The auditor and restatement reads above come from the EDGAR form census and the proxy instead — weaker evidence, and an item-4.01 8-K would not be visible in a form census.
  • No analyst coverage count. Not available from the rungs used without spending an Equibles call; the skill lists it under S2.
  • Proxy compensation structure not extracted — only the ownership table was read. What the bonus is paid on is unread.
  • Next earnings date is derived from filing cadence, not a calendar feed; market data's earnings calendar needs a paid plan on this account (established in slot names-ai-capex-1).
  • The 2026-06-30 13F quarter is incompletely ingested — see S5. Every "exit" reading is suppressed accordingly.