Analog Devices ADI

Reference table

> EARNINGS BANNER — fiscal Q3 2026 reports 2026-08-19, FIFTEEN DAYS from this dive. Consensus non-GAAP EPS $3.34 and revenue $3.915B, implying +35.9% revenue growth year on year and +8.0% sequentially. ADI has beaten non-GAAP EPS in each of the last four quarters. The verdict below is set with this timing as an explicit input: we are not asking a reader to take a position two weeks ahead of a print in a name at 30.7x forward that has already risen 72% in a year.

1. What the business is, and how the disclosure works

Chief executive Vincent T. Roche; approximately 24,500 employees at 2025-11-01, of whom approximately 13,000 are in engineering roles (10-K); NASDAQ-listed; CIK 0000006281; fiscal year ends the Saturday nearest 31 October. Reporting currency USD and profile.currency USD — no mixing trap.

Four end markets, and seg_prod ties exactly. The four lines sum to $11,020.0M against reported fiscal-2025 revenue of $11,020M — an exact match. This is a clean vendor-to-filing reconciliation and worth recording after the segment failures elsewhere in this batch.

Read the last column, because it is the single most important corrective to the current narrative. Fiscal 2025 revenue was still 10.4% BELOW the fiscal-2023 peak, and Industrial — the largest and highest-margin end market — was 24.8% below its own peak. The fiscal-2026 consensus of $14.746B would put revenue 19.8% above the fiscal-2023 high, which is a genuine new plateau if it lands. But the growth being celebrated in fiscal 2026 is, in Industrial's case, still partly the recovery of ground lost in fiscal 2024.

Automotive is the exception and the real secular story: $1.249B in fiscal 2021 to $3.278B in fiscal 2025, up 162% in four years, and 12.4% above its fiscal-2023 level while the company overall was 10.4% below. That is content per vehicle, not cycle.

Geography — and seg_geo double-counts in the fiscal-2025 entry. The fiscal-2025 block contains both a Non-US line of $7,781.6M and the five regional lines that compose it (China $2,858.3M, Europe $2,285.6M, Japan $989.9M, Rest of Asia $1,485.3M, Rest of North and South America $162.5M = $7,781.6M exactly), plus United States $3,238.1M. Summing every field gives $18.80B against revenue of $11.020B. The correct read is US $3,238.1M + Non-US $7,781.6M = $11,019.7M, an exact tie. The fiscal-2024 and earlier entries do NOT contain the Non-US line, so the block is not internally consistent across years and any multi-year geographic series built from it naively will show a spurious 71% jump in fiscal 2025.

China rose from 22.6% to 25.9% of revenue in a single year, and is now the second-largest market. The 10-K's risk factors name "recently announced and future tariffs and other trade restrictions" explicitly.

Distribution concentration is disclosed and material. The 10-K: "Sales to third-party distributors accounted for approximately 56% of our revenue in the year ended November 1, 2025." Within that, the concentration table shows Distributor 1 at 24% of revenue (24%, 24%, 25% across the three years) and Distributor 3 at 13% (13%, 12%, —). A single distributor is roughly a quarter of the company, and distributor inventory is where a semiconductor cycle turns before it shows in end demand. There is no vendor field for this.

Manufacturing is hybrid and internal capacity matters. The 10-K: "In addition to leveraging an outsourcing model for certain manufacturing operations, we also rely on our internal manufacturing operations located in the United States, Ireland, the Philippines, Thailand and Malaysia." This is the mechanical reason gross margin swings 630 basis points on a 37% revenue move, and it cuts both ways.

2. The upturn, quantified

Five things follow.

First, the sequential acceleration is real: +6.8%, +2.7%, +14.7%. The May quarter is the largest sequential step in the file.

Second, the margin move is the story. The 10-Q states gross margin of 67.3% against 61.0% for the quarter and 66.1% against 60.1% for the half. Operating margin went from 25.7% to 38.1% in the same twelve months — a 1,240 basis point expansion, roughly double the gross-margin move, because operating expenses grew far more slowly than revenue.

Third, four consecutive non-GAAP beats, and the margin of beat is widening: +8.8%, +5.1%, +0.9%, +6.5%, +6.9%.

Fourth, the GAAP-to-non-GAAP gap is the widest in this batch and it is shrinking. Trailing GAAP diluted EPS is $6.73; trailing non-GAAP is $9.86 — a 46.5% gap. The cause is acquired-intangible amortisation from the Linear Technology and Maxim Integrated deals: intangibles fell from $9.585B (FY2024) to $8.014B (FY2025) to $7.255B at 2026-05-02, implying roughly $1.57B a year of amortisation, or about $3.20 per share pre-tax. This drag is mechanically finite. As it runs off, GAAP earnings converge upward toward non-GAAP without the business doing anything — which is a real, underappreciated tailwind to the reported number, and a reason not to anchor on the 56.6x trailing GAAP multiple.

Fifth, and this is the caution: the same fabs that produced the margin expansion produced the collapse. Fiscal 2024 revenue fell 23.4% from $12.306B to $9.427B, and gross margin fell to 57.1% (from $5.381B on $9.427B) from 64.0% in fiscal 2023. The 630 basis points now being added were, in large part, the 690 basis points lost then. The honest question is not whether margin expanded — it plainly did — but whether 67.3% is the new normal or the top of a utilisation curve.

Research and development was $1.766B in fiscal 2025 — 16.0% of revenue, and researchAndDevelopementToRevenueTTM is 14.9%. That is a company investing properly, and it is 53% of the workforce.

3. Balance sheet

Correction 1 — netDebt omits short-term investments and is six months stale. The vendor's $6,165.0M is total debt less cash and equivalents only, ignoring $1,152.9M of short-term investments. Corrected to the same date that is $4,939.4M; corrected to the most recent filed balance sheet (2026-05-02) it is $5,245.5M. The vendor reports $6.165B; the filing supports $5.246B; we use $5.246B — a 17.5% overstatement corrected.

Correction 2 — the vendor's totalDebt of $8,665.0M does not match the fiscal-2025 balance sheet. The 10-Q's November 2025 column shows commercial paper $446.6M plus long-term debt $8,145.1M = $8,591.7M; the vendor adds capitalLeaseObligations of $72.9M to reach $8,665.0M. A $73M difference, immaterial and defensible, but recorded. The current figure at 2026-05-02 is $8,684.8M across three lines the vendor does not separate: current debt $899.2M, commercial paper $550.2M and long-term debt $7,235.4M. Note the structural change: $899.2M of long-term debt reclassified to current between November and May, i.e. a maturity inside twelve months.

The rebuilt enterprise value:

> Market cap $185.483B (487,087,040 shares × $380.80)

> + total debt $8.685B (10-Q, 2026-05-02)

> − cash $2.437B − short-term investments $1.002B

> = corrected enterprise value $190.729B

The vendor reports $191.760B, an overstatement of $1.031B (0.54%) — small, because the equity value dominates. The leverage picture changes more:

Interest coverage is 12.1x. The current ratio is 1.75. There is no preferred stock outstanding — the 10-Q balance sheet is explicit: "Preferred stock, $1.00 par value, 471,934 shares authorized, none outstanding." This is a comfortable balance sheet and we say so.

Goodwill of $26,973M plus intangibles of $7,255M is $34,229M — 71.4% of total assets and 1.01x total equity. That is the Linear (2017) and Maxim (2021) purchase price sitting on the books. It is the entire explanation for returnOnInvestedCapitalTTM of 7.78% and returnOnEquityTTM of 9.79% in a business earning a 26.0% net margin and a 38.1% quarterly operating margin. We use ROIC as the honest measure and note plainly that the operating business earns far more on the capital it deploys than 7.78% — but shareholders paid for the acquisitions, and ROIC is the measure that remembers that. priceToBookRatioTTM of 5.50x and debtToEquityRatioTTM of 0.257x are both correct and both dominated by the same goodwill.

4. Cash flow and capital returns

A 38.8% free-cash-flow margin is exceptional and it is the best single argument for owning this business. Capital expenditure fell to $533.6M — 4.8% of revenue and 27% of depreciation — while revenue rose 16.9%, because the capacity built during the last upcycle is now being filled rather than extended. freeCashFlowYieldTTM is 2.46% and priceToFreeCashFlowRatioTTM is 40.6x, so the cash is excellent and the price for it is not.

Buyback tripled: $615.6M in fiscal 2024 to $2.165B in fiscal 2025 — about 1.17% of the current market capitalisation. The 10-Q states that as of 2026-05-02 the board had authorised $26.7 billion in aggregate under the repurchase programme with $8.5 billion remaining — roughly 4.6% of market cap in dry powder. Shares outstanding fell from 489,654,097 (2025-11-01) to 487,087,040 (2026-05-02), a 0.52% reduction in six months.

Dividend, and the usual vendor labelling trap. The vendor reports lastDividend: 4.18 and dividendPerShareTTM: 4.18, yielding 1.10%. That $4.18 is the trailing-twelve-month total. The 10-Q gives the actual rate: "On May 19, 2026, our Board of Directors declared a cash dividend of $1.10 per outstanding share of common stock." The forward run-rate is $4.40 — a 1.16% yield, not 1.10%. Payout is 44.6% of trailing non-GAAP EPS (the vendor's dividendPayoutRatioTTM of 60.3% is computed on GAAP and is also correct on its own basis — the two payout ratios differ by 16 points purely because of intangible amortisation, which is a good illustration of why the basis has to be stated).

Total shareholder yield is roughly 2.3% — 1.16% dividend plus about 1.17% buyback.

5. Valuation — priced in or room?

At $380.80 (market cap $185.483B, corrected enterprise value $190.729B):

Estimate coverage is good on FY2026 and FY2027 (18-20 analysts) and thin beyond. The FY2029 row rests on 2 analysts and FY2030 on 4 — both are excluded from every conclusion. The FY2028 EPS row carries 6 analysts and is used only for cross-checking and for the bull case, with the coverage disclosed at each use.

est.ebitAvg is BELOW est.netIncomeAvg in every forward year — arithmetically impossible, and REJECTED. FY2028 shows ebitAvg of $4.854B against netIncomeAvg of $8.150B; FY2030 shows $6.491B against $11.871B. A company cannot earn more net income than earnings before interest and tax while paying both interest and tax. The ratios are also fixed: ebitdaAvg / revenueAvg = 44.46% and ebitAvg / revenueAvg = 25.88% in all eight years in the file, to four significant figures. Per the data contract, all forward valuation runs on epsAvg and revenueAvg.

Peer context. The vendor peer set is unusable in aggregate — it contains Strategy Inc ($32.4B), a bitcoin treasury company, and Dell ($311.4B), neither of which is an analog semiconductor comparable. The defensible peers are Texas Instruments ($259.1B at $283.67), NXP Semiconductors ($60.0B at $237.83), STMicroelectronics ($48.4B at $54.20) and Monolithic Power Systems ($65.7B at $1,337.00). ADI at $185.5B is the second-largest of that group. No peer multiples are supplied in the file, so no peer-multiple comparison is drawn. Note that KLAC also appears in the peer set — a wafer-fab-equipment name, not an analog comparable.

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

At $380.80 — 30.7x FY2026 consensus and 25.2x FY2027 — the price embeds:

• Revenue reaches $14.746B in fiscal 2026 and $17.099B in fiscal 2027 — respectively 19.8% and 38.9% above the fiscal-2023 peak of $12.306B. (Consensus; 20 analysts each.) The near check is 2026-08-19: consensus revenue of $3.915B would be +35.9% year on year, and the fourth quarter would then need roughly $4.05B to hit the full-year number.
• Gross margin holds at or above roughly 67%. (Our number, derived from the 10-Q's 67.3% and the consensus EPS-to-revenue relationship.) This is the most fragile assumption in the price. ADI owns its fabs. Margin was 61.0% twelve months ago and roughly 57% at the fiscal-2024 trough; the 630 basis points came from utilisation, and utilisation is a function of a demand cycle that has now run five consecutive up quarters.
• Non-GAAP EPS reaches $15.086 in fiscal 2027 — up 94.4% from the $7.762 delivered in fiscal 2025. (Consensus; 17 analysts.) Revenue over the same span rises 55.2%, so roughly 39 points of the EPS growth is margin and buyback rather than volume.
• The market keeps paying 25x or better on the forward year at every date. (Our number.) At 20x FY2027E the stock is $302; at 30x it is $453. The multiple is where essentially all of the outcome variance sits, and it has traversed a wide range recently: at the 52-week low of $220.68 the stock was roughly 14.6x the current FY2027 estimate.
• China, at 25.9% of revenue and rising, is not disrupted. (Filing-confirmed share; our assumption on the direction.) The 10-K names tariffs and trade restrictions as a live risk factor and the knowledge base carries an independent claim that Chinese domestic suppliers are now "matching and undercutting" ADI's closest competitors in microcontrollers and power chips.

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

Expected return over roughly two years decomposes as: non-GAAP EPS growth (+39.9%, from FY2026E $12.408 to FY2028E $17.352) + multiple drift (COMPRESSION, from 30.7x to 23x on the respective forward year, −25%) + shareholder yield (+2.3% a year, ~4.7% compounded) ≈ +15% to +20% total, or roughly +8% annualised.

Our base assumes substantial multiple compression and the reason is the shape of the growth, not a view on the business. Today the market pays 30.7x on a year in which EPS grows 59.9%. Consensus itself has EPS growth decelerating to 21.6% and then 15.0%. A multiple appropriate to 60% growth is not appropriate to 15% growth, and semiconductor cyclicals de-rate ahead of the deceleration rather than after it. The compression from 30.7x to 23x is not a bearish assumption; it is the arithmetic of a cycle maturing.

Almost all of the expected return is earnings growth and the multiple is fighting it hard. If the multiple simply HELD at 30.7x on FY2028E the price would be $533 (+40%); if it compressed to 18x — roughly where ADI traded at the fiscal-2024 trough on then-forward estimates — the price would be $312 (−18%). The entire spread between a good and a bad outcome sits in a multiple we do not control.

The bull case at $534 holds essentially today's multiple against the FY2028 consensus HIGH — it requires no re-rating, only an earnings beat and no de-rating. The bear case at $281 is 20x the FY2027 consensus LOW of $14.034 — a de-rating to a normal analog multiple against a slightly disappointing number, which is what the last cycle turn actually looked like.

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

• We differ from the street on the durability of the gross margin, and it is the whole disagreement. The 10-Q's 67.3% against 61.0% is a fact. Our reading is that a large part of it is fab utilisation, because the same line ran at roughly 57% at the fiscal-2024 trough on 23.4% lower revenue, and because ADI's own 10-K describes internal manufacturing in five countries as a core dependency. Consensus EPS of $15.086 for fiscal 2027 requires the margin to hold. Watchable number: gross margin in the 2026-08-19 print. A figure at or above 68% argues structural mix improvement; a figure below 66.5% on sequentially higher revenue argues the utilisation ceiling has been reached.
• We differ on the balance sheet, modestly. Screens show net debt of $6.165B and 1.00x EBITDA; the corrected figures are $5.246B and 0.84x. Small in absolute terms, but it is the difference between "levered one turn" and "effectively unlevered", and it is why the risk score is 5 rather than 6.
• We think the 56.6x trailing GAAP multiple is the wrong lens and will fix itself. Intangible amortisation of roughly $1.57B a year is running off — intangibles fell from $8.014B to $7.255B in six months. GAAP earnings converge upward toward non-GAAP mechanically as the Maxim and Linear intangibles amortise, without the business improving at all. This is a real, quantifiable and generally under-discussed tailwind to the reported number.
• We have essentially no independent conviction and we say so. Zero entity-tagged claims name this company across 51,928 distilled claims. The two usable claims are category-level, and one comes from a competitor's chief executive. A name with a 5.0% base-case return, a 12.0% gap below consensus, and no conviction lane is a Watch by construction.
• Positive surprise that would force a re-rate: the 2026-08-19 print beating $3.915B with gross margin at or above 68% and a fourth-quarter guide above $4.1B, which would simultaneously validate the full-year consensus and the margin; a materially higher Automotive contribution, which is the one end market genuinely above its prior peak; or an acceleration in Industrial back toward the $6.555B it did in fiscal 2023.
• Negative surprise that would break the thesis: gross margin flat or down sequentially; distributor inventory commentary turning cautious, which matters disproportionately because distributors are 56% of revenue and one is 24%; any China revenue disruption at 25.9% of the total; or a book-to-bill below one. At 30.7x forward with the stock 73% above its 52-week low, none of these is absorbed by the valuation.

Synthos fair values

Anchors are multiples of the fiscal-2027 and fiscal-2028 consensus non-GAAP EPS distributions (FY2027 mean $15.086, low $14.034, high $16.365, 17 analysts; FY2028 mean $17.352, low $15.711, high $19.079, 6 analysts — disclosed).

• Bear ~$281 — 20x the FY2027 consensus LOW of $14.034. Cross-check: 16.2x FY2028E mean; 22.7x FY2026E. Roughly 27% above the 52-week low of $220.68, which is where this stock traded eleven months ago. The scenario: the cycle peaks in fiscal 2027, gross margin gives back 300 basis points as utilisation normalises, distributor inventory corrects, and the multiple returns to a mid-cycle analog level. −26.2%.
• Base ~$400 — 26.5x the FY2027 consensus MEAN of $15.086. Cross-check: 23.1x FY2028E; 32.2x FY2026E; corrected EV of $210.0B on FY2027E revenue of $17.099B is 12.3x sales. Sensitivity, stated openly: 24x FY2027E gives $362, 5% BELOW spot, and 29x gives $437, close to the street's median. The whole answer lives in a 24-29x band. The scenario: fiscal 2026 lands near consensus, gross margin holds in the mid-to-high sixties, growth decelerates to the mid-teens as forecast, and the multiple compresses toward the mid-twenties as it does. +5.0%, plus about 4.7% of compounded shareholder yield.
• Bull ~$534 — 28x the FY2028 consensus HIGH of $19.079, i.e. roughly today's multiple held against an earnings beat. Cross-check: 35.4x FY2027E. Just below the street's $550 high. The scenario: Industrial recovers fully to and beyond its fiscal-2023 $6.555B, Automotive content keeps compounding at the 162%-in-four-years rate, edge inference genuinely lifts analog content as the knowledge base's one relevant claim argues, and gross margin proves structural at 68%-plus. +40.2%.

Base is 5.0% above spot; asymmetry roughly 1.53:1 to the upside (26.2% down, 40.2% up). That ratio is respectable. What makes this a Watch rather than a Buy is that the base case is only 5% above the price, sits 12% below the street, rests on a margin we think is partly cyclical, is offered on a day the stock rose 5.23%, and is fifteen days from a print.

6. Knowledge base — zero claims name this company

Raw entity hits: 0. Raw text hits: 5. Used: 2. Discarded: 3.

The search covered Analog Devices, ADI, Texas Instruments and analog semiconductor across all 51,928 distilled claims. Not one claim carries "Analog Devices" or "ADI" in its entity list. This is the thinnest lane in this batch and we state it as a finding rather than dressing category claims as company conviction.

The two claims used, both text-level:

> 2025-09-20 · bullish · conviction 65 · horizon: thesis · speaker: Jordi Visser · role: independent · skill 2.0

> "Analog semiconductor names that haven't moved in four years (autos, machines, appliances) get re-rated as NPUs put decision-making brains inside humanoids, phones and cars with no latency allowed."

> 2026-06-18 · bullish · conviction 60 · horizon: thesis · speaker: Lip-Bu Tan · channel: no_priors

> "Power and thermal are bottlenecks; voltage conversion (e.g. 48V down to 1V) loses a lot of power, making power-management ICs a very good area — e.g. ADI acquiring Empower."

On the first: this is the highest-skill independent voice in the file (skill 2.0) and it is directly on ADI's thesis — edge inference requires local sensing and local power conversion. It is also gradeable and currently right in a way that argues for caution now rather than enthusiasm: it was made on 2025-09-20, when the stock was near its $220.68 low, and the "re-rating" it called for has since delivered +72% in twelve months. The claim has largely played out.

On the second: the speaker is the chief executive of a competing semiconductor company. He is not ADI management, so the half-weight management discount does not strictly apply — but he is an industry principal with a book, and we weight the claim as informed third-party colour rather than as independent analysis. Note also that the Empower acquisition it references is NOT corroborated anywhere in the filings in this archive — acquisitionsNet was only $45.7M in fiscal 2025 and the 10-Q's goodwill moved only $28.0M in the first half of fiscal 2026. We flag the claim as uncorroborated by the primary sources available to us and do not treat the transaction as a fact.

The three discards, and why they matter anyway:

• 2026-04-25 · bullish · conviction 80 · Jordi Visser (skill 2.0) · entities: ON, TXN — "Power semiconductors (On Semi, Texas Instruments) are going through the roof." Discarded: about different securities. Category-relevant and noted.
• 2026-04-26 · bullish · conviction 75 · Jordi Visser (skill 2.0) · entities: TXN — "Texas Instruments, an industrial-demand barometer, reported industrial demand improving while data-center flourishes; AI sensing/power penetrating edge form factors — robots, EVs, test equipment, not just cloud GPUs." Discarded as a TXN claim, but it is the single most useful piece of external corroboration for ADI's Industrial recovery in the file, and it is dated three weeks before ADI's own May-quarter print showed +37.2%.
• 2026-02-05 · bullish · conviction 60 · Dylan Patel · role: independent — "China has built a real semiconductor industry, insourcing more, with microcontrollers and power chips now matching and undercutting Texas Instruments/ST Micro, though still behind on leading edge." Discarded as naming other companies — but it is the only bearish structural input available and it points at exactly the 25.9% of revenue ADI ships to China.

Conclusion, stated as a finding. The Synthos knowledge base has no view on Analog Devices. It has a bullish view of the analog category from a high-skill independent voice, a bullish view of power management from an industry principal, a bullish read on industrial demand via a competitor, and a bearish read on Chinese domestic competition via two competitors. None is name-level conviction, none justifies a conviction premium in the fair value, and the thinness is a direct input to the Watch verdict. Conviction Very Low, breadth 2. No ADI management voice appears anywhere.

7. Data integrity — what we rejected from the vendor file and why

Nine findings. ADI's file is clean on segments and share count and defective on net debt, the estimate block and the price fields.

1. netDebt of $6,165.0M omits $1,152.9M of short-term investments and is six months stale — REJECTED and rebuilt. The vendor nets total debt against cash and equivalents only. The vendor reports $6.165B; the filing supports $4.939B at the same date and $5.246B at 2026-05-02; we use $5.246B. Consequential: netDebtToEBITDATTM 1.003x → 0.84x; enterpriseValueTTM $191.760B → $190.729B; evToSalesTTM 15.05x → 14.97x; evToEBITDATTM 30.79x → 30.62x.

2. est.ebitAvg is BELOW est.netIncomeAvg in every forward year — arithmetically impossible, REJECTED. FY2028: ebitAvg $4.854B against netIncomeAvg $8.150B. FY2029: $5.515B against $9.771B. FY2030: $6.491B against $11.871B. Net income cannot exceed earnings before interest and tax for a company that pays both. The ratios are additionally fixed: ebitdaAvg / revenueAvg = 44.46% and ebitAvg / revenueAvg = 25.88% in all eight years, to four significant figures. Note that the EBITDA ratio is coincidentally close to reality (actual fiscal-2025 EBITDA was 45.6% of revenue) which makes the row more dangerous, not less — it looks plausible and is constructed. All forward valuation runs on epsAvg and revenueAvg.

3. seg_geo double-counts by 1.71x in the fiscal-2025 entry, and the block is not consistent across years. The fiscal-2025 block contains a Non-US subtotal of $7,781.6M alongside the five regional lines that compose it exactly ($7,781.6M) plus United States $3,238.1M. Summing all seven fields gives $18.80B against revenue of $11.020B. The fiscal-2024 and earlier entries do not contain the Non-US line, so a naive multi-year series shows a spurious 71% jump in fiscal 2025. The correct read is US $3,238.1M + Non-US $7,781.6M = $11,019.7M, an exact tie, and that is what we use.

4. seg_prod ties exactly — a clean check worth reporting. Industrial $4,929.4M + Automotive $3,277.9M + Consumer $1,434.6M + Communications $1,377.9M = $11,019.7M against reported fiscal-2025 revenue of $11,020M. Verified clean, across all six years used.

5. quote.price ($380.80) and tech.last ($381.88) disagree by 0.28% — the quote is used and the discrepancy is flagged. The two fields are meant to be the same close. Every price-derived figure in this dive uses quote.price of $380.80, which reconciles with the market capitalisation and the filed share count to four decimal places. Every technical percentage uses the tech block, which is computed from the close series that produced the moving averages — mixing the two would make the percentages inconsistent. We state the mismatch rather than silently choosing. Related: quote.yearHigh/yearLow ($445.91 / $218.37) disagree with tech.hi52/lo52 ($445.48 / $220.68) by 0.1% and 1.1%; we use tech.

6. lastDividend: 4.18 is the trailing-twelve-month total, not the last declared dividend — corrected from the 10-Q. "On May 19, 2026, our Board of Directors declared a cash dividend of $1.10 per outstanding share." The forward run-rate is $4.40 — a 1.16% yield, not the 1.10% the field implies.

7. totalDebt differs from the filing by $72.9M — recorded, immaterial. The vendor adds capitalLeaseObligations of $72.9M to the balance sheet's $8,591.7M. Defensible, disclosed, and it changes nothing.

8. The GAAP-versus-non-GAAP gap is 46.5% and unlabelled in the payload — established explicitly. Trailing GAAP diluted EPS is $6.73 (sum of inc_q) against trailing non-GAAP of $9.86 (sum of earn_cal actuals). est.epsAvg is on the non-GAAP basis — verified because the fiscal-2025 row of $7.762 matches the sum of the four earn_cal quarterly actuals and does not match GAAP diluted EPS of $4.56. A reader comparing the vendor's priceToEarningsDilutedRatioTTM of 56.59x to our 30.7x forward figure without knowing the basis differs would draw a completely wrong conclusion about the implied growth. The driver is roughly $1.57B a year of acquired-intangible amortisation from the Linear and Maxim deals, which is quantifiable from the balance sheet (intangibles $9.585B → $8.014B → $7.255B) and is finite.

9. Vendor composite rating — mechanism disclosed, partially rejected. rating gives B− / 3, with 1 out of 5 on price-to-earnings and 2 on price-to-book and debt-to-equity, against 4 on return on assets and 3 on return on equity. The price-to-earnings 1 is computed on the 56.6x GAAP trailing figure and therefore embeds the intangible amortisation as if it were a cash cost — the non-GAAP trailing multiple is 38.6x, still high but a different judgement. The price-to-book 2 is driven by $34.2B of merger goodwill. We reject the price-to-earnings sub-score as computed and note that our own conclusion on valuation happens to agree with its direction for different reasons.

Share count — checked and verified exactly clean. The 10-Q balance sheet states 487,087,040 shares outstanding at 2026-05-02. Market cap of $185,482,729,600 divided by $380.80 gives 487,086,999 — agreement to within 0.00001%. This is the cleanest share-count verification in this batch. Diluted weighted-average shares of 490.5M in the May quarter are the normal dilution wedge.

Non-equity tripwire — checked and passed. ADI is common stock, par value $0.16⅔, NASDAQ-listed, with 471,934 preferred shares authorised and NONE outstanding per the 10-Q balance sheet. Price $380.80 is not par-like; beta 1.188; the dividend is variable and was just raised to $1.10 quarterly; volume 2.58M shares (~$982M of turnover); the 52-week band of $220.68 to $445.48 is a 102% range. This is common equity.

8. Technicals

• Price $380.80. −14.3% from the 52-week high of $445.48; +73.0% above the low of $220.68. Position within the annual range: 71st percentile. Note the width of that range — a 102% spread between high and low in twelve months is the signature of a cyclical, and it is the single best argument for waiting.
• Moving averages, split: −4.0% below a 50-day average of $396.82, which is falling; +14.8% above a 200-day of $332.59, which is rising. The long-term uptrend is intact; the intermediate trend has stalled.
• RSI 45.7 — neutral, neither stretched nor washed out. MACD −7.89 — negative, i.e. the correction since May has not resolved.
• Maximum drawdown from peak over the trailing year: −14.3% — the current price is the worst point of a shallow drawdown inside a very large advance.
• Relative performance — the best in this batch: 3-month −3.8% vs SPY +7.6% and QQQ +7.7%; 6-month +20.5% vs SPY +11.1% and QQQ +15.7%; 12-month +72.2% vs SPY +24.3% and QQQ +30.8%. A forty-eight point outperformance of the index over twelve months. That is the return that has already been earned, and it is money that is no longer available to the buyer today.
• Sentiment: 0 strong buy, 43 buy, 11 hold, 0 sell across 54 analysts. Consensus target $454.76 (+19.4%), median $450, high $550 (+44.4%), low $360 — 5.5% below spot. Well-covered, directionally near-unanimous, and with only one target below the price.

Today's move and the entry

ADI closed 2026-08-04 at $380.80, up 5.23% or $18.92 from $361.88. It opened at $372.39 — a 2.9% gap — traded $368.91 to $382.42, and closed within 0.4% of the day's high on 2.58M shares. The whole semiconductor complex moved together on this session. No company-specific news appears in this file for 2026-08-04, and the last company event was the 2026-05-20 second-quarter release. This was sector beta, not an ADI event.

The honest read, and it is unfavourable. Unlike KLAC in an adjacent batch, ADI did not gap from a deep drawdown into resistance — it gapped from 73% above its 52-week low, at the 71st percentile of its range, after a 72.2% twelve-month return, and it is still 4.0% below a falling 50-day average. Buying today means paying a 5.23% one-day move for a 5.0% base-case return, fifteen days before a print, in a cyclical at 30.7x forward. That is not a trade with an edge in it.

What would change the answer, and the numbers are specific:

• A price near $320 — roughly 21x FY2027E and 18.4x FY2028E — which would put our $400 base at +25% and make the risk/reward genuinely asymmetric. That is 16% below today's close and well inside the range this stock has traversed this year.
• Or the 2026-08-19 print landing above $3.915B of revenue with gross margin at or above 68% and a fourth-quarter guide above $4.1B, which would validate both the revenue and the margin legs of 5a at once. At that point paying 30x is defensible, because the uncertainty being paid for has been resolved.

9. Insiders — genuine open-market sales, and they are small

Unlike most of this batch, ADI's insider file contains actual S-Sale transactions on the open market, not withholding events:

The reading, carefully. These are real dispositions at real prices, which distinguishes them from the F-InKind tax-withholding events that dominate most insider files. But the sizes are trivial. Karen Golz sold 1,000 shares of an 11,019-share holding — 9.1%. Ray Stata's tranches are 12, 62, 199 and 144 shares against a holding of roughly 118,747 shares — the fragmentation across many small lots at ascending prices is the signature of an automated 10b5-1 plan executing into liquidity, not a discretionary decision.

What matters more is the absence on the other side: zero open-market purchases. No insider bought a share during a twelve-month period in which the stock rose 72%. We report the net as mildly negative and no more than that — small scheduled sales into strength by a founder-director and one other director, with no offsetting buying. The reference prices ($371-390) bracket today's $380.80, so nobody sold the top.

10. Verdict, kill-criteria and flip conditions

Watch.

What is genuinely excellent: 67.3% gross margin against 61.0% a year earlier and 38.1% operating margin against 25.7%; revenue +37.2% year on year and +14.7% sequentially in the May quarter; four consecutive non-GAAP beats with the margin of beat widening; a 38.8% free-cash-flow margin on capital expenditure of just 4.8% of revenue; corrected net debt of $5.246B at 0.84x EBITDA with 12.1x interest coverage and no preferred stock; $8.5B remaining under a $26.7B repurchase authorisation; 13,000 engineers out of 24,500 employees; an Automotive franchise up 162% in four years; and a GAAP earnings drag from intangible amortisation that is mechanically running off.

What we are declining to pay for: 30.7x the FY2026 consensus and 56.6x trailing GAAP earnings after a +72.2% twelve-month return and a +5.23% day; a 630 basis point gross-margin expansion we believe is substantially a fab-utilisation effect in a company whose margin was roughly 57% at the fiscal-2024 trough; fiscal-2025 revenue still 10.4% below the fiscal-2023 peak, with Industrial 24.8% below its own; 56% of revenue through distributors with one at 24%; 25.9% of revenue in China and rising; 71.4% of the balance sheet in Linear and Maxim goodwill, producing a 7.78% return on invested capital; zero entity-tagged knowledge-base claims; a base fair value of $400 that is 5.0% above spot and 12.0% below the street's $454.76; and a print in fifteen days.

The distinction that matters. We are not saying the upturn is fake — it is the most convincing operational recovery in this batch and it is filing-verified line by line. We are saying that a cyclical at 30.7x forward, 73% above its own twelve-month low, two weeks before a report, with our fair value 5% above the price and 12% below consensus, offers no margin. In the Synthos frame, a name where our number sits below the crowd's, the conviction lane is empty, and the entry is dominated by a scheduled event two weeks out is a Watch by construction — and the wanted price is specific.

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

• Gross margin flat or down sequentially in the 2026-08-19 or subsequent print. It has expanded every quarter for five quarters; the first flat quarter is the utilisation ceiling and it breaks the FY2027 EPS bridge.
• Revenue below approximately $3.80B on 2026-08-19 (against $3.915B consensus), or a fourth-quarter guide below $3.95B, which would put the $14.746B full-year consensus out of reach.
• Distributor inventory building, or Distributor 1 (24% of revenue) reducing orders. This is where an analog cycle turns first and 56% of revenue runs through the channel.
• China revenue disruption at 25.9% of the total, on tariffs or export controls the 10-K already names as a risk.
• Industrial revenue growth decelerating below 10% year on year. It is 44.7% of the company and still 24.8% below its fiscal-2023 peak; a stall there before the peak is recovered would mean the cycle topped early.

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

• Price at or below approximately $320 (roughly 21x FY2027E) with the operational picture intact. This is the most likely path to a Buy and it is 16% below today's close — inside the range this stock has traversed in the last six months. At $320 the base case is +25% and the bear case is 12% away.
• The 2026-08-19 print above $3.915B with gross margin at or above 68% and a fourth-quarter guide above $4.1B — both legs of 5a validated at once, at which point paying 30x is defensible.
• Industrial revenue exceeding its fiscal-2023 quarterly run-rate, which would establish the recovery as complete rather than in progress and shift the argument from cycle to slope.
• A reclaim of the 50-day average at $396.82 together with one of the above.
• Genuine name-level knowledge-base conviction appearing. There is currently none — zero claims in 51,928 — and that alone is worth roughly one tier.

Where ADI fits in the Synthos Framework Portfolio. The analog and mixed-signal semiconductor sleeve, at 0% today with a 1.5-2% target on a fill near $320 or a confirmed August print. On batch overlap: ADI and QCOM are the two semiconductor names in this batch and they are not the same trade — ADI is an industrial and automotive content-per-unit business with 67% gross margins and a fab-utilisation cycle; QCOM is a handset and licensing business with concentration risk of a completely different kind. Holding both is defensible. Holding ADI alongside a wafer-fab-equipment name would be concentration dressed as diversification, since both key off the same capital-expenditure cycle. Logged as a tracked Synthos call (Watch) as of 2026-08-04 at $380.80.

Single biggest risk: the 630 basis points of gross-margin expansion is a utilisation artefact. ADI manufactures internally in five countries. Gross margin was roughly 57% at the fiscal-2024 trough on revenue 23.4% below the prior peak, 61.0% twelve months ago, and 67.3% in the May 2026 quarter on revenue up 37.2%. That is textbook operating leverage on a fixed cost base, and operating leverage is symmetric. Consensus fiscal-2027 non-GAAP EPS of $15.086 requires the margin to hold while revenue grows a further 16%. If gross margin gives back 300 basis points as the cycle normalises, fiscal-2027 EPS is nearer $13.00 than $15.09, and at 26.5x that is $345 — nine percent below today's price. At 30.7x forward with the stock 73% above its low, there is nothing in the valuation to catch that.

Most fragile assumption in the price: that fiscal 2027 revenue of $17.099B is a plateau rather than a cycle peak. That figure is 38.9% above the fiscal-2023 high in a business whose largest end market is still 24.8% below its own fiscal-2023 level. Every other assumption in 5a is observable quarterly. This one asks a reader to believe that a company which fell 23.4% two years ago will, three years later, be running 39% above its previous best — and to pay 25.2x for the privilege before the first half of that increase has been reported.

Provenance & disclosures

• Traceability: ZERO tagged knowledge-base entity claims name Analog Devices (raw entity hits 0 across 51,928 distilled claims; raw text hits 5; used 2; discarded 3; breadth 2, net conviction positive-negligible). This is the thinnest lane in this batch and it is stated as a finding, not padded. The two claims used are text-level: Jordi Visser (named speaker, role independent, skill 2.0, 2025-09-20, bullish, conviction 65) on analog names re-rating as neural processing units push decision-making to the edge — noted as having already largely played out, since it was made near the $220.68 low and the stock is +72% since; and Lip-Bu Tan (named speaker, 2026-06-18, bullish, conviction 60, channel no_priors) on power-management integrated circuits, citing "ADI acquiring Empower" — the speaker is the chief executive of a competing semiconductor company, so this is informed third-party industry colour rather than independent analysis, and it is weighted accordingly; the Empower transaction it references is NOT corroborated by any filing in this archive (fiscal-2025 acquisitionsNet was $45.7M and first-half fiscal-2026 goodwill moved $28.0M), and we flag it as uncorroborated rather than treating it as fact. Three claims discarded as being about different securities — two Jordi Visser claims naming ON Semiconductor and Texas Instruments (2026-04-25 and 2026-04-26) and one Dylan Patel claim on Chinese domestic competition naming Texas Instruments and STMicroelectronics (2026-02-05) — all three are category-relevant and their content is reported in the body, but none is an ADI claim. No ADI management voice appears anywhere in the lane, so the half-weight management discount does not apply. Searches: Analog Devices, ADI, Texas Instruments, analog semiconductor. All quotes are verbatim from the stored claim text.
• Data as-of: fundamentals — annual statements to 2025-11-01 (fiscal 2025) from the vendor payload; quarterly income statement to 2026-05-02; most recent filing-verified financials 2026-05-02, from the 10-Q filed 2026-05-20 · estimates 2026-08-04 · prices 2026-08-04, quote timestamp 1785873542 = 2026-08-04T19:59:02Z ($380.80, +5.23%; 50-DMA $396.82; 200-DMA $332.59; RSI 45.7; MACD −7.89) · knowledge-base claims 2026-08-04. Fiscal note: ADI's fiscal year ends the Saturday nearest 31 October. "FY2025" means the year ended 2025-11-01 and "Q2 FY2026" means the quarter ended 2026-05-02. All figures come from the Synthos vendor data file for ADI or from the SEC filings in the ADI archive; no figure comes from memory, recall or external retrieval.
• Filing archive contents: 10-K filed 2025-11-25 (fiscal 2025); 10-Q filed 2026-02-18 (January 2026 quarter); 10-Q filed 2026-05-20 (May 2026 quarter); 8-K filed 2026-05-20 (second-quarter results); 8-K filed 2026-07-02; 8-K filed 2026-07-29 (reporting an event dated 2026-06-23). All six carry preserved financial tables (tables: true), which is why the balance sheet, gross-margin series, share count, buyback authority and dividend declaration in this dive are filing-sourced. The 2026-05-20 8-K is present as the cover document only and carries no guidance figures, so the fiscal third-quarter outlook is taken from the vendor's earnings calendar rather than from the company.
• Where the filings contradicted or corrected the vendor (detailed in Section 7): net debt rebuilt from $6.165B to $5.246B (the vendor omits $1.153B of short-term investments and carries a six-month-old balance sheet; the 10-Q shows cash $2,436.9M, short-term investments $1,002.4M and total debt $8,684.8M at 2026-05-02), with enterprise value rebuilt to $190.729B and net-debt-to-EBITDA corrected from 1.003x to 0.84x; est.ebitAvg rejected as arithmetically impossible — it is BELOW est.netIncomeAvg in every forward year ($4.854B against $8.150B in FY2028) and both rows are fixed at 44.46% and 25.88% of revenue in all eight years; seg_geo corrected for a 1.71x double-count (a Non-US subtotal reported alongside its own five components plus the US line, summing to $18.80B against $11.020B of revenue) with the additional finding that the Non-US line exists only in the fiscal-2025 entry, making a naive multi-year series show a spurious 71% jump; gross margin sourced from the 10-Q's stated 67.3% and 61.0% rather than derived; lastDividend of $4.18 identified as the TTM total against a filing-declared $1.10 quarterly rate, a $4.40 forward run-rate; quote.price of $380.80 and tech.last of $381.88 flagged as inconsistent, with the quote used for all price-derived figures and the tech block used for all technical percentages; and the GAAP-versus-non-GAAP basis established explicitly at a 46.5% gap driven by roughly $1.57B a year of Linear and Maxim intangible amortisation, quantified from the balance-sheet run-off ($9.585B → $8.014B → $7.255B). Where vendor and filing AGREED — worth recording: the share count (10-Q's 487,087,040 at 2026-05-02 against an implied market-cap count of 487,086,999 — agreement to five decimal places, the cleanest verification in this batch); seg_prod summing exactly to reported fiscal-2025 revenue across all six years used; total debt to within $72.9M; and the quarterly revenue series against the earnings calendar to the dollar.
• Basis note: analyst estimates in est.epsAvg are non-GAAP EPS — verified because the fiscal-2025 row of $7.762 matches the sum of the four earn_cal quarterly actuals and does NOT match GAAP diluted EPS of $4.56. The GAAP-to-non-GAAP gap is 46.5% on the trailing twelve months ($6.73 GAAP against $9.86 non-GAAP), the widest in this batch, and is almost entirely acquired-intangible amortisation. Forward multiples in this dive use consensus non-GAAP EPS, and the trailing multiple is shown on both bases (38.6x non-GAAP, 56.6x GAAP). A reader comparing the vendor's 56.59x trailing figure to our 30.7x forward figure without the basis stated would infer an implied growth rate roughly double the real one.
• Estimate coverage: 18 analysts on FY2026 EPS and 20 on revenue; 17 and 20 on FY2027; only 6 and 12 on FY2028; 2 and 6 on FY2029; 4 and 6 on FY2030 — FY2029 and FY2030 are excluded entirely. The FY2028 EPS row, used for the bull case and for cross-checks, carries 6 analysts and that is disclosed at every point of use.
• Peer note: the vendor peer set is not usable in aggregate — it includes Strategy Inc ($32.4B), a bitcoin treasury company, and Dell ($311.4B), neither of which is an analog comparable, alongside KLA (wafer-fab equipment) and the design-software names Cadence and Synopsys. The defensible peers are Texas Instruments ($259.1B), NXP Semiconductors ($60.0B), STMicroelectronics ($48.4B) and Monolithic Power Systems ($65.7B). No peer multiples are supplied, so no peer-multiple comparison is drawn.
• Fair-value caveat: the $281 / $400 / $534 anchors are multiples of the fiscal-2027 and fiscal-2028 consensus non-GAAP EPS distributions — 20x the FY2027 low of $14.034, 26.5x the FY2027 mean of $15.086, and 28x the FY2028 high of $19.079 (the last on 6 analysts, disclosed). Stated arithmetic, not a discounted cash flow. The base is sensitivity-disclosed: 24x FY2027E gives $362, five percent BELOW spot; 29x gives $437, close to the street's median. The whole answer lives in a 24-29x band and no greater precision is claimed. The bear case assumes both a de-rating and a modest earnings shortfall; the bull case assumes essentially no re-rating, only an earnings beat.
• Timing: fiscal third-quarter 2026 results are due 2026-08-19, FIFTEEN DAYS after this dive — consensus non-GAAP EPS $3.34 and revenue $3.915B, implying +35.9% year on year. This proximity is an explicit input to the Watch verdict, not an oversight. The last report was 2026-05-20, 76 days ago. 2026-08-04 was a sector-wide semiconductor up-day and ADI's +5.23% carried no company-specific news in this file.
• 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.