Adobe ADBE

Reference table

1. What the business is — and a reporting change the vendor file has not caught up with

Adobe Inc. is a $102.4B software company headquartered in San Jose, California, with 31,360 employees, listed on NASDAQ since 1986, run by chief executive Shantanu Narayen — a point Section 10 returns to. Trailing revenue is $25.198B.

The vendor's segment file describes a company that no longer exists. seg_prod reports three lines for FY2025:

Two checks, both performed. First, the three lines sum to $23.769B, exactly the reported FY2025 revenue — no double-count, no missing residual, which is the failure mode the data contract warns about most often. Second, and this is where the filing wins: the Q1 FY2026 10-Q states that "we combined our former segments—Digital Media, Digital Experience and Publishing and Advertising—into a single operating and reportable segment." Adobe now has one reportable segment; the vendor's FY2025 split is correct history with no fiscal-2026 successor.

What replaced it is thinner. The 10-Q now discloses subscription revenue across two customer groups:

Total subscription revenue for the quarter was $6.42B, so roughly $30M sits outside the two named groups. Total revenue was $6.618B, meaning 97.0% of revenue is subscription.

This is a disclosure downgrade and it should be logged as one. Investors have lost the Digital Media ARR series, which the FY2025 10-K last reported at $19.20B, up 11.5%, and with it the ability to separate the creative franchise from the marketing-software franchise. What remains is one blended ARR number and two broad customer groups. In the middle of an argument about which half of the business is exposed, the company has stopped splitting it.

Geography (FY2025) is stable and sums cleanly to reported revenue: Americas $14.120B (59.4%), EMEA $6.289B (26.5%), Asia $3.360B (14.1%). The 10-Q notes that a weaker dollar against EMEA currencies added roughly $116M to quarterly revenue, partially offset by $22M of hedging losses — so about 1.4 points of the reported 13% growth is currency.

One acquisition matters. On 2026-04-28 Adobe completed the purchase of Semrush Holdings for $1.87B in cash, described in the 10-Q as a "brand visibility platform" bought for "search engine optimization, generative engine optimization and agentic search optimization." It contributed approximately $480M of ARR. Goodwill rose from $12.86B to $14.04B as a result, partly offset by a $70M goodwill impairment on the Publishing and Advertising reporting unit — the shrinking legacy line, written down.

2. The single most important thing to understand — the order book says one thing, the price says another

For a subscription business, the honest forward indicator is not revenue, which is a backward-looking recognition schedule, and not ARR, which the company defines itself. It is remaining performance obligations: contracted revenue not yet recognised, disclosed under a standard, audited annually. Adobe publishes it every quarter, and the series is the most important table in this dive.

Read it correctly and it is unambiguous. The 10-Q states plainly: "Remaining performance obligations of $22.27 billion as of May 29, 2026 increased by 13% from $19.69 billion as of May 30, 2025." The 10-K states the same for the year-end: "$22.52 billion as of November 28, 2025 increased by 13% from $19.96 billion as of November 29, 2024." Thirteen percent growth at both the half and the year-end.

Read it carelessly and you get the opposite answer, which is why it is worth spelling out. The book has fallen $250M since the fiscal year-end, from $22.52B to $22.27B. That looks like deceleration. It is not: the prior year shows the identical shape, $19.96B at year-end falling to $19.69B at the half — a 1.4% dip against this year's 1.1% dip. Adobe's fiscal fourth quarter is its enterprise renewal quarter, so the book peaks in November and drifts through the first half. The correct comparison is year on year, and year on year the book is compounding at 13%. This year's seasonal dip is slightly shallower than last year's.

Approximately 67% of the book converts to revenue within twelve months, with the remainder thereafter — so roughly $14.9B of the next twelve months' revenue is already contracted, against a consensus FY2026 revenue estimate of $26.524B.

Now put that against the price. At $257.64 the market capitalisation is $102.41B and the corrected enterprise value is approximately $103.9B (Section 4) against trailing free cash flow of $10.63B. Solve the perpetuity — enterprise value equals free cash flow times (1 + g) divided by (r − g) — at a 9% cost of capital:

> $103.9B = $10.63B × (1 + g) ÷ (0.09 − g) → g = −1.1%

Today's price implies Adobe's free cash flow declines about 1.1% every year, forever. At a 10% discount rate the implied rate is −0.2%; at 8% it is −2.1%. Under no reasonable discount rate does the current price imply growth. This is our arithmetic, labelled as such, and it is the central fact of the dive.

A contracted book growing 13% and a price implying perpetual decline cannot both be right. One resolves within about eight quarters. That is the trade, and it is why the September print is the catalyst rather than a routine update.

3. Growth and margin trajectory — and the one ugly number

Revenue and margins, annual:

The FY2024 operating margin needs a normalisation and we make it explicitly. The 10-K states that FY2024 operating cash flow "was adversely impacted by payment of the $1 billion Figma termination fee" — the break fee on the abandoned acquisition. Adding $1.0B back to FY2024 operating income gives $7.741B, a 36.0% margin, against FY2025's 36.6% and FY2023's 34.3%. So the true trajectory is 34.3% → 36.0% → 36.6%: steady, modest expansion, not the violent swing the unadjusted line implies. The 10-K confirms the cash-flow impact; it does not itemise the expense line, so we label this an adjustment rather than a restatement.

Quarterly, where the recent deterioration lives:

Gross margin is not moving. 89.2% in the most recent quarter against 89.1% a year earlier, and inside a 88.6%-89.3% band for two years. The 10-Q places "AI inferencing costs" explicitly inside cost of subscription revenue, so whatever those costs are, they have not yet bent the line. That is a fact worth holding against the disruption case.

Operating margin is moving, and downward. 33.8% against 37.8% one quarter earlier and 35.9% one year earlier. The 10-Q gives three causes: the $70M goodwill impairment on Publishing and Advertising, loss contingencies associated with a legal settlement (the FTC/DOJ subscription-cancellation action, settled 2026-03-12 per the legal-proceedings note), and "increases in advertising expenses and, to a lesser extent, compensation costs." The impairment alone is 1.1 points of revenue. The advertising increase is not a one-off — a franchise defending its position spends more on marketing — and it is the item to watch.

The ugly number. Operating expenses grew 17% against revenue's 13%, and GAAP net income grew 1.2% — $1.712B against $1.691B. The 10-Q's own summary says net income "remained relatively flat." Yet GAAP earnings per share grew 7.6%, from $3.95 to $4.25. The entire difference is the 7.2% reduction in the diluted share count, from 433M to 402M. Say it plainly: in the most recent quarter, Adobe's earnings-per-share growth was buyback, not business.

The ARR series, with the acquisition stripped out:

The 10-Q attributes approximately $480M of the latest figure to Semrush. Strip it and ARR is $26.62B, up 10.5% — a shade below the 10.9% printed in January. Reported ARR growth accelerated; organic ARR growth did not. We report both.

Consensus, five for five. Adobe has beaten the vendor's consensus revenue in every quarter in earn_cal, and the beat has widened monotonically: +1.3%, +1.2%, +1.4%, +1.9%, +2.5%. Non-GAAP EPS beats: +$0.09, +$0.13, +$0.10, +$0.19, +$0.14. Whatever is happening to this business, it is not yet showing up as a miss.

4. Balance sheet and the enterprise-value rebuild

FY2025 audited balance sheet (2025-11-28), the clean rebuild:

• Cash and cash equivalents $5.431B · short-term investments $1.164B · total $6.595B
• Senior notes, carrying value $6.210B — which ties exactly to the 10-K's statement that "the carrying value of our senior notes was $6.21 billion, net of fair value of the interest rate swaps and unamortized discount and debt issuance costs"
• Lease obligations $438M · total debt $6.648B
• Corrected net debt: $6.648B − $6.595B = $53M. Effectively zero.
• Total assets $29.496B · total liabilities $17.873B · equity $11.623B · goodwill $12.857B

The double-counting check the data contract demands was performed and it PASSED. longTermDebt ($6.210B) and capitalLeaseObligations ($438M) are not identical, are not duplicates, and sum precisely to totalDebt ($6.648B). The long-term-debt figure ties to the filing to the dollar. This is one vendor failure mode that does not occur here, and it is worth saying so rather than only reporting the failures.

The net-debt error does occur, and it is the textbook version. The vendor reports netDebt of $1.217B for 2025-11-28. That is total debt of $6.648B minus cash and cash equivalents of $5.431B — ignoring the entire $1.164B of short-term investments. The correct figure is $53M. The vendor overstates net debt by roughly 23 times. In absolute terms $1.164B is only 1.1% of the market capitalisation, so the enterprise-value distortion is small; in characterisation terms it is the difference between "a company carrying a billion of net debt" and "a company that owes nothing net."

As at Q2 FY2026 (2026-05-29), from the 10-Q narrative:

• Senior notes carrying value $6.15B, of which $1.35B reclassified to current (the February 2027 and April 2027 maturities). The 10-Q states Adobe "intend[s] to refinance the current portion."
• Commercial paper $494M — newly drawn; there was none at the fiscal year-end
• Revolving credit facility $1.5B, entirely undrawn
• Total borrowings ≈ $6.64B, plus lease liabilities of roughly $0.44B on the fiscal-year-end basis

A stated limitation. The condensed balance-sheet tables in the extracted 10-Q text are not machine-recoverable, and the vendor payload carries no quarterly balance sheet — only three annual ones. We therefore cannot read Q2'26 cash and short-term investments directly from the filing. Derived two independent ways, they agree: the vendor's cashPerShareTTM of $13.995 against the 402M diluted count implies $5.63B; rolling the fiscal-year-end $6.595B forward through the 10-Q's disclosed cash flows (+$5.12B operating, −$1.87B Semrush, −$4.59B repurchases, +$0.494B commercial paper, less capex and net-share-settlement taxes) lands in the $5.3-5.7B range. We use ~$5.6B and label it derived.

Corrected enterprise value:

> Market cap $102.412B + borrowings $6.64B + leases ~$0.44B − cash and short-term investments ~$5.63B = ~$103.9B

Against the vendor's enterpriseValueTTM of $104.566B — an overstatement of roughly $0.7B, or 0.6%. Small, but arising from the same mechanism as the $1.164B error at the fiscal year-end.

Ratios that follow, and ratios rejected. Interest coverage 67.3x; net debt to EBITDA 0.22x; debt to market cap 6.5%. Against that, three trailing metrics are arithmetically correct and economically meaningless and are rejected as lenses: returnOnEquityTTM of 62.4%, priceToBookRatioTTM of 8.99x, and tangibleBookValuePerShareTTM of negative $8.79. All three are artefacts of the buyback: shareholders' equity fell from $16.518B (FY2023) to $14.105B (FY2024) to $11.623B (FY2025) because $25.2B of repurchases over three years ran through it, while goodwill of $14.04B now exceeds total equity. A 62% return on equity is not a quality signal here; it is a signal that the denominator has been bought away. The vendor's own rating model scores debtToEquityScore 1 and priceToBookScore 1 for exactly this reason, and those sub-scores should be disregarded.

Similarly, currentRatioTTM of 0.751 and working capital of negative $3.010B are correct and are not a liquidity warning: $7.25B of deferred revenue (10-Q) sits in current liabilities and converts to revenue rather than to a cash payment.

5. Cash flow and capital returns — the actual engine

FY2025 operating cash flow of $10.031B is confirmed by the 10-K, which reports +25% growth against FY2024 and attributes part of that to the absence of the $1B Figma termination fee. Normalising FY2024 to ~$9.06B, the underlying growth is nearer +10.7% — still healthy, and roughly in line with revenue.

Capital intensity is negligible. Capex of $179M in FY2025 is 0.75% of revenue and has fallen every year since FY2022. The company converts 98.2% of operating cash flow into free cash flow. Against a market capitalisation of $102.41B this is a 10.4% free-cash-flow yield and a 9.64x price-to-free-cash-flow multiple — for an 89% gross-margin subscription business.

The buyback is the return engine and it is enormous. FY2025 payments of $11.281B are confirmed verbatim by the 10-K ("payments totaling $11.28 billion") — vendor and filing agree. In the first half of FY2026 Adobe paid $4.59B, an annualised $9.2B. Against $102.41B that is a 9.0% gross buyback yield.

Two honest deductions. First, stock-based compensation of $1.942B (8.2% of revenue) means a meaningful share of the repurchase is offsetting dilution rather than shrinking the count; net shareholder yield is nearer 7.1%. Second, there is no dividend — dividendYieldTTM, dividendPerShareTTM and dividendPayoutRatioTTM are all zero, four years of cash-flow statements show no dividend paid, and the 10-K states "We do not anticipate paying any cash dividends in the foreseeable future." The profile.lastDividend field of $0.026 contradicts all of this and is rejected (Section 8).

Authority remaining is large and recently topped up. The 8-K dated 2026-04-21 discloses a new $25B authorisation through 2030-04-30, on top of the March 2024 programme. The 10-Q states $26.78B remained available as of 2026-05-29 — 26% of the market capitalisation. At the current pace that is roughly three years of repurchase already authorised.

The count is genuinely shrinking. Diluted weighted-average shares: 481M (FY2021) → 471M → 459M → 450M → 427M (FY2025), and quarterly from 433M (Q2'25) to 402M (Q2'26), −7.2% in a year. The 10-Q cover page reports 397.5M shares issued and outstanding as of 2026-06-11. At 7% a year the count halves in a decade — and at 9.4x forward earnings, every dollar repurchased buys back an outsized share of the earnings stream.

6. Valuation — priced in or room?

At $257.64. All forward multiples are on the NON-GAAP consensus basis, and this is a material choice, made explicit here. FY2025 consensus EPS of $20.822 against reported GAAP diluted EPS of $16.70 is a 25% gap; the quarterly actuals in earn_cal ($5.06, $5.31, $5.50, $6.06, $5.96) are non-GAAP and consistently 30-40% above the GAAP figures in inc_q. The estimate block is non-GAAP throughout.

Estimate-coverage caveat, stated before anything is built on it. Coverage is deep near-term (22-26 analysts on FY26E and FY27E) and thin beyond (9 on FY2028 EPS, 6 on FY2029, 7 on FY2030). The FY2028 EPS range is $28.78 to $36.60 — a 27% spread. And the FY2029 row is internally odd: EPS growth of +2.7% against revenue growth of +9.2%, despite an ongoing 7% annual share-count reduction, which is arithmetically implausible and reflects a six-analyst sample. The FY2027 row is our primary forward anchor; the FY2028 row is used only for the base-case cross-check and is flagged; nothing rests on FY2029 or FY2030.

Peer context: none available. The peers block supplies price and market capitalisation only — no multiples — so no peer-multiple comparison is drawn. The set is also poorly constructed for this purpose: KLAC, TXN, ADI, DELL and SONY are semiconductors and hardware, not application software. The only genuinely comparable names in it are Synopsys and, loosely, the security vendors.

The vendor's own rating model scores Adobe B+ with an overall score of 3, discounted-cash-flow score 4, return-on-equity 5, return-on-assets 5 — dragged down to B+ by the debt-to-equity and price-to-book sub-scores of 1, which Section 4 explains are buyback artefacts.

6a. What today's price assumes (the inversion)

This is the section that matters most for this name, because the street's consensus target of $261.16 sits 1.4% above a spot of $257.64 — the lowest implied upside in this batch by a factor of six, with the next-lowest at +9.4%. A consensus that has converged on the current price is a statement, and it needs decoding. The rating distribution decodes it: 31 buy, 27 hold, 5 sell across 63 analysts — a constructive majority. But the median target is $245, which is 4.9% BELOW spot, while the mean is $261.16. Mean above median means right-skew: a minority of high targets (the range runs to $379) dragging the average up over a bunched middle sitting at or under the market. The street's ratings say own it. The street's targets say it is fully valued or worse. That disagreement is the finding, and it is the signature of analysts who cannot underwrite the terminal value and have let the price set the target.

At $257.64 — 9.37x FY27E non-GAAP EPS, a 10.4% free-cash-flow yield, a corrected enterprise value of ~$103.9B — the price embeds the following falsifiable claims:

• Free cash flow declines about 1.1% a year in perpetuity. (Our arithmetic, from corrected EV $103.9B, trailing FCF $10.63B and a 9% cost of capital. At 10% the implied rate is −0.2%; at 8%, −2.1%.) Under no plausible discount rate does this price imply growth. Falsifier: FY2027 free cash flow above ~$10.6B, which consensus revenue and margin assumptions comfortably imply.
• The contracted book stops compounding. (Filing-derived: remaining performance obligations $22.27B at 2026-05-29, +13% from $19.69B; $22.52B at 2025-11-28, +13% from $19.96B.) A price implying perpetual decline is inconsistent with two consecutive 13% readings on contracted forward demand. Falsifier, with a date: remaining-performance-obligation growth below ~8% at the 2026-09-10 print.
• Consensus is wrong beyond about two years. (Consensus-derived: revenue $26.524B FY26E to $37.330B FY30E, +8.9% compound; EPS $24.402 to $34.840, +9.3% compound.) The price pays 7.4x FY2030 consensus EPS. Either the estimates are fiction beyond FY2027 or the multiple is. Falsifier: whether the FY2028 EPS mean of $31.422 holds through the September revision cycle.
• The share-count reduction stops or is offset. (Filing-derived: 433M diluted shares Q2'25 to 402M Q2'26, −7.2%; $4.59B repurchased in H1 FY2026; $26.78B of authority remaining.) At an unchanged multiple, a 7% annual count reduction alone delivers a 7% annual return. Falsifier: quarterly repurchase spend below ~$2B.
• THE MOST FRAGILE ASSUMPTION: that gross margin holds near 89% while inference costs scale. (Filing-derived and our reading.) Gross margin is 89.2% and has not moved, and the 10-Q places AI inference costs inside cost of subscription revenue — so far, so good. But the same filing warns that developing and deploying AI models "may continue to increase the cost of our solutions" and that "these costs could adversely impact our margins." Operating expenses already grew 17% against revenue's 13%. This is the leg that could break quietly, over several quarters, without a headline. Falsifier: gross margin below 87% for two consecutive quarters.

6b. The return bridge (why the multiple moves)

Expected return over 12-24 months ≈ EPS growth (+12.7%) + multiple drift (assumed FLAT) + net shareholder yield (~7.1%).

Our base case assumes NO multiple expansion whatsoever, and that is deliberate. At $257.64 the stock trades at 9.37x FY2027 consensus non-GAAP EPS of $27.509. Hold that identical multiple and roll forward to FY2028 consensus EPS of $31.422 and the price is $294 — within 1% of our $296 base. The base case requires only that the estimates are roughly met and that the multiple stops falling. It does not require the crowd to change its mind, and given that our knowledge base says the crowd is right, that matters.

That is the cleanest statement of the opportunity: at 9.4x forward earnings with a 9% gross buyback yield, you are paid by earnings growth and share-count reduction alone. A re-rating is upside, not thesis.

The bull case at $385 does require re-rating, to roughly 14.0x FY27E or 12.3x FY28E, and that leg is genuinely fragile — it needs the terminal-value argument to be visibly retired, which one quarter cannot do. The bear case at $190 needs no estimate cut at all: 6.9x FY27E on unchanged consensus, i.e. pure further de-rating, or equivalently a ~30% estimate cut at today's multiple. Note that $190 is also the vendor's targetLow and sits 1.8% below the 52-week low of $193.41 — the market has already traded there this year.

Guardrail applied. The rolldown from 10.56x FY26E to 9.37x FY27E to 8.20x FY28E at a constant price is mechanical, not compression. We call no re-rating in the base case and label the bull case's expansion explicitly.

6c. Variant perception (where we differ, what would surprise)

• We differ on the weight of evidence, not the direction of the risk. The bear case is a terminal-value argument with, as of today, zero numerical corroboration in the filings: remaining performance obligations +13%, ARR +12.5%, revenue +13%, gross margin 89.2% and stable, five consecutive revenue beats with the beat widening from +1.3% to +2.5%. That is not what disruption looks like in the accounts — yet. The concrete surprise that would reprice this negatively: remaining-performance-obligation growth below 8% on 2026-09-10. Positively: growth holding at or above 13% with gross margin intact.
• We differ from the street's targets, not its ratings. Sixty-three analysts are 49% buy and only 8% sell, yet the median target sits below the market. Our base of $296 is 13% above the mean target and 21% above the median. We think the target set has been anchored to a falling price rather than to a book that is growing, and that the $190-$379 dispersion is the honest admission that this is a binary the sell side cannot resolve. Watchable: whether targets are raised after the September print, which would be the first genuine sentiment change in a year.
• We do NOT differ on the central weakness, and this is why the verdict is Watch rather than Buy. Adobe discloses no generative-monetisation metric of any kind — we searched the FY2025 10-K and both fiscal-2026 10-Qs for credit consumption, AI-attributable revenue, or usage-based generative pricing, and found nothing beyond boilerplate that some offerings are billed on transaction volume. The company that must prove generative tooling expands its market publishes no number that would show it doing so. We cannot have a strong variant perception on a question the disclosure does not address.
• A negative surprise already in motion: leadership. The 8-K dated 2026-06-11 discloses that chief financial officer Daniel Durn resigned effective 2026-06-15, with Steven Day appointed interim. The 8-K dated 2026-07-17 discloses a retention letter for the general counsel whose protections "sunset 12 months following the date the new Chief Executive Officer commences employment with the Company" — an unambiguous statement that a chief-executive transition is underway, with no successor named in any filing here. Two of three top seats unsettled, going into the print that decides the argument. This is a filing-sourced fact absent from the vendor file entirely, and it is a direct input to the risk score and the verdict.
• A negative surprise the company itself wrote down. The competition risk factor was materially expanded in the Q1 FY2026 10-Q relative to the FY2025 10-K, adding that AI is "transforming how digital work is performed or automated and how, and the extent to which, our solutions are accessed and used," extending competition to "AI-enabled workflows, some of which are embedded within or operating across third-party AI platforms or domain-specific solutions," and extending the interference language to cover third parties limiting Adobe's ability to "make, use or sell our own AI solutions." A company does not add that language about a risk it considers remote. Half-weight is not applicable here — this is a risk factor, not guidance, and risk factors carry the opposite bias to promotional disclosure.
• A positive surprise that is cheap and available: an accelerated repurchase against the $26.78B authority at 9.4x forward earnings, or the first disclosure of a consumption-based generative metric. The latter would convert this from Watch to a position in a single filing, because it would make the unmeasurable measurable.

Synthos fair values

All three anchors are struck on consensus NON-GAAP EPS and cross-checked against the free-cash-flow perpetuity implied at a 9% cost of capital.

• Bear ~$190 — 6.9x FY27E EPS / 6.0x FY28E. Implied perpetual free-cash-flow growth: −4.3%. Cross-check: 1.8% below the 52-week low of $193.41 and exactly the vendor's targetLow. The scenario: seat counts erode, professional pricing power breaks, remaining performance obligations decelerate below high single digits, and gross margin gives way under inference costs. −26.3%.
• Base ~$296 — 10.8x FY27E EPS, or equivalently 9.37x FY28E EPS at today's exact multiple, requiring no re-rating at all. Cross-check: a zero-growth perpetuity on trailing free cash flow of $10.63B at a 9% cost of capital gives an enterprise value of $118.1B, or $294 a share after net debt — within 1% by an entirely independent route. The base case values Adobe as a business that never grows again and never shrinks. Given a book compounding at 13%, that is a conservative anchor. +14.9%.
• Bull ~$385 — 14.0x FY27E EPS / 12.3x FY28E. Implied perpetual free-cash-flow growth: +2.0% — still below inflation. Cross-check: 1.6% above the vendor's targetHigh of $379 and 4.8% above the 52-week high area. The scenario: generative monetisation is disclosed and is additive, remaining-performance-obligation growth holds, a credible chief executive is named, and the multiple re-rates toward the low teens. +49.4%.

Base is 14.9% above spot; asymmetry roughly 1.9:1 to the upside (26% down, 49% up). That asymmetry, on its own, argues for a position. What holds it at Watch is that the KB lane is net negative, the disclosure gap is total, the C-suite is unsettled, and the resolving event is 37 days away — so the cost of waiting is 37 days of carry and the cost of being early is a retest of $193.

7. Knowledge base — 65 raw hits, 43 used, 22 discarded, and a net stance that is NEGATIVE

Raw hits: 65 on the terms ADBE, Adobe and Figma. Used: 43. Discarded: 22. We report the discards in full because the composition of this lane is as informative as its direction.

Discards, by reason:

1. Nine claims are about Figma, not Adobe — entity Figma with no Adobe or ADBE entity. Figma is a separate listed company (the KB tracks its 2025 IPO at a ~$25-26B valuation) and Adobe's failed acquisition target; Adobe paid a $1B termination fee on the abandoned deal, confirmed in the 10-K. These are competitive colour, not claims about Adobe, and they are excluded from the stance tally.

2. Nine claims carry thinker: quarantine_misattributed — the house attribution quarantine. All are entity or text matches on Adobe, all bearish, all with document identifiers pointing to the same speaker as the properly-attributed claims. We discard them rather than double-count an unresolved attribution — and note that discarding them does NOT change the net stance, because the same view is represented by fifteen correctly-attributed claims.

3. Two claims are same-appearance duplicates — three near-identical theses dated 2026-11-16 from one video identifier, split across jordi_visser, jordi_visser_ai and jordi_visser_m. One counted, two discarded.

4. Two match: text claims are about other companies entirely — one is a Palantir/Snowflake buy thesis in which Adobe is the foil; one is a Datadog/Fortinet rotation claim. Excluded.

Of the 43 used: 34 are match: entity (first-class) and 9 are match: text (supporting colour only, and treated as such).

Stance of the 43 used: 24 bearish · 14 bullish · 5 neutral. The net stance is NEGATIVE. Skill-weighted it is worse than the count suggests, roughly 3:1 bearish, because the negative claims cluster on a skill-2.0 voice and the positive claims sit at skill 0.9-1.1.

The bearish lane — deep, high-skill, and narrow.

Eighteen of the twenty-four bearish claims are attributable to one speaker, Jordi Visser (speaker_role: independent, skill 2.0), appearing across three distinct channels — his own, Anthony Pompliano's, and Bankless. Where speaker and channel differ we attribute to the speaker and name the channel separately, per house policy; several of these hits carry doc_id: anthony_pompliano-* with speaker: jordi_visser, and it is the speaker who owns the view.

> 2026-02-07 · bearish · conviction 85 · speaker Jordi Visser (channel: Anthony Pompliano) · entities: CRM, ADBE, MSFT, IGV

> "Enterprise SAS (Salesforce, Adobe, Microsoft) is directly at odds with AI-native agents; multiples must be re-rated down like Macy's — companies survive but stop growing with nominal GDP."

> Test attached: "Enterprise-software names (Salesforce, Adobe) should NOT bounce significantly; their earnings/revenues decouple from nominal GDP."

> 2025-09-07 · bearish · conviction 72 · Jordi Visser · entity: ADBE

> "Adobe's ~$13B Creative Cloud revenue from 37M subscribers is threatened by free [consumer generative image tool — product name withheld per house style], which preserves character likeness across edits and eliminates the learning curve."

> 2026-07-19 · bearish · conviction 78 · Jordi Visser · entities: ADBE, CRM

> "Adobe and Salesforce have downward-sloping 200-day moving averages — 'these stocks are toast'; may bounce to the 200-day but must pivot their business, disrupted by AI agents."

> Test attached: "200-day moving averages keep declining; no durable recovery without business pivot."

Two things must be said about this lane, and they point in opposite directions.

First, it has been right. The stock is −25.9% over twelve months and −62.5% from its peak. The 200-day moving average at $275.66 is still declining and the price is 6.5% below it — the July 19 claim's own test is currently passing. A conviction lane whose pre-registered tests are passing is not one to wave away, and it is the primary reason this dive is not a Buy.

Second, its conviction is collapsing. Tracking the same speaker's conviction on ADBE entity matches through time: 72 (Sep 2025) → 80 (Jan 2026) → 85 (Feb 2026, the peak) → 80 (Apr) → 80 (18 Jul) → 78 (19 Jul) → 66 (25 Jul) → 45 (2 Aug 2026). The most recent claim, two days before this dive, is the lowest conviction he has held on the name in a year, and its framing has softened from "toast" to a valuation observation:

> 2026-08-02 · bearish · conviction 45 · Jordi Visser · entities: ADBE, CRM

> "Software built on code (Adobe, Salesforce) faces terminal-value uncertainty as AI compresses their durability — can fall 40% without earnings changing, driving multiple compression."

That claim is, read carefully, almost our own thesis with the sign flipped. "Can fall 40% without earnings changing" is a statement that the move is about terminal value rather than about the numbers — which is exactly what Section 6a's inversion shows the price already reflects.

Other bearish voices are thinner but independent: Chamath Palihapitiya (all_in, 2026-04-24, conviction 80) on enterprises spinning up bespoke agent alternatives to vertical software and cutting renewals 50-75%; compound_and_friends (2026-04-10, conviction 65) on a software "bloodbath" in which "the market ignores record earnings and prices a much lower terminal value"; and, dating back much further, Dan Nathan (2022-03-25, conviction 60) flagging Adobe's sales multiple as unsustainable at a ~$220B capitalisation — a call that has since been vindicated by a 53% decline in market value.

The bullish lane — broader in voices, lower in conviction, and closer to the price.

> 2026-06-21 · bullish · conviction 50 · speaker Stig Brodersen · we_study_billionaires · entity: ADBE

> "Adobe near a 7-year low (~$270, was $224 recently), ~$100B cap, 96% of $23B revenue subscription, 41M+ paying users, switching-cost moat — unloved SaaS with a lot to love despite AI-disruption fear."

> 2025-12-04 · bullish · conviction 75 · speaker Sean Ali (guest) · we_study_billionaires · entity: ADBE

> "Adobe's core business isn't disrupted by generative AI — Pixar won't use [general-purpose chat model — name withheld per house style]; results stay stellar yet the stock fell over a third in a year with nothing in the numbers to justify it."

> 2026-04-14 · bullish · conviction 55 · lyn_alden · entity: Adobe

> "Adobe bull case: cheap, quality product, record revenue/FCF, Firefly AI usage growing ~50% QoQ — business gets resized, not made irrelevant."

> 2026-01-19 · bullish · conviction 60 · speaker Alex Rampell (host) · a16z · entities: SAP, ADBE, WDAY, BILL

> "Incumbent software systems-of-record (bill.com, SAP, Adobe, Workday) get stronger with AI — they can charge for new AI features and keep pricing power because customers are 'hostages'."

The bullish lane's highest conviction reading is 75 against the bearish lane's 90, and its three most recent entries sit at 50-55. It is a lane of measured constructiveness, not enthusiasm. Note also that the 2026-04-14 claim's "Firefly AI usage growing ~50% QoQ" is the only quantification of generative adoption anywhere in this dive, and it comes from a knowledge-base claim rather than from Adobe — a point that reinforces Section 6c's complaint about disclosure.

Two stance tags in this file are wrong and we correct them rather than inherit them. The 2024-05-02 macrovoices claim is tagged bullish, conviction 75, but its text — "AI image tools crossed the uncanny valley to photorealism — an 'atom bomb' disrupting the illustration, photography and design industries" — is a disruption claim, not an endorsement. And the 2026-03-03 compound_and_friends claim is tagged bullish while stating "avoided Adobe after getting cut catching that knife", which is explicitly negative on Adobe specifically. Corrected, the split would be 26 bearish / 12 bullish / 5 neutral — worse, not better. We report the uncorrected tally in the front matter and the corrected reading here.

A management-voice check. No claim in this lane carries speaker_role: "management" or an Adobe executive as speaker, so no half-weighting is required. One claim (a16z) is host-attributed at a venture firm with a structural interest in incumbent software surviving — flagged, not discounted.

Why this lane produces Low-Moderate conviction with a negative sign. Twelve sources and 43 claims is genuine breadth of coverage, but the negative direction is carried disproportionately by one speaker repeating a consistent view across channels — depth of repetition, not independent agreement. Set against that, the view has been right on the price, has a pre-registered test that is currently passing, and is echoed independently by at least three other voices. We treat the lane as a material headwind to conviction and a direct input to the Watch — and we note, without over-reading it, that the loudest bear's conviction has halved in five weeks.

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

Fifteen findings. Three are places a filing contradicted the vendor; several are checks that passed and are reported as such.

1. seg_prod describes a segment structure that no longer exists — FILING WINS. The vendor reports Digital Media / Digital Experience / Print and Publishing for FY2025. The Q1 FY2026 10-Q (filed 2026-03-25) states: "In the first quarter of fiscal 2026, we combined our former segments—Digital Media, Digital Experience and Publishing and Advertising—into a single operating and reportable segment." Adobe has ONE reportable segment from FY2026. The vendor history is correct and will have no successor. Replacement disclosure is subscription revenue by two customer groups (Section 1). We use the vendor split for FY2025 history and the 10-Q customer groups for FY2026.

2. Segment double-count check — PERFORMED, PASSED. The three FY2025 vendor segment lines sum to $23.769B, exactly the reported revenue. No sub-segment is listed both inside a parent and standalone. seg_geo likewise sums to $23.769B. Both are clean.

3. netDebt omits short-term investments — FILING WINS. bal_a reports netDebt of $1.217B at 2025-11-28, computed as total debt $6.648B less cash and cash equivalents only ($5.431B), ignoring $1.164B of short-term investments. The correct figure is $53M — effectively zero net debt. The vendor overstates by roughly 23x. Corrected enterprise value ~$103.9B against the vendor's $104.566B.

4. longTermDebt / capitalLeaseObligations double-count check — PERFORMED, PASSED. longTermDebt of $6.210B ties exactly to the 10-K's "carrying value of our senior notes was $6.21 billion." capitalLeaseObligations of $438M is a distinct lease liability. totalDebt of $6.648B is their sum. No duplication. This is the failure mode the data contract flags most often and it does not occur here.

5. Vendor debt and cash are STALE by two quarters — FILING WINS. The payload contains no quarterly balance sheet (three annual only), so the most recent balance data is 2025-11-28. Since then, per the 10-Q: $494M of commercial paper issued (there was none at year-end), $1.35B of senior notes reclassified to current, and $1.87B spent on Semrush. Q2'26 borrowings are ~$6.64B. Cash and short-term investments at Q2'26 are not extractable from the filing text and are derived at ~$5.6B by two independent routes (Section 4), labelled as derived.

6. profile.lastDividend of $0.026 is false — REJECTED. It contradicts dividendYieldTTM (0), dividendPerShareTTM (0), dividendPayoutRatioTTM (0), four consecutive years of cash-flow statements showing no dividend, and the 10-K's explicit "We do not anticipate paying any cash dividends in the foreseeable future." Almost certainly a decades-old artefact. Adobe pays no dividend; all shareholder return is repurchase.

7. Estimates are NON-GAAP; reported EPS is GAAP — basis stated, one row rejected. FY2025: epsAvg $20.822 against GAAP diluted $16.70, a 25% gap. Quarterly earn_cal actuals ($5.06 / $5.31 / $5.50 / $6.06 / $5.96) are non-GAAP, 30-40% above the matching inc_q GAAP figures. All forward multiples in this dive are on the non-GAAP basis and labelled. The 2023-11-28 estimate row (epsAvg $11.203) is REJECTED: it sits below that year's GAAP actual of $11.83, which is inconsistent with every other row and indicates a stale or mixed basis.

8. est.ebitdaAvg and ebitAvg are mixed-basis — NOT USED. FY2025 ebitdaAvg of $9.500B is close to GAAP operating income plus D&A ($8.706B + $818M = $9.524B), while the same row's niAvg of $8.729B is clearly non-GAAP (GAAP net income was $7.130B). A single estimate row cannot be GAAP on EBITDA and non-GAAP on net income. Less corrupt than the ~70% failure rate the data contract warns of, but unusable. No EV/EBITDA multiple appears in this dive.

9. tech and quote disagree on the 52-week band and the moving averages — both reported, tech used. tech gives hi52 $367.46 / lo52 $193.41; quote gives yearHigh $370.86 / yearLow $190.12. tech gives sma50 $228.57 / sma200 $275.66; quote gives priceAvg50 $228.34 / priceAvg200 $276.45. tech.last of $257.81 also differs from quote.price of $257.64. We use tech for the band and averages (it is computed from six years of closes) and quote.price as the strike, and state both.

10. Book-value and return-on-equity metrics are arithmetically right and economically meaningless — REJECTED as lenses. returnOnEquityTTM 62.4%, priceToBookRatioTTM 8.99x, tangibleBookValuePerShareTTM negative $8.79. All three are buyback artefacts: equity fell $16.518B → $14.105B → $11.623B across FY2023-FY2025 as $25.2B of repurchases ran through it, while goodwill of $14.04B now exceeds total equity. The vendor's own debtToEquityScore of 1 and priceToBookScore of 1 rest on these and should be disregarded.

11. currentRatioTTM 0.751 and workingCapitalTTM −$3.010B are correct and are NOT a liquidity signal. $7.25B of deferred revenue (10-Q) sits in current liabilities and converts to revenue, not to a cash outflow. Not used as a solvency input.

12. Market capitalisation and share count — CHECKED AGAINST THE FILING, CONFIRMED. $102.4119B ÷ $257.64 = 397.50M shares implied. The 10-Q cover page states: "As of June 11, 2026, 397.5 million shares of the registrant's common stock… were issued and outstanding." Exact agreement. The Q2'26 diluted weighted-average count of 402M is a different measure over a different period and is not a contradiction. Given how often vendor share counts fail on heavy repurchasers, this is worth stating as a pass.

13. earn_cal EPS is non-GAAP while inc_q.epsdiluted is GAAP — both used, each labelled. Q2'26: $5.96 versus $4.25.

14. peers supplies no multiples and a poor comparison set — NO peer comparison drawn. Price and market capitalisation only. Five of ten names (KLAC, TXN, ADI, DELL, SONY) are semiconductors or hardware.

15. MISSING ENTIRELY: any generative-monetisation disclosure. We searched the FY2025 10-K and both fiscal-2026 10-Qs for a generative-credit metric, AI-attributable revenue, consumption-based pricing detail, or any adoption quantification. There is none, beyond a generic note that some offerings are billed on transaction volume. Stated as a gap, not estimated around. It is also the single most consequential gap in this dive.

Not defects, correctly reported: capexToRevenueTTM of 0.80% is genuine (capex $179M on $23.769B in FY2025); stockBasedCompensationToRevenueTTM of 8.05% ties to the $1.942B in the cash-flow statement; interestCoverageRatioTTM of 67.3x is consistent with $6.15B of notes against $9.09B of trailing operating income.

9. Technicals — mid-recovery, below a declining long average

• Price $257.64, -29.8% below the 52-week high of $367.46, +33.3% above the 52-week low of $193.41. Maximum drawdown from the all-time peak: -62.5% — this equity has lost nearly two-thirds of its value from top to bottom, which is the single most important thing to know about its volatility.
• Above the 50-day, below the 200-day. +12.7% over a rising 50-DMA of $228.57; -6.5% under a 200-DMA of $275.66 that is still declining. The knowledge base's most specific bearish test names precisely this configuration.
• RSI 64.5 — elevated but not extreme. MACD +8.42 and positive. Momentum is genuinely positive on the short measures and negative on the long one.
• Relative performance: 3-month +1.5% vs SPY +7.6% and QQQ +7.7% (a 6-point deficit even during the recovery); 6-month -12.1% vs SPY +11.1% (-23 points); 12-month -25.9% vs SPY +24.3% and QQQ +30.8% (-50 to -57 points).
• Volume 3.20M shares. The payload supplies no average-volume figure, so no relative-volume conclusion is drawn.
• Non-equity tripwire — checked and passed. Common stock, $0.0001 par value, NASDAQ-listed since 1986, beta 1.433, no dividend, 52-week range spanning 90% of the low, market capitalisation $102.4B. This is not a preferred, a baby bond or a tracking unit inheriting parent fundamentals.

Today's move and what it does to the entry

ADBE closed +2.51% at $257.64 on a day when the semiconductor complex went vertical. In this twelve-name batch, INTC rose 10.9%, MRVL 12.8%, LRCX 7.9%, MU 7.6%, AMD 7.0%, KLAC 7.0%, AMAT 5.5% and ASML 4.2%. Adobe, Salesforce (+2.71%) and Accenture (+2.88%) — the three software-and-services names — captured roughly a quarter of that. On the single most emphatic buildout day of the batch, the software complex managed a fifth of the semis' move. That is the "long the buildout, short the software" configuration playing out inside one session, and it is the same configuration our knowledge base has been describing for a year.

But the intraday structure is the opposite of a gap you are being asked to chase. Adobe opened at $247.81, 1.4% below the prior close of $251.34, traded down to $245.46 (−2.3% on the session), and then rallied 5.0% off that low to close at $257.64, within $0.90 of the session high of $258.54. A gap-down that closes at the high of the day is an accumulation session, not a squeeze. Nobody chased Adobe today; buyers absorbed a morning decline.

So the honest answer on entry is: no, you are not being asked to buy into a one-day gap here — and that is not the reason to wait. The reasons to wait are structural and they are three. First, the name has already advanced 33.3% off its 52-week low, so the easy part of the mean reversion is behind it; you are entering the middle of a recovery leg, not the bottom of a capitulation. Second, it remains 6.5% below a 200-day moving average that is still falling, which means the primary trend has not turned and the KB's pre-registered technical test is still passing. Third and decisively, the event that resolves the argument — the remaining-performance-obligation and ARR readings in the 2026-09-10 print — is 37 days away, and the cost of waiting for it is 37 days of a 10% free-cash-flow yield you do not yet own, against the cost of being early, which is a retest of the $193 low, 25% lower.

Practical read: this is a Watch with a date, not a Watch without one. The trigger is the September print, and the specific number is remaining performance obligations. Reclaiming the 200-DMA at $275.66 on that print would be the first genuine trend change in a year and would convert this to a staged buy. Losing $228.57 (the 50-DMA) before it would say the bears' 200-day test is holding and the entry should move materially lower.

10. Insiders — one small sale, and a chief executive who has not sold anything

Eight transactions are on file. Seven of them belong to one person on one day and none is a discretionary sale.

The reading, carefully. The chief executive's seven entries are 8,298 shares acquired at $0 (option and restricted-stock-unit vesting) against 3,022 shares withheld in kind at $224.56 to cover the tax on that vesting. Net position change: +5,276 shares, ending at 365,140 held indirectly. These are mechanical compensation events, not signals — but the absence is worth naming: there is no open-market sale by the chief executive in this file at all, in a year when the stock has been as high as $367 and as low as $193.

The only discretionary transaction is a 416-share sale by the chief accounting officer at $264.33 on 2026-07-29, roughly 9.8% of her 4,240-share position and worth about $110,000. It is too small to signal anything and should not be over-read in either direction.

What is genuinely material sits in the 8-Ks, not in the Form 4s. The 2026-06-11 8-K discloses that chief financial officer Daniel Durn resigned effective 2026-06-15, with Steven Day — a twenty-year internal finance executive — appointed interim, effective immediately. No permanent appointment appears in any filing here. The 2026-07-17 8-K approves a retention letter for general counsel Louise Pentland whose severance protections "sunset 12 months following the date the new Chief Executive Officer commences employment with the Company." That sentence discloses, indirectly but unmistakably, that Adobe is hiring a new chief executive and has not announced one. The 2026-04-21 8-K's annual-meeting results show all eleven directors re-elected and say-on-pay approved, so this is not a contested-governance situation — but a company entering the most consequential strategic debate in its history with an interim finance chief and an unnamed incoming chief executive is carrying execution risk that no vendor field captures.

11. Verdict, kill-criteria and flip conditions

Watch.

What is true and good. An 89.1% gross margin. A 42.2% free-cash-flow margin producing $10.63B of trailing free cash flow on $179M of annual capital expenditure. Effectively zero net debt once the vendor's $1.164B short-term-investment omission is corrected. 97% subscription revenue. A contracted forward book of $22.27B, up 13% year on year at both the half and the fiscal year-end. Five consecutive revenue beats with the beat widening from +1.3% to +2.5%. A share count falling 7.2% a year against $26.78B of remaining repurchase authority. And a price of 9.37x FY2027 consensus non-GAAP earnings which, inverted, assumes free cash flow declines 1.1% a year in perpetuity.

What is true and bad. The knowledge base is net negative — 24 bearish against 14 bullish — and the negative lane has been right about the price, with a pre-registered technical test that is currently passing. GAAP net income grew 1.2% last quarter on 13% revenue growth, with operating expenses up 17% and operating margin down 400 basis points sequentially. Adobe collapsed three reporting segments into one and stopped publishing Digital Media ARR, reducing disclosure exactly when scrutiny rose. It materially expanded its competition risk factor in the Q1 FY2026 10-Q. It publishes no generative-monetisation metric of any kind. Its chief financial officer resigned in June and an unnamed chief executive is being hired. And the street, while 49% buy-rated, has a median price target 4.9% below the current price.

Why Watch and not Buy — Tactical. The asymmetry (1.9:1) and the valuation would ordinarily support a staged position. Three things override it. One: the central question — whether generative tooling expands Adobe's market or collapses its moat — cannot be answered from Adobe's own disclosure, and buying a name whose decisive variable is undisclosed is speculation dressed as analysis. Two: the resolving event is 37 days away and carries the exact number that matters. Three: the stock has already run 33% off its low into a still-declining 200-day average, so the timing premium for waiting is small and the timing risk of being early is a 25% retest. Waiting 37 days for a number that settles the thesis is not indecision; it is the correct use of a known catalyst.

Pre-registered FLIP TO BUY — TACTICAL (the specific trigger):

• Remaining performance obligations at the 2026-09-10 print growing 11% or better year on year (the last two readings were both +13%). This is the single trigger. It says the contracted book has not cracked, and at 9.4x forward earnings that is enough.
• AND gross margin at or above 88% in the same print, confirming inference costs are not eating the model.
• Either of these alone is insufficient; together they convert the position. Add: a named permanent chief executive would allow a larger initial size.

Pre-registered FLIP TO BUY — CORE (upsize):

• First disclosure of a consumption- or credit-based generative revenue metric showing meaningful scale. This would convert the unmeasurable to the measurable and is the highest-value single event available.
• An accelerated repurchase materially above the ~$2.3B quarterly run rate at these prices, signalling management agrees with the arithmetic.
• Reclaim of the 200-day moving average at $275.66 on the print, retiring the knowledge base's own technical test.
• A price below ~$200 (near the $193.41 low) with the forward book still compounding — at which point the bear case is fully paid for and the position should be built rather than reconsidered.

Pre-registered KILL criteria — what removes this from the watchlist entirely:

• Remaining-performance-obligation growth below 8% at the September print, or any sequential decline that exceeds the ~1.1-1.4% seasonal pattern established in the last two years. This is the primary kill.
• Gross margin below 87% for two consecutive quarters, indicating inference costs are structurally impairing the model — the most fragile assumption in the price (Section 6a).
• Organic ARR growth (ex-Semrush) falling below 8%.
• A downward revision cycle taking FY2028 consensus EPS below ~$27, which would put the current price at 9.5x FY2028 rather than 8.2x and remove the valuation support.
• The repurchase pace falling below ~$2B a quarter while the stock sits under 12x forward earnings, which would signal either that management does not share the valuation view or that cash is needed elsewhere.
• A permanent chief executive appointed from outside with a mandate to restructure, which would introduce a multi-year transition of a kind this valuation does not price.

Where ADBE would fit in the Synthos Framework Portfolio. The quality-value / cash-return sleeve, at a 1-1.5% initial position on the trigger, not before, built over 2-4 weeks after the print rather than in one trade. It is explicitly not a growth or exponential allocation — Exponential Potential is a 3 — and it should not be sized as one.

On overlap within the batch. Adobe and Salesforce are the two software names here and the knowledge base treats them as a single trade with two expressions — nine claims name both together. The instructive difference is what the street has done to each: ADBE's consensus target is +1.4% above spot; CRM's is +39.1%. The sell side has marked Adobe's targets down to the market and has not done the same to Salesforce. If the disruption thesis is correct, that makes Adobe the less dangerous of the two, because less of the recovery is already written into the target set. If only one software name is held, the valuation case is stronger here — but the correct number of software names to hold before the September print is zero.

Single biggest risk: an unmeasurable terminal value. Not that Adobe is being disrupted — nothing in the filings shows that — but that the question cannot be answered from what the company publishes. Adobe discloses total ARR, subscription revenue across two broad customer groups, and remaining performance obligations, and nothing at all about how much revenue its generative capabilities produce, how consumption is trending, or how pricing is being restructured around them. Every bull argument in our knowledge base and every bear argument rests on an assertion about a number nobody outside the company has. That is why the risk score is 5 rather than 3, and why this is a Watch.

Most fragile assumption in the price: that gross margin holds near 89% while inference costs scale with usage. It has held so far — 89.2% in the latest quarter against 89.1% a year earlier — and the 10-Q confirms those costs are already inside cost of subscription revenue. But the same filing warns that deploying AI models "may continue to increase the cost of our solutions" and that the costs "could adversely impact our margins." Adobe's entire valuation support is a 42% free-cash-flow margin. Take gross margin from 89% to 84% and the free-cash-flow yield goes from 10.4% to roughly 8%, and the whole valuation argument in this dive becomes ordinary. It would not arrive as a headline; it would arrive as four quarters of quiet erosion.

Provenance & disclosures

• Traceability: 65 raw knowledge-base hits on the terms ADBE, Adobe and Figma; 43 used, 22 discarded (9 Figma-only claims not naming Adobe; 9 carrying thinker: quarantine_misattributed under the house attribution quarantine; 2 same-appearance duplicates split across three document identifiers from one 2025-11-16 video; 2 match: text claims whose subject is Palantir/Snowflake and Datadog/Fortinet respectively). Of the 43 used, 34 are match: entity and 9 are match: text, the latter treated as supporting colour only. Breadth 12 sources: lyn_alden, jordi_visser, anthony_pompliano (channel), bankless, we_study_billionaires, compound_and_friends, all_in, a16z, no_priors, macrovoices, gavin_baker, bill_gurley. Net stance NEGATIVE: 24 bearish / 14 bullish / 5 neutral, and roughly 3:1 bearish on skill weighting. Eighteen of the 24 bearish claims are one speaker (Jordi Visser, skill 2.0) across three channels — depth of repetition rather than breadth of agreement; where speaker and channel differ, attribution is to the speaker and the channel is named separately. That speaker's conviction on the name has fallen from 85 (2026-02-07) to 45 (2026-08-02). Two vendor stance tags are demonstrably wrong (2024-05-02 macrovoices tagged bullish on a disruption claim; 2026-03-03 compound_and_friends tagged bullish while stating it avoided Adobe); correcting them would make the split 26/12/5, i.e. more negative. All quotes are verbatim from stored claim text except where a third-party AI product or laboratory name has been withheld under house style, which is marked inline in square brackets at each occurrence.
• Data as-of: income statement 2026-05-29 (fiscal Q2 2026, 10-Q filed 2026-06-15) · balance sheet and cash flow 2025-11-28 audited (10-K filed 2026-01-15), supplemented by 10-Q narrative disclosures at 2026-05-29 · no quarterly balance sheet exists in the payload · filings through 2026-07-17 · estimates 2026-08-04 · prices 2026-08-04, quote timestamp 2026-08-04T19:58:59Z ($257.64; 50-DMA $228.57; 200-DMA $275.66; RSI 64.5) · knowledge-base claims 2026-08-04. Note the fiscal-calendar mismatch: Adobe's fiscal year ends in late November, so "FY2025" ended 2025-11-28 and the 10-K was filed 2026-01-15. Every number on this page traces to the Synthos vendor payload for ADBE or to one of the six SEC filings on file; none comes from memory, recall or external retrieval.
• Data rejections (detailed in Section 8): seg_prod structure obsolete from FY2026 (single reportable segment per the Q1 FY2026 10-Q) — filing wins; netDebt corrected from $1.217B to $53M (omits $1.164B of short-term investments) and enterprise value from $104.566B to ~$103.9B — filing wins; vendor balance-sheet data stale by two quarters against the 10-Q ($494M commercial paper, $1.35B notes reclassified current, $1.87B Semrush) — filing wins; profile.lastDividend $0.026 rejected as contradicting a zero dividend across four cash-flow statements and the 10-K's explicit statement; the 2023-11-28 est row rejected as mixed-basis (estimate below that year's GAAP actual); est.ebitdaAvg/ebitAvg rejected as mixed-basis and no EV/EBITDA multiple used; returnOnEquityTTM (62.4%), priceToBookRatioTTM (8.99x) and tangibleBookValuePerShareTTM (−$8.79) rejected as buyback artefacts; currentRatioTTM (0.751) and negative working capital reported but not treated as a liquidity signal given $7.25B of deferred revenue; tech and quote 52-week bands and moving averages disagree, tech used and both stated; peers supplies no multiples and half the set is semiconductors, so no peer comparison drawn. Checks performed and PASSED, reported as such: segment lines sum exactly to reported revenue (no double-count); longTermDebt ties to the 10-K senior-notes carrying value to the dollar and does not duplicate capitalLeaseObligations; implied share count (397.50M) matches the 10-Q cover page (397.5M) exactly.
• Basis note — the most important one in this dive: analyst estimates are NON-GAAP; reported epsdiluted is GAAP. FY2025 consensus EPS $20.822 against GAAP $16.70. Quarterly earn_cal actuals are non-GAAP and run 30-40% above the GAAP figures. All forward multiples here (10.56x FY26E, 9.37x FY27E, 8.20x FY28E) are struck on the non-GAAP consensus basis. The trailing GAAP multiple of 14.74x and the trailing non-GAAP multiple of 11.28x are both reported for calibration. Non-GAAP excludes stock-based compensation of $1.942B a year, which is 8.2% of revenue and a real cost; the free-cash-flow analysis in Section 5 deducts it explicitly to give a net shareholder yield of ~7.1% against a gross buyback yield of 9.0%.
• Cash-flow note: free cash flow is operating cash flow less capital expenditure, with no adjustment for stock-based compensation in the headline figure. FY2024 operating cash flow was depressed by the $1B Figma termination fee, confirmed in the 10-K; normalising for it, FY2025 operating-cash-flow growth is ~+10.7% rather than the reported +25%. Trailing free cash flow of $10.63B is derived from the vendor's freeCashFlowPerShareTTM against the 402M diluted count and cross-checks against freeCashFlowYieldTTM.
• Estimate coverage: deep near-term and thin beyond — 26 analysts on FY26E/FY27E revenue and 22 on EPS, falling to 9 on FY2028 EPS, 6 on FY2029 and 7 on FY2030. The FY2028 EPS range is $28.78-$36.60 (a 27% spread) and the FY2029 row shows +2.7% EPS growth on +9.2% revenue growth despite a 7% annual buyback, which is implausible and reflects the small sample. FY2027 is the primary anchor; FY2028 is used only as a base-case cross-check and is flagged; nothing rests on FY2029 or FY2030.
• Peer note: the vendor peer set (ACN, ADI, CRWD, DELL, FTNT, KLAC, PANW, SNPS, SONY, TXN) supplies price and market capitalisation only, with no multiples, and half of it is semiconductors and hardware rather than application software. No peer-multiple comparison is drawn. The batch cross-reference to Salesforce (spot $190.99, consensus target $265.75, +39.1% implied) is drawn from the batch table supplied with this assignment and is used only to contrast the street's posture on the two names.
• Fair-value caveat: the $190 / $296 / $385 anchors are multiples of consensus non-GAAP EPS (6.9x / 10.8x / 14.0x FY2027E), cross-checked independently against the free-cash-flow perpetuity implied at a 9% cost of capital (−4.3% / 0.0% / +2.0% perpetual growth respectively). Stated arithmetic, not a discounted cash flow. The base case is reachable with no re-rating at all: today's 9.37x multiple applied to FY2028 consensus EPS of $31.422 gives $294. The bull case does require re-rating and is flagged as the fragile leg. All three anchors are sensitive to the non-GAAP basis; on GAAP earnings every multiple in this dive is roughly 30% higher.
• Timing: fiscal Q2 2026 was reported 2026-06-11 (8-K) and filed 2026-06-15 (10-Q), 54 days before this dive, so the most recent quarter is fully reflected. The next print is 2026-09-10, 37 days away — inside the medium horizon and the explicit trigger for the verdict. No earnings-timing banner is warranted (the threshold is ~5 days), but the 37-day gap is an explicit verdict input rather than an incidental fact.
• Governance disclosure sourced from filings, not the vendor: chief financial officer Daniel Durn resigned effective 2026-06-15 (8-K, 2026-06-11), with Steven Day appointed interim; a chief-executive transition is disclosed indirectly by the 2026-07-17 8-K's retention letter, which sunsets "12 months following the date the new Chief Executive Officer commences employment." Neither appears anywhere in the vendor payload.
• 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.