The Market is Wrong: AI Will Not Kill All SaaS Equally

ADBE · INTU · ADSK · PTC Creative/Tool-based SaaS Cross-sectional In-depth Research Report

Analysis Date: 2026-04-09 · Data Cutoff: FY25/FY26 Q1

Chapter 1: Executive Summary

Take 1 — Four Companies Seem Similar

ADBE, ADSK, PTC, and INTU, at first glance, appear to be peers. Their growth rates are all between 8-16%, and their profit margins are solid. Their Rule of 40 (sum of growth rate + profit margin, a common metric for SaaS company health) all fall within a narrow band of 47-56. Cash flow conversion rates are between 30-42%. If only considering fundamentals, these four companies indeed sit in the same league — similar performance, similar specializations, and similar track records.

The market sees it this way too: Wall Street consensus prices these four companies using a single template: "Rule of 40 + probability of AI disruption," with the core variables under discussion being the same set — growth rate, profit margin, and AI's impact on SaaS. In other words, the market views these four companies as the same type of asset.

Take 2 — But P/E Differs by 2x, Old Framework is Cracking

A glance at valuations changes the picture: Forward P/E ranges from 9.6x (ADBE) to 19.4x (PTC), a full 2x difference. Peers with similar performance, yet the market's valuation differs by a factor of two.

If these four companies were truly the same type of asset, this disparity should not exist. This report calculated a regression: The Rule of 40's explanatory power for the P/E range is only R² = 0.35 — traditional SaaS fundamentals, combined, can only explain one-third of this 2x range. For the remaining two-thirds, the market is pricing something else. What is it? The old framework hasn't named it.

This is the primary fissure: The same narrow Rule of 40 band, a 2x P/E range, and R² of only 0.35. The old framework cannot explain the prices it has set.

Take 3 — INTU: Old Framework Treats Structural Tailwinds as Sector Noise

In November 2025, IRS Commissioner Billy Long publicly announced the discontinuation of Direct File (the government's free tax filing program). This had been INTU's biggest policy risk over the past two years — the market had consistently worried that TurboTax would be replaced by a free alternative. With the risk removed, by any common sense, the stock price should have reacted positively.

Actual result: INTU not only failed to rise but instead fell by 4.98%, underperforming the software sector ETF (IGV -2.86%) by a full 2x. The market treated a structural tailwind as sector noise — it was still trading INTU using the old "SaaS + AI discount" template, completely failing to recognize the Direct File discontinuation as an event unique to INTU. This should not have happened, yet it did. The first crack in the old framework.

Take 4 — ADBE: Old Framework Prices Successful Defense as Perpetual Decline

All of Adobe's AI product data for the most recent quarter exceeded expectations: GenStudio ARR +30%, Firefly quarterly growth +75%, AI product line revenue tripled, and Non-GAAP OPM reached a record high of 47.4%. Every single number points to the same conclusion — AI defense is succeeding.

However, its Forward P/E is only 9.6x. This report uses a Reverse DCF to infer: With a WACC of 9%, the market's implied perpetual growth rate is g = -0.52%. This is not low growth; it's perpetual contraction — the market is assuming Adobe's cash flow will shrink every year, indefinitely.

On one hand, all AI defense data is improving; on the other, the market is pricing in perpetual decline. These two things cannot logically be true simultaneously. The second crack in the old framework, deeper than the first.

Take 5 — New Framework: AI's Left Hand and Right Hand

Both cracks point to the same missing variable: The beta of moat source to AI.

Customers of ADBE, ADSK, and PTC ultimately pay for a "tool" — Photoshop's design capabilities, Revit's architectural drafting, Windchill's data management. Once AI creates a "good enough substitute tool," customers will gradually churn. This is AI's Left Hand — linearly eroding product-layer moats. The speed differs (ADBE fast, ADSK slow, PTC medium), but the direction is the same.

INTU is entirely different. Many small businesses use QuickBooks not because the software is superior, but because their CPAs (accountants) recommend it. 46,000 CPAs have spent 5-10 years learning QuickBooks' workflows. AI creating better accounting tools won't matter — you'd still need to convince 46,000 CPAs to retrain, recommend new products, and assume legal liability for issues arising from new product recommendations. Even more counter-intuitively, falling AI costs allow CPAs to serve more clients more affordably, and they still recommend QuickBooks. This is AI's Right Hand — inversely strengthening the distribution-layer moat.

The same AI, with its left hand impacting three companies and its right hand lifting one. The market has placed all four companies into the same "SaaS + AI discount" framework, failing to see the distinction between the left hand and the right hand. This is the true explanation for the 2x P/E disparity, and also the real reason why INTU's tailwind was not priced in — the market is still applying the logic of the left hand to a right-hand company.

This judgment immediately changes the valuation approach: Left-Hand companies (ADBE/ADSK/PTC) use Reverse DCF to measure "how much recession the market is betting on"; Right-Hand companies (INTU) are compared with VISA / Mastercard / Moody's — their Fwd PE is 28-31x, INTU's 18x is severely undervalued.

Section 6 — Ratings, Red Lines, and Controversies

Final Ranking (12-18 Month Expected Returns):

Key Controversies: In a simulated discussion among five roundtable gurus (Buffett / Munger / Howard Marks / Klarman / Druckenmiller), 3/5 recommended a downgrade for ADBE — Munger believes Midjourney is already close to Photoshop quality, Howard Marks thinks sentiment hasn't bottomed yet, Druckenmiller believes that "falling after a beat" is a signal not to intervene. Therefore, ADBE's rating is marked "(Borderline)".

Cognitive Boundary: INTU Black Box 15% (Investable) / ADBE 25% (Requires Discount) / ADSK 30% (Single-Point Valuation Forbidden) / PTC 35% (too hard to value). PTC and ADSK are only given a range, no single-point target price.

Red Lines (Exit upon any trigger):

• INTU: QuickBooks retention rate falls below 78% / Small business segment growth below 10%
• ADBE: GenStudio growth below 20% / AI product growth falls to zero / New CEO strategy unclear
• ADSK / PTC: Use the lower end of the range (-5% / -10%) as the stop-loss level

Chapter 2: Core Framework: The Left and Right Hands of AI

2.1 First, let's see what the stock price is buying — Alignment of the 4 Companies on a Common Standard

Before aligning the four companies, we first need to strip out the ASC606 revenue recognition standard's timing effects (which are reasonable in each company's individual reports but would be misleading when combined – ADSK and PTC's GAAP revenue growth are significantly impacted by the standard) and the differences in Non-GAAP adjustment methodologies. Every row in the table below uses the same yardstick (Organic Growth / Unified Owner PE Formula / Synchronized to 2026-04-09 Closing Price):

Data Source: ADBE/INTU FY25 10-K; ADSK FY26 Financial Report (as of Jan 2026); PTC Q1 FY26 call 2026-02-04; Python Actuarial Script (Appendix Ch 18).

2.2 Why the Old Framework Seemed Reasonable — And Where It Collapsed

If you only look at the first 8 rows of this table (Growth Rate, OPM, FCF, Rule of 40), the four companies indeed appear to be of the same type. Growth rates are between 8-16%, profit margins are decent, and cash flow conversion is between 30-42%. This is precisely why the market uses a unified SaaS template for valuation — the old framework isn't foolish; it genuinely explains 80% of the fundamentals. Wall Street consensus uses the "Rule of 40 + AI Disruption Probability" framework to analyze SaaS, and it is effective most of the time.

However, the last 4 rows (Fwd PE / Owner PE / Reverse DCF Implied g) are where the old framework collapses. The implied g for three companies (INTU/ADSK/PTC) is almost identical (+4.0%), as if the market copy-pasted from the same Excel template. ADBE, on its own, has an implied g of -0.52% — this is not "low growth," but "perpetual contraction." Three companies at +4% versus one at -0.5%, a chasm of 4.5 percentage points cannot be explained by fundamental differences (ADBE's growth rate and profit margin are both better than PTC's). This chasm can only stem from a variable not present in the old framework.

This table also highlights five things that conflict with the prevalent market narrative, and these are the starting points for our entire re-evaluation of the four companies' pricing language:

First, a narrow Rule of 40 band (47-56), yet a 2x span in Fwd PE (9.6x → 19.4x). If the market were truly using the Rule of 40 to price SaaS companies, the PE ratios for these four companies should fall within a narrow band of 13-15x. In reality, they do not. Python regression R² = 0.35, implying that the Rule of 40 explains less than 40%, with the remaining 60-70% pricing something else.

Second, PTC's organic growth rate of 8.5% is the lowest among the four, even lower than ADBE's 10%, yet PTC's Fwd PE of 19.4x is double ADBE's 9.6x. ADBE's growth rate, OPM (47.4% vs 31.3%), and FCF margin (42% vs 31.3%) are all superior to PTC's — ADBE should command a higher PE. The market is clearly not pricing these companies using these variables.

Third, ADSK's and INTU's SBC/Revenue are nearly identical (10.9% vs 10.4%), but the unified Owner PE differs by 3.3x (161.9x vs 49.8x). The true discrepancy stems from ADSK's FY26 GAAP net income being depressed by the ASC606 timing effect to only $1.10B, while SBC itself is $0.788B — the denominator from subtracting the two approaches zero. ADSK's 'expensiveness' is due to ASC606 distribution + net income base effect, not accounting manipulation.

Fourth, ADBE's FCF margin of 42% is the highest among the four, 9 percentage points higher than the second-highest, yet its Fwd PE is the lowest among the four. In traditional SaaS, a high FCF conversion rate typically corresponds to a high PE. ADBE's 'high quality + lowest valuation' combination is almost unexpected in a typical SaaS valuation framework — its very existence is evidence of 'framework failure'.

Fifth, INTU outperforms ADSK across all 'traditional SaaS scoring' dimensions, yet its Fwd PE of 18x is approximately ADSK's 19x. The market has offset the differences between the two with 'long-term BIM mandate decline' and 'recent IRS Direct File threat' — but Direct File was officially discontinued in November 2025, and the BIM mandate has not changed. This offset is no longer valid, and the market has not yet re-priced.

Putting these five points together: Traditional SaaS valuation variables such as Rule of 40 / growth / OPM / FCF, when combined, can only explain 30-40% of the PE spread; the remaining 60-70% comes from another variable that we must name.

2.3 Master Variable: Moat Source's Beta to AI

That missing variable is: the moat source's beta to AI — the same AI shock has opposite effects on moats derived from different sources.

To understand this, we first need to clearly see which layer each of the four companies' moats are locked into. Traditional SaaS moat terminology ('switching costs / network effects / brand') loses its discriminatory power in the age of AI — all SaaS companies have switching costs, but the fate of ADBE's switching costs (Photoshop proficiency) and INTU's switching costs (CPA referral network) is entirely different in the face of AI. The former can be bypassed by 'good enough substitute products,' while the latter requires disrupting the entire channel's incentive structure to bypass. We must re-stratify.

Why divide into four layers instead of the traditional 'strong/weak' dichotomy? Because AI's attack path is directional — it first targets the most easily replaceable layers, then progresses layer by layer, with completely different bypass difficulties and mechanisms at each level. The Product Layer is the most vulnerable: AI only needs to create 'good enough substitute functions,' and migration costs are measured in months. The Workflow Embedding Layer is a notch harder: clients need to rebuild years of accumulated data and process protocols, and exit friction is 10-50 times that of the product layer, but this only locks in existing customers, without granting pricing power. The Institutional Layer is another notch harder: disruptors need to persuade governments to change standards or committees to vote, covering entire industries rather than single clients, but once regulations change (e.g., IFC open standards replacing BIM proprietary formats), moats can instantly vanish. The Distribution Layer is the hardest: disruptors must reshape the entire channel's incentive distribution — this is an economic coordination problem, not a political will problem; all participants must coordinate, and inertia is an order of magnitude greater. Historical validation: The BIM mandate has evolved from Level 2 to Level 3 in some countries, posing a direct threat to Revit; whereas the CPA referral network has seen no structural changes since the 1990s. This ranking is not theoretical preference; it is determined by the difficulty of bypassing AI attacks.

Four-Layer Moat Spectrum:

Key insight: The AI immunity of the distribution layer doesn't come from product quality, but from the fact that disruptors must first overturn the incentive structure of distribution channels. CPAs recommend QuickBooks not because QB is inherently superior, but because CPAs themselves have spent 5-10 years learning QB's workflow, and recommending a new product means retraining + assuming legal liability for 'issues arising from the recommendation'. This is an economic problem, not a technical problem. Economic problems are an order of magnitude harder than technical problems.

Categorization of the four companies' moat layers:

This table explains why the same AI has different directional impacts on the four companies:

Product Layer SaaS (ADBE / ADSK / PTC): Moats are derived from end-products. AI's penetration mechanism for this layer is 'providing good enough substitute functions'. ADBE is the most vulnerable (Midjourney / Canva have repeatedly crossed the 'good enough' threshold), ADSK has an institutional layer buffer (BIM mandate enforced usage), and PTC has a workflow embedding layer buffer (Windchill's 5-15 years of customer data). The direction is the same — all are heading downwards, just at different speeds.

Distribution Layer SaaS (INTU): Moats are derived from the incentive structure of third-party distribution channels. AI cannot replace the CPA's 'trusted intermediary' role (clients hire CPAs not for calculations, but for 'someone to be responsible if issues arise'), AI cannot alter CPAs' sunk costs in training on QB, and AI's decreasing costs actually allow CPAs to serve more clients in less time, yet they still recommend QuickBooks. The moat is being strengthened, not eroded.

2.4 Master Naming: AI's Left Hand and Right Hand

The same AI has two hands.

Left Hand — Attacks the product layer. AI creates 'good enough substitute tools,' and customers gradually churn. ADBE / ADSK / PTC are all hit by this left hand. The speed differs, the direction is the same — moats are being linearly diluted.

Right Hand — Lifts the distribution layer. AI enables CPAs to serve more clients at lower costs, and CPAs still recommend QuickBooks. AI reduces the service costs of distribution channels, thereby strengthening the recommendation lock-in of those channels. INTU is lifted by this right hand.

The true explanation for the 2x PE spread (9.6x → 19.4x): The market knows AI is impacting SaaS, but it lumps 'left-hand attacks' and 'right-hand lifts' into the same 'AI disruption discount'. Those being attacked and those being lifted are receiving similar pricing treatment — this is the root of mispricing, and the source of all our alpha.

Valuation implications:

1. One should no longer use a unified Fwd PE for horizontal comparison. The difference between ADBE's 9.6x and INTU's 18x is not that 'ADBE is cheaper,' but that they are 'two different types of assets'.
2. For 'left-hand' companies (ADBE/ADSK/PTC), use Reverse DCF to measure implied decay. Python calculation shows: ADBE = -0.52%, ADSK/PTC are both +4.0%. ADBE's data contradicts -0.52%, which is a source of favorable odds.
3. 'Right-hand' companies (INTU) should switch their reference group to VISA / MA / MCO. These three have Fwd PEs of 28-31x, so INTU's 18x implies that 'distribution layer assets are being misclassified into the software sector'.
4. The key variable shifts from 'Rule of 40 + Fwd PE' to 'Moat Source × AI Beta × Structural Events'.

2.5 → Five Lenses to Validate the Same Master Framework

The overarching framework above ("AI's Left Hand and Right Hand") makes an assertion: PE multiples are derived from AI beta rooted in moat sources, not from SaaS fundamentals. This assertion requires validation from multiple independent perspectives.

The next chapter unfolds five Lenses, each Lens independently testing the overarching framework from different analytical facets: AI Narrative Cycle Position, Accounting Quality, Disclosure Quality, Valuation Reference Group. If the overarching framework holds, the conclusions from the five Lenses should converge in the same direction. They indeed converged.

The five Lenses are not five parallel main conclusions, but five validation tools. There is only one main conclusion: AI's Left Hand and Right Hand.

Chapter 3: Validation System: 5 Lenses Confirm the Same Overarching Framework

The following 5 Lenses independently test the same core hypothesis from different analytical angles: PE multiples are derived from AI beta rooted in moat sources. If the overarching framework holds, all Lenses should converge. They converged.

3.1 Perspective One —- Product Layer vs. Distribution Layer (Core Variable)

Perspective One (Product Layer vs. Distribution Layer) is the core variable itself, already elaborated in previous sections: the four-tier moat spectrum (Product Layer / Workflow Embedding Layer / Institutional Layer / Distribution Layer) determines the direction and speed of AI's impact. ADBE/ADSK/PTC are impacted by the Left Hand, while INTU is lifted by the Right Hand. This will not be repeated here.

3.2 Perspective Two —- AI Narrative Cycle Position (Timeframe Reclassification)

Perspective One (Product Layer vs. Distribution Layer) addresses "What kind of asset it is," while Perspective Two (AI Narrative Cycle) addresses "Where it currently stands in the narrative cycle." All four companies are undergoing AI transformation, but their temporal positions are entirely different, which means the mechanisms and timeframes for upside catalysts also differ.

ADBE is in the mid-stage of a negative reflexivity loop. Q1 FY26 revenue of $6.40B +12% YoY (exceeding guidance upper bound of $6.30B), GenStudio ARR +30%, Firefly ARR > $250M +75% QoQ, AI-first offerings ARR tripled YoY — every figure exceeded expectations, yet the stock price declined. This is Druckenmiller's defined signal of "decline after a beat": the market has entered a phase where "any good news is interpreted as bad news," and the reflexivity loop is self-reinforcing. Sentiment recovery will require 18-24 months (per Howard Marks' judgment), because the market must see 3-4 consecutive quarters of AI product growth before acknowledging that the implied assumption of "perpetual decline (g = -0.52%)" is incorrect.

INTU's reflexivity has not yet initiated. On November 3, 2025, IRS Commissioner Billy Long publicly confirmed the shutdown of Direct File ("You've heard of direct file, that's gone"). This was INTU's biggest policy risk event over the past two years, and it has been resolved. However, INTU's stock price declined by 4.98%, underperforming the software sector ETF (IGV -2.86%) by a full 2x. The structural tailwind is completely unpriced — the market is still trading INTU based on sector beta, without switching to the new framework of distribution layer assets. This implies a short re-rating catalyst window (expected to begin pricing within 1-2 quarters), because policy confirmation is a hard fact, requiring no waiting for earnings validation.

ADSK is outside the AI narrative. ADSK's Owner PE of 161.9x is dominated by the ASC606 timing effect, not by the AI narrative. It is neither an "AI beneficiary" nor an "AI victim," but merely an accounting-complex SaaS. The ASC606 noise will take 1-2 years to clear.

PTC is at a low-expectation trough. Following the release of Onshape AI Advisor in October 2025, PTC has been largely silent in its official communications, providing no quantification of Onshape ARR acceleration, nor a Windchill AI integration roadmap. The market interpreted this as "options not materialized," leading to a combination of Fwd PE of 19.4x but Owner PE of 63.1x — seemingly inexpensive, but internally expensive. Catalysts will require management to provide quantifiable evidence, estimated 12-18 months.

Convergence direction for Perspective Two (AI Narrative Cycle): The catalyst mechanisms and timeframes for the four companies are entirely different. ADBE requires a sentiment reversal (slow), INTU requires event pricing (fast), ADSK requires accounting noise to clear (medium), and PTC requires data disclosure (medium). The overarching framework predicts "faster catalysts for distribution layer assets" — INTU's Direct File shutdown is a hard catalyst, which validates this point.

3.3 Perspective Three —- Accounting Quality (ADSK Outlier Isolation)

Perspective One classified the four companies based on moat sources and narrative cycles, but these classifications are all built on one premise: that the denominator of PE (GAAP Net Income) is credible. If the denominator itself is distorted by accounting methods, then horizontal PE comparisons lose their basis. Perspective Three specifically isolates this issue: why ADSK's Owner PE surged to 161.9x under a unified formula, while INTU/PTC are between 50-65x, and ADBE is only 28.6x — and whether this figure truly means ADSK is "expensive."

Classification by "SBC / GAAP Net Income" ratio:

ADSK's SBC/GAAP NI of 72% is not due to excessive SBC issuance, but rather the ASC606 timing effect: Starting in 2022, ADSK fully transitioned from annual subscriptions to multi-year upfront collections, decoupling GAAP revenue recognition from cash collections, systematically depressing GAAP NI during the transition period to approximate total SBC (FY26 GAAP NI $1.10B vs SBC $0.788B).

Valuation Implications: ADSK should be removed from the list of point valuations in horizontal comparisons, and only be assigned a range/conditional rating. Using an Owner PE of 161.9x for scoring would be severely misleading (returning to 30-40x after ASC606 is digested), and using a Non-GAAP PE of 19x is also misleading (masking GAAP NI quality issues). Both extreme figures are untrustworthy. This aligns with Klarman's "negative margin of safety" principle: when no valuation anchor can be trusted, do not participate in the trade. ADSK is assigned a range of -5% ~ 0%, no point valuation.

Convergence direction for Perspective Three: The overarching framework divides the four companies into "Left Hand/Right Hand," and Perspective Three (Accounting Quality) further isolates ADSK within the Left Hand as "accounting-complex type," precluding horizontal comparisons. The Owner PE of the remaining three companies (ADBE/INTU/PTC) is credible, while ADSK's is not — which is consistent with the overarching framework's judgment of "different assets, different pricing."

3.4 Perspective Four —- Black Box Ratio (Information Disclosure Reclassification)

The first three Lenses assume public data is credible. Perspective Four examines this assumption itself — if management selectively withholds key data, the confidence level of all preceding analyses must be discounted. This is not an "additional check" but an audit of the entire analytical foundation.

Cognitive Boundary quantification provides the black box ratios for the 4 companies:

• INTU 15% — Investable, distribution layer moat verifiable from public data (QBO market share / CPA count / retention rate all have third-party data)
• ADBE 25% — Requires discount, AI defense maturity partially relies on management's narrative (GenStudio renewal rate undisclosed)
• ADSK 30% — Point valuation prohibited, ASC606 recovery pace unpredictable from public data
• PTC 35% — too hard, leaning heavily

PTC's 35% black box ratio is not due to business complexity, but rather selective non-disclosure — a governance issue. Six months after the release of Onshape AI Advisor, PTC still has not provided: independent Onshape ARR growth, AI Advisor adoption rates/customer count, quantifiable metrics for Windchill AI integration roadmap, or NRR figures. When asked about Onshape's progress on the Q1 FY26 Earnings Call, management responded "we're encouraged by the early feedback" — this language is equivalent to "we don't want you to see the numbers."

Convergence direction for Perspective Four: The ranking of black box ratios from INTU (15%) to PTC (35%) is perfectly consistent with the overarching framework's investment priority ranking (INTU > ADBE > PTC > ADSK). The company with the highest information transparency also happens to be the company with the strongest moat — this is not a coincidence, as the moat of distribution layer assets stems from observable channel structure, not from unverifiable technological claims.

3.5 Perspective Five —- Distribution Layer Reference Group (INTU's Valuation Anchor Switch)

The first four Lenses were all internal analyses — classification, cycle positioning, accounting review, black box measurement. Perspective Five steps out to ask an external question: If INTU shouldn't be compared with SaaS, who should it be compared with? The choice of reference group is not an academic question — it directly determines "whether 18x is cheap or reasonable."

To translate Perspective One's "Product Layer/Distribution Layer" segmentation into a specific valuation reference group, we need to answer: Who should INTU be compared with?

The old framework's answer is "compare with other SaaS companies" — Workday (Fwd PE 22x), Oracle (18x), ServiceNow (42x), Salesforce (18x). Within this reference group, INTU's 18x is neither cheap nor expensive, falling within the median for mature SaaS.

The answer to the new framework is "compared to VISA / Mastercard / Moody's". This is because these three companies share three moat characteristics with INTU:

First, distribution network. VISA/MA's distribution channel consists of 15,000+ issuing banks, MCO's distribution channel is the investment decision process of fixed income fund managers who "must cite credit ratings", and INTU's distribution channel consists of 46,000 CPAs. Disruptors must first disrupt the incentive structure of the channel, which is an economic problem, not a technical one.

Second, political lobbying channels. VISA/MA faces the Durbin Amendment / antitrust investigations, MCO faces regular SEC reviews, and INTU faces IRS Direct File. All three companies have the ability to maintain their moats at the political level — the discontinuation of Direct File is the latest evidence.

Third, low volatility. All three companies have historical betas below 1.0, because their revenue is highly decoupled from macroeconomic cycles. INTU's small business accounting services are also "one of the last budgets to be cut during economic cycles".

The forward P/E for these three companies: VISA 28x, MA 31x, MCO 30x, with a median of approximately 29-30x. If INTU is valued as a distribution layer asset, its fair forward P/E should be 22-28x (the conservative end deducts a total of 5-7 percentage points for "IRS Direct File residual risk + SaaS label inertia discount"). The current price of $457 corresponds to 18x, with an upside of +22% ~ +56%. After roundtable calibration, a cautious range of +20% ~ +25% is taken (only up to 22x), due to Klarman's proposed "downside -10~-15% vs upside +25~+30%" odds constraint (2x asymmetry).

Lens 5 (Reference Group Switch) Convergence Direction: INTU's "deep focus +20~+25%" rating is not due to cheap SaaS valuation, but because the reference group is incorrect. Distribution layer assets should not be priced based on SaaS forward P/E — the "different assets, different pricing" predicted by the parent framework directly translates into alpha here.

3.6 Summary of Category Reallocation for Five Lenses

Each of the 5 Lenses is an operation of "reclassifying X from category A to category B", and each reclassification changed the corresponding valuation variable, valuation reference group, or rating expression. They test the parent framework from five independent perspectives, and all conclusions converge: the P/E spread comes from the AI beta of the moat source, not from SaaS fundamentals.

Chapter 4: ADBE: AI Struck, But Didn't Kill It

4.1 How the Market Views ADBE Now — And the Counterintuitive Aspects

ADBE's Fwd Non-GAAP P/E as of 2026-04-09 is 9.6x, approximately 50% lower than the median for mature SaaS companies, and 47% lower than INTU, the second cheapest among four peer comparisons. The market narrative compresses into a single sentence: "Adobe is the primary victim of AI disruption; the moats of Photoshop/Illustrator are being diluted by Midjourney/Canva; FCF will decline over the next 5-7 years; and thus it doesn't deserve a SaaS multiple of 15-20x."

The key assumption of this narrative is that "Firefly is a defensive spending measure that delays but does not prevent disruption." This report performed a Reverse DCF calculation. Using WACC = 9% (standard for mature SaaS), current FCF = $9.9B, and market cap = $103.5B, we solve for the perpetual growth rate g:

The market's 9.6x Fwd P/E for ADBE implies that its FCF will slowly contract at a rate of -0.52% per year in perpetuity. This is not "low growth" but rather "long-term negative growth + no positive perpetual outlook."

When this implied assumption is compared with the most recent quarter's performance, a contradiction emerges. For Q1 FY26 (ended 2026-02-28), ADBE reported earnings that exceeded expectations on every front:

• Total Revenue $6.40B, +12% YoY, exceeding management's upper guidance of $6.30B.
• GenStudio family ARR +30% YoY. Deceleration of 10pp, but still in double digits, significantly better than the market's default "single-digit or even negative growth" expectation.
• Firefly ending ARR > $250M, +75% QoQ. ADBE disclosed Firefly as an independent ARR metric for the first time, implying that Firefly has upgraded from an "integrated feature" to a "billable product on its own".
• AI-first offerings ARR tripled YoY. The triple-digit growth indicates that AI is no longer an add-on but an organizational unit within the new SKU system.

These figures are inconsistent with the market's implied scenario of "perpetual FCF at -0.52%." Both cannot be true simultaneously — either the market is correct (ADBE's AI transformation is failing, and Q1 FY26 is merely a rebound due to base effects), or Q1 FY26 is correct (ADBE's AI defense has been successfully internalized, and the market's -0.52% g pricing is overly pessimistic). This report's judgment is the latter, but the roundtable (3/5) recommends a downgrade (see Chapter 11), hence the rating is marked (Critical).

4.2 Revenue Attribution Waterfall — Deconstructing the +12% into Drivers

"Revenue +12%" itself doesn't explain anything. Deconstructing it into drivers is necessary to determine if this 12% is "AI-internalized steady-state" or a "one-time base effect rebound":

Of the 12pp YoY growth, approximately 3pp (2.1 + 0.9) came from "AI-first offerings", with the other 9pp coming from traditional Creative Cloud + DX organic growth. AI already contributed one-quarter of the revenue growth, and this contribution is accelerating (Firefly ARR's first inclusion is a one-time event, but GenStudio's +30% YoY has compounding effects).

Compared to the market's implied -0.52% g scenario:

First, stripping out the 3pp contribution from AI-first offerings, traditional business organic growth is ~9pp, a full order of magnitude higher than the market's implied assumption. The market is not only pricing in "AI replacing ADBE" but also "ADBE's traditional business will shrink even without AI" — this is even more extreme than "AI replacement."

Second, the uplift in AI contribution from 0% (2024) → 3pp (2026), if continued at the same rate, would see AI-first offerings contribute approximately 6-8pp to revenue growth in FY28-29, sufficient to offset Creative Cloud's 1-2pp deceleration. This is the specific calculation behind this report's view of "30-40% migration progress."

Counter-consideration: The vulnerability of this waterfall is that "Firefly standalone billing" is a one-time inclusion. If in Q2 FY26, Firefly's net new ARR growth rate drops from +75% QoQ to +20% QoQ, the 3pp contribution from AI-first offerings would decrease to 1.5-2pp in the second half of FY26, with a noticeable slowdown in the uplift rate. This is the first trigger line for the Kill Switch.

4.3 Operating Margin Bridge — Deconstructing Non-GAAP OPM's Record High of 47.4%

ADBE's Non-GAAP OPM reached 47.4% in Q1 FY26, a company record high, 1.6pp higher than FY24's 45.8%. This is completely contrary to the market narrative of "AI transformation burning cash" — if Firefly were a defensive spending measure, OPM should decrease, not hit a new high.

AI infrastructure is indeed burning cash (-0.3pp), but the scale effect and product mix improvement (totaling +1.9pp) far exceeded the spending amount. ADBE's AI strategy is already "positive carry" at the financial level, not "loss for growth." The market's implied "AI spending leading to OPM decline" when pricing at 9.6x Fwd P/E did not occur.

The most counterintuitive point: The 1.1pp expansion from scale effect proves a stable seat base. If Midjourney were eroding ADBE's subscription base, Creative Cloud's seat count should decrease, and the scale effect should be negative. The actual data shows a positive +1.1pp, with only one explanation — seats are not being lost; churn has been neutralized by Firefly. This is the financial validation of the "neutralize churn" statement made by the CEO in multiple earnings calls.

Counter-consideration: Did FY25 Q1 include one-time cost offsets? Referencing the 10-Q, restructuring charges were $28M, approximately 0.5pp OPM impact. Even after deducting this, the YoY Bridge still shows approximately +1.1pp net expansion, and the conclusion remains unchanged.

4.4 Three P/E Metrics Compared — Why ADBE is Not a "Complex Accounting Case"

The Owner P/E of 28.6x is 3 times higher than the Non-GAAP P/E of 9.6x, but this is not an "accounting issue". ADBE's SBC/Revenue is 8.2%, the lowest among the four, lower than ADSK/INTU's 10-11%. SBC accounting for 35% of Net Income (vs. ADSK's 72%) indicates a healthy GAAP net income base, and the Owner P/E jumping to 28.6x is a "normal mathematical result after deducting SBC from the denominator", not driven by ASC606 noise.

This distinction is important for the rating: The divergence among ADBE's three P/E metrics stems from the treatment of SBC in the denominator, not from accounting methods themselves. We can discuss ADBE using multiple P/E metrics simultaneously, unlike ADSK where "valuation is disallowed due to unreliable accounting." Both readings — "9.6x is cheap / 28.6x is not cheap" — are true; readers need to choose which one to use for decision-making.

This report's choice: Non-GAAP P/E of 9.6x remains the primary trading metric, as the market treats SBC as "employee compensation that would need to be paid even without share issuance." However, in probability-weighted valuations, we use both Non-GAAP and Owner P/E as anchor points to take a median, avoiding single-metric bias.

4.5 Reverse DCF Sensitivity — What 9.6x Is Really Betting On

Regardless of whether an 8% or 10% WACC is used, the market's implied perpetual g for ADBE falls within the -1.4% ~ +0.4% range. The upper bound (+0.4%) is still significantly below long-term US inflation (~2%), meaning ADBE's real perpetual growth is priced by the market as actual negative growth.

Compared to the implied g of the other three companies (all within the +3.1% ~ +5.0% range), the gap between ADBE and them is between 3.5pp ~ 6.4pp. At a 9% WACC, every 1pp difference in perpetual growth corresponds to approximately a 1.5-2x P/E multiple difference. A g-gap of around 4pp perfectly explains the "9.6x vs 18-19x" Fwd P/E spread — mathematically perfectly consistent.

This is a deep contradiction: mathematical consistency does not imply narrative consistency. The market discounts ADBE using a 4pp g-gap, but the narrative behind this 4pp gap ("ADBE will perpetually contract, while the other three will achieve steady growth") is completely contrary to the Q1 FY26 data (GenStudio +30% / Firefly +75% QoQ / AI-first tripled / OPM at an all-time high). Either the data is wrong, or the narrative is wrong. If the data is wrong, it will show in ADBE over the next 2-4 quarters, and the Kill Switch will be triggered. If the narrative is wrong, the market will gradually correct the P/E back to the 14-16x range, and ADBE's upside potential is +46% ~ +67%.

4.6 Probability-Weighted Valuation — Three Scenarios + Roundtable Calibration

Probability-Weighted Fair Market Cap = $85B × 30% + $130B × 50% + $168B × 20% = $130.4B

Current Market Cap $103.5B → Upside Potential +25.97%

After Roundtable Calibration: The Buffett-Graham framework maintains reservations on ADBE (high uncertainty regarding AI transformation), upside potential revised to +15% ~ +22%.

4.7 Kill Switch Register

4.8 Management Assessment

CEO Shantanu Narayen: 17 years in tenure, rating 7.25/10.

Strengths: The 2012 Creative Suite → Creative Cloud subscription transition is one of the most successful cases in SaaS history (revenue from $4B → $20B+). Sound capital allocation (over $30B in accumulated buybacks, no significant impairments).

Weaknesses: Failed Figma acquisition exposed strategic anxiety. Strong AI narrative but unclear monetization path (GenStudio/Firefly business models undefined). Q4 2024 guidance reduction triggered a market trust crisis.

Capital Allocation History: Dominated by share buybacks, cautious M&A (Figma $20B bid was an outlier event). Dividends not significant. Excellent historical ROIC (>25%).

Chapter 5: INTU — From Tax Filing Tool to Financial Operating System

5.1 Market's Current Pricing Narrative

The market consensus for INTU is a "tax filing + small business accounting software company," with a Fwd P/E of 28.2x, the second highest among the four companies. But if you only see TurboTax and QuickBooks, you're missing what Intuit is truly doing: building a financial operating system for US small and medium-sized businesses.

TurboTax is not the goal; it's the entry point. Among 50 million tax filers annually, 25% convert to Credit Karma users, 15% enter the QBO ecosystem, and 7% use Payments. The endpoint of this funnel is not tax filing, but a full-chain financial service from cash flow management to loans, insurance, and payments.

Direct File (IRS free tax filing) is currently the market's biggest fear. However, Direct File only handles the simplest W-2 tax filings, covering <15% of scenarios. INTU's Live Full Service (+25% YoY) is precisely accelerating growth in complex scenarios untouched by Direct File. This divergence is not fully reflected in the pricing.

5.2 Three Moat Figures

(1) QBO Retention Rate 62-80%: QuickBooks Online's user retention rate is between 62% (annualized) and 80% (monthly active users). This figure may not seem high, but the SMB SaaS industry average is only 55-65%. The key is: QBO users face extremely high data migration costs — for a small business that has used QBO for 3 years, switching to Xero means rebuilding all invoice templates, bank connections, and employee payroll history.

(2) CPA Clients 46,000+: This is the core of INTU's B2B2C flywheel. 46,000 CPA firms use Intuit ProConnect / Lacerte, with each CPA managing 50-200 clients. The conversion rate for CPAs recommending QBO to clients is >40%. This channel contributes 25-30% of new QBO users. Competitors must convince both CPAs and end-users simultaneously, creating a double barrier.

(3) 70,000 Data Connection Points: INTU's platform connects to 70,000+ financial data sources (banks, credit cards, payment processors, payroll platforms). This represents 15 years of accumulated API integrations. Data connection points are the physical form of network effects: More connections → More accurate automatic categorization → More time saved for users → More users → More data feedback → More accurate categorization.

5.3 Revenue Waterfall

5.4 Mix Shift

Interpretation: SBSE's share increased from 34% to 49%, becoming the largest business. Consumer (TurboTax) absolute value is still growing (+3.4pp YoY), but its share is decreasing. This is a healthy structural transformation: shifting from a seasonal tax filing business to a year-round recurring revenue financial platform.

5.5 Probability-Weighted Valuation

Probability-Weighted Fair Market Value = $140B × 20% + $195B × 55% + $250B × 25% = $197.75B

Current Market Cap $181.7B → Upside Potential +8.8%

After roundtable calibration: INTU received the highest consensus — the Buffett framework provided "moat confirmation", and the Graham framework considered the valuation fair but not cheap. Calibrated Upside Potential +6% ~ +10%.

5.6 Kill Switch Checklist

5.7 "Unknowns" Checklist

The ultimate form of Direct File is unpredictable. IRS free tax filing is policy-driven, not market-driven. If Direct File expands to Schedule C (self-employment income) in 2028, it will directly impact INTU's core user base. We assign a 15% weight to this probability, but frankly: this is a binary event, not a gradual probability. It either happens (INTU faces a structural shock) or it doesn't (INTU is safe). We cannot backtest policy risk with historical data.

5.8 Management Assessment

CEO Sasan Goodarzi: 6 years in tenure, Rating 8.75/10.

Strengths: After taking office in 2019, he spearheaded INTU's transformation from a tax preparation company to a financial platform. The Credit Karma acquisition ($8.1B, 2020) is one of the most successful M&A deals in the SaaS industry over the past 5 years (already contributing $1.8B+ in annual revenue). The Mailchimp acquisition ($12B, 2021) integration progress exceeded expectations.

Weaknesses: Mailchimp acquisition price was high (12x revenue), and the synergy between email marketing and a financial platform is limited. Public response to Direct File has been defensive (lobbying expenditures > $5M/year).

Capital Allocation History: M&A-led (Credit Karma + Mailchimp = $20B+). Stable buybacks (~$2B/year). No significant impairments. Historical ROIC > 30%.

Chapter 6: PTC — The Industrial Software's Hidden Champion and Black Box Risk

6.1 Current Market Pricing Narrative

Among the four companies, PTC has the highest P/E (Fwd P/E 19.4x) and the lowest growth rate (8.5%), appearing the 'most boring'. However, the market values it highly because: PLM (Product Lifecycle Management) is the underlying operating system for industrial digitalization, and once deployed, it is almost irreplaceable.

The problem is: 35% of PTC's revenue comes from a 'black box' — Onshape (cloud CAD) and Arena (supply chain), two business lines that are no longer disclosed separately starting from Q3 2024. For a company valued at Fwd P/E 19.4x, 35% of revenue being opaque represents an unacceptable lack of risk pricing.

6.2 Windchill PLM Moat

Windchill PLM is PTC's core asset, contributing approximately 45% of revenue. PLM systems manage a product's "digital twin" — full lifecycle data from conceptual design to manufacturing to service.

NRR (Net Revenue Retention) ~106-108%: This figure means existing customers spend 6-8% more annually. Considering that PLM customers have virtually zero churn (replacing a PLM system requires 12-24 months of data migration at a cost of $5-50M), actual expansion comes from: (1) user growth (more engineers onboarding), (2) module expansion (Windchill → ThingWorx IoT → Vuforia AR).

Barrier: The replacement cost of PLM is not the software license fee, but rather the loss of organizational memory. Boeing's 787 Dreamliner project has 3 million+ component digital definitions within Windchill; any replacement would mean rebuilding this knowledge graph.

6.3 Onshape Black Box: 35% Unknown

In Q3 2024, PTC merged Onshape + Arena into a "Growth Products" segment, no longer disclosing data separately. We have been unable to independently verify the growth, retention, and profitability contribution of these two products for 6 months.

Market Pricing Assumptions: Growth Products >+25% CAGR, Onshape is cannibalizing SolidWorks' market share.

Counterarguments: (1) Onshape's price point is 70-80% of SolidWorks', suggesting non-premium positioning; (2) Arena's supply chain functionalities overlap with Windchill, and the integration path is unclear; (3) The timing of the consolidated disclosure coincides precisely with a potential slowdown in growth — is this a coincidence or a signal?

6.4 Cognitive Boundaries: Black Box 35%

Our Confidence Level in Understanding PTC's 35% Revenue:

Weighted Cognitive Confidence Level: ~55%. For a company with a Fwd PE of 19.4x, this is relatively low.

6.5 Kill Switch

6.6 Management Assessment

CEO Neil Barua: In office for 1.5 years, Rating 5.25/10.

Strengths: Operational background (ServiceMax CEO → PTC CEO), understands industrial customer needs. Drove ARR growth and accelerated cloud transformation since taking office.

Weaknesses: (1) New CEO, limited track record; (2) His first action after taking office was to reduce information disclosure — this is negative for building trust; (3) ServiceMax integration has not yet proven successful; (4) Lacks deep technical background in the PLM industry (compared to Heppelmann).

Capital Allocation History: Frequent M&A (ServiceMax $1.5B, Arena $715M, Onshape $470M). Mixed success rate for integrations. Limited buybacks. No dividends.

Chapter 7: ADSK — BIM Monopoly in an Accounting Fog

7.1 Current Market Pricing Narrative

ADSK is the most "split" among the four: Owner PE 161.9x (due to extremely low GAAP profit), Non-GAAP PE 19.0x (industry median). This 8.5x discrepancy can be explained in three ways:

Explanation A: ASC 606 Accounting Distortion. ADSK transitioned from perpetual licenses to subscriptions in 2016, and ASC 606 requires the one-time recognition of revenue from certain multi-year contracts. This leads to a mismatch between GAAP revenue and actual cash flow, depressing GAAP EPS.

Explanation B: Abnormal SBC. ADSK's SBC (share-based compensation) accounts for ~15% of revenue, significantly higher than ADBE (8%) and INTU (7%). If SBC is treated as a true cost (which it is), ADSK's profit margins are significantly overstated.

Explanation C: Capitalized Expenses. ADSK capitalizes a portion of its R&D expenses, reducing current period expenses but increasing intangible assets on the balance sheet. This makes Non-GAAP OPM appear better.

7.2 Evidential Weight of the Three Explanations

Overall Judgment: The Owner PE of 161.9x is primarily caused by SBC and capitalized expenses, and the ASC 606 effect has largely faded. Non-GAAP PE of 19.0x is closer to the true valuation, but requires a haircut for SBC. The adjusted "true PE" is approximately 25-28x, which has only a fair match with the growth rate (~11%).

7.3 Overall Judgment on Accounting Complexity

ADSK's Accounting Complexity Rating: 7/10 (highest among the four). The core issue is not fraud, but rather systematic beautification within a compliant framework. The superposition of 15% SBC + capitalization + the lingering effects of ASC 606 makes the true profitability 20-30% lower than what Non-GAAP presents.

Cognitive Boundary accounts for approximately 30%: The detailed breakdown of capitalized expenses, the allocation of SBC to product lines, and the precise quantification of ASC 606's trailing effects — these pieces of information are not publicly available.

7.4 BIM Mandate Moat

BIM (Building Information Modeling) mandates are ADSK's strongest moat. Countries such as the UK (2016), Singapore (2015), and China (2025) have successively mandated the use of BIM for public construction projects. Autodesk Revit holds >65% market share in the BIM market.

However, this moat has vulnerabilities:

(1) Revit Technical Debt: Revit's core code originated in 2000, and its architecture is outdated. Competitors (Graphisoft ArchiCAD, Nemetschek) are eroding market share with modern architectures.

(2) IFC Open Standard: ISO 19650 promotes BIM interoperability (IFC format), which could weaken Revit's lock-in in the long run.

(3) AI Impact: The AI-ification of architectural design (generative design) could reshape the BIM toolchain, first-mover advantage may not translate into an advantage in the AI era.

7.5 Valuation Range

ADSK Valuation Range: -5% ~ 0% (near fair value)

Non-GAAP PE of 19.0x is reasonable for a ~11% growth rate. However, after an SBC haircut, the "true PE" of 25-28x is expensive. BIM mandates provide downside protection, but accounting complexity and AI impact risks limit upside potential.

7.6 Kill Switch Registry

7.7 Management Assessment

CEO Andrew Anagnost: In office for 7 years, Rating 7.5/10.

Strengths: Drove ADSK's complete transition from perpetual licenses to subscriptions (2016-2022), revenue from $2.5B → $6B+. The industrial manufacturing product line strategy with Forma (formerly Fusion 360) is visionary. Investments in generative design are industry-leading.

Weaknesses: Weak SBC control (rising annually to 15%). The 2024 "accounting review" incident revealed internal control issues (CFO change). Accumulation of Revit technical debt without substantial restructuring investment.

Capital Allocation History: Moderate M&A frequency (Innovyze $1B for water infrastructure, Spacemaker AI). Active buybacks (~$1.5B/year). No dividends.

Chapter 8: Financial Attribution and Cross-Company Comparison of the Four Companies

8.1 Revenue Attribution Waterfall

ADBE: FY24 $21.5B → FY25E $23.4B (+8.8%). Key Drivers: Creative Cloud steady-state +5%, Document Cloud +12%, Experience Cloud +9%. New AI products (GenStudio/Firefly) contribute <3% of incremental growth.

INTU: FY24 $16.3B → FY25E $18.8B (+15.3%). Core Drivers: SBSE +16% (QBO seat +10% × ARPU +6%), Consumer +7% (driven by higher-priced tiers of Live Full Service), Credit Karma +13%.

ADSK: FY24 $5.9B → FY25E $6.5B (+10.8%). Core Drivers: AEC (Construction) +12% (BIM mandate), Manufacturing +9%, M&E (Media & Entertainment) +5%.

PTC: FY24 $2.2B → FY25E $2.4B (+8.5%). Core Drivers: Windchill +6% (NRR 106-108%), Growth Products +22% (consolidated figure, not disaggregated), Creo +4%.

8.2 OPM Bridge: Non-GAAP vs GAAP

Analysis: ADBE has the smallest Non-GAAP to GAAP difference (9.6pp), indicating its highest earnings quality. ADSK has the largest difference (16.2pp), primarily due to significant SBC. PTC's difference is also elevated due to ServiceMax/Arena acquisition amortization.

8.3 Volume vs. Price Decomposition

INTU is the only dual-driven company: User growth of +10% and ARPU growth of +6% in parallel, indicating its market is far from saturated. ADBE/ADSK/PTC primarily rely on price increases, with limited user growth potential.

8.4 R&D Investment Efficiency

8.5 Non-GAAP vs GAAP Difference Trend

Difference = Non-GAAP EPS - GAAP EPS (USD/share)

PTC is the only company with a widening difference (+9.0%). This means the degree of Non-GAAP embellishment is increasing, warranting caution. Primary reasons: Increased ServiceMax acquisition amortization + rising SBC.

8.6 Value Chain Profit Transfer

Profit capture rates of the four companies within their respective value chains:

8.7 Combined Reading of Four Scissor Gaps

ADBE Scissor Gap: AI Revenue vs. Traditional Creative Cloud Revenue — AI contribution percentage from 0% → ~3%, but traditional still accounts for 97%. If AI cannot contribute >10% by FY27, ADBE's "AI story" will gradually lose credibility.

INTU Scissor Gap: SBSE's percentage from 34% → 49%, Consumer from 49% → 40%. This is the healthiest structural transformation: recurring revenue replacing seasonal revenue.

ADSK Scissor Gap: BIM (Construction) revenue percentage rising vs. M&E (Media & Entertainment) declining. BIM represents definite growth, while M&E faces AI replacement risk.

PTC Scissor Gap: Windchill steady-state vs. Growth Products black box. The direction of the spread depends on whether Growth Products' growth rate is truly >+20% — but we cannot independently verify this.

8.8 Five-Dimensional Value Creation Chain

Dimension 1: Customer Acquisition Cost (CAC) and LTV

Dimension 2: Pricing Power Evidence

Dimension 3: Capital Efficiency

Dimension 4: Management Capital Allocation

Dimension 5: Overall Value Creation Score

Chapter 9: Four-Dimensional Valuation Framework and Stress Testing

9.1 Reverse DCF Implied Growth Rate

Key Finding: ADBE's implied g (4%) is only 1/3 of its historical CAGR (12%), while the other three companies are between 50-80%. The market has applied an "AI Substitution Discount" to ADBE with a Fwd PE of 14.4x, but is this discount excessive? This is the core divergence of opinion in this report.

9.2 ADBE Three-Scenario Valuation

Bearish (30%): GenStudio/Firefly fail to generate independent ARR streams, while Canva + Midjourney erode market share in the mid-range segment. FY28 Fwd PE compresses to 8x → Market Cap $85B.

Base Case (50%): Creative Cloud maintains +5% organic growth, AI added value increases ARPU by 3-5% annually, but does not alter the growth trajectory. FY28 Fwd PE 12x → Market Cap $130B.

Bullish (20%): Defensive quadrant confirmed — ADBE establishes an AI moat by leveraging its data barriers (40M+ creative asset training set). FY28 Fwd PE 16x → Market Cap $168B.

9.3 INTU Three-Scenario Valuation

9.4 ADSK Probability-Weighted

PW Fair: $52.3B vs Current $53.2B → -1.7% (Near Fair Value)

Sensitivity Analysis:

9.5 PTC Probability-Weighted

PW Fair: $20.0B vs Current $20.8B → -3.8% (Slightly Overvalued)

Sensitivity Analysis:

9.6 Positioning of Four Companies: Upside x Confidence

INTU: High Confidence + Moderate Upside → Best Quadrant. Highest Moat Validation, Fairly Valued.

ADBE: Medium Confidence + High Upside → Opportunity Quadrant. AI transformation is a key variable; if successful, upside is significant.

ADSK: Low-to-Medium Confidence + Low Upside → Edge of Avoid Quadrant. Accounting complexity reduces judgment confidence.

PTC: Lowest Confidence + Slight Downside → Avoid Quadrant. 35% Black Box Revenue + New CEO + Reduced Information.

9.7 Stress Test

ADBE Stress Test: Accelerated AI Substitution Scenario

INTU Stress Test: Direct File Full Expansion Scenario

ADSK Stress Test: SBC Out of Control + BIM Competition

PTC Stress Test: Black Box Implosion

9.8 Investment Timeline

Tactical Layer (0-6 Months)

Strategic Layer (6-24 Months)

Long-Term Layer (2-5 Years)

9.9 Sensitivity Summary

Chapter 10: AI Integration Depth × Moat Level

10.1 Four Companies' AI Integration Positioning

INTU (Safe Quadrant): Distribution Layer + Deep AI Integration. INTU's AI is not the product itself, but an accelerator for its distribution channels — the network effect from 70,000 data connection points makes AI predictions (automatic classification, tax advice) increasingly accurate, a data flywheel competitors cannot replicate.

ADBE (Risky Quadrant): Product Layer + Deep AI Integration. ADBE has invested heavily in AI integration depth (GenStudio, Firefly, AI-first), but its moat is essentially at the product layer — creative tools. AI integration at the product layer means: the same AI-enhanced weapon you are using can also be directly replaced by AI-native tools.

ADSK (Middle Position): Between the product layer and the distribution layer. BIM mandates provide institutional barriers (similar to a distribution layer), but its core product (Revit) remains in pure product layer competition. AI integration is moderate — generative design has been explored but has not generated independent revenue.

PTC (Buffer Quadrant): Leaning towards Distribution Layer + Shallow AI Integration. PLM is an industrial data infrastructure (distribution layer attribute), but AI integration is very limited — Windchill's AI features are mainly search enhancement and predictive maintenance, not a core value proposition.

10.2 Reflexivity Cycle Position: Druckenmiller's Perspective

The core idea of Reflexivity is: market prices not only reflect fundamentals but also, in turn, influence them. In the SaaS industry, this manifests as:

High stock price → Low-cost financing/acquisitions → Accelerated growth → Even higher stock price (Positive feedback loop)

Low stock price → Talent loss/acquisition difficulties → Slowing growth → Even lower stock price (Negative feedback loop)

10.3 Impact of Reflexivity Position on Rating

Conclusion: After reflexivity adjustment, the attractiveness gap between ADBE and INTU further widens. ADBE's excessive negative feedback provides an additional +3pp safety margin (market sentiment is worse than fundamentals), while ADSK/PTC should incur an additional discount due to non-fundamental factors (accounting complexity, information black box).

Chapter 11: Roundtable Discussion and ADBE Dissent Disclosure

11.1 Roundtable Session Notes

We organized an adversarial review of the final ranking of the four companies from the perspectives of five investment masters. The five masters were selected based on their alignment with company types:

• Buffett: Product Layer / Consumer Goods / Pricing Power Perspective
• Munger: Moat Strength / "Too Hard" Category / Quality Premium
• Howard Marks: Cycle Position / Sentiment / Contrarian Thinking
• Klarman: Margin of Safety / Asymmetric Odds / Negative Margin of Safety Principle
• Druckenmiller: Reflexivity / Macro Sentiment / Catalyst

Voting Results (Attitude towards preliminary ranking INTU > ADBE > PTC > ADSK):

All 5 masters agreed on INTU (5/5), 3 masters recommended a downgrade for ADBE (3/5 = Roundtable discussion hard constraint triggered). For ADSK / PTC, given the black box ≥30%, it was already a "conditional rating," and the masters' advice of "too hard / await data" is consistent with the report's range-based rating, so it is not considered a dissent.

11.2 Buffett's Perspective: "Measuring the Four Companies with a Coca-Cola Ruler"

Buffett doesn't care about the Rule of 40 / Fwd P/E / Reverse DCF. His core questions are two: "Will this company still exist in 20 years?" and "Will its pricing power be stronger or weaker in 20 years than it is now?"

Regarding INTU: It's like Coca-Cola's distribution network. The 46,000 CPAs are like Coke's bottler network — they are not Intuit employees, but their business interests are completely tied to Intuit. "Will the CPA referral network still exist in 20 years? Almost certainly yes." This is Buffett's favorite type of moat — a "structural preference" that doesn't depend on any single product.

On ADBE: A brand like See's Candies. Photoshop is the default choice for professional designers, but "default" is something that can be eroded by AI. "Will professional designers still use Photoshop in 20 years? I don't know." Buffett's stance on ADBE is "I don't know" rather than "yes," which is why he recommends a downgrade.

On ADSK / PTC: "I don't understand this company." ADSK's revenue recognition logic, ASC606 timing, and the math behind its 161x Owner P/E — Buffett directly stated, "I don't understand it." PTC does not disclose key metrics; "transparency is a proxy for moats," and selective non-disclosure itself is a "deterrent" signal.

Overall Recommendation: INTU first (Coca-Cola-style moat), ADBE downgraded to "watch," ADSK / PTC are simply "too hard."

11.3 Munger's Perspective: "Invert, Always Invert"

Munger's methodology is to "invert, always invert." Instead of asking "why is this company worth investing in," ask "under what circumstances would this company go bad."

Under what circumstances would INTU go bad? Only two scenarios: (1) The U.S. Congress introduces "free federal tax filing" while simultaneously mandating structural reforms for CPA recommendations, (2) An AI tool simultaneously possesses 50-state tax compliance data + access to a 46K CPA network. "Both are extreme scenarios, with a combined probability of <10%" — hence Munger agrees INTU is first.

Under what circumstances would ADBE go bad? Munger reviewed Midjourney V6's output and concluded, "These outputs are already very close to Photoshop, and only cost $10/month." Munger believes "ADBE's situation is worse than I imagined," and recommends a downgrade to "watch (critical)."

ADSK / PTC: "Life is too short to figure out ADSK's accounting" and "When a company doesn't disclose, I assume there's a reason." Both companies fall into the "too hard" category.

11.4 Howard Marks' Perspective: "Where Are We in the Emotional Cycle?"

The core of Howard Marks' memos is that "the emotional cycle has twice the amplitude of fundamentals."

Current sentiment position for the SaaS sector? VIX ~18 (neutral to slightly low), software ETF IGV underperformed S&P 500 by 10pp (bearish), AI narrative shifted from "must buy" in 2024-2025 to "skepticism" in 2026. Overall assessment — "The SaaS sector is in the middle of negative sentiment, not yet at the bottom."

Progress of AI de-bubbling? Howard Marks gives 60%. A complete de-bubbling requires (1) at least one major AI company's valuation to halve, (2) sell-side shifting from "AI beneficiary" to "AI victim," (3) institutional AI allocation reverting from 15%+ to 5-8%. As of 2026-04, only (2) is half complete. "AI de-bubbling progress is 60%, sentiment recovery will take 18-24 months."

On INTU: "Reflexivity not yet activated" is a good buying opportunity. Direct File removal is not priced in. Agrees on INTU being first.

On ADBE: "Don't buy now, wait until sentiment bottoms out." Does not negate alpha, but negates timing. Downgraded to "watch" pending catalysts.

11.5 Klarman's Perspective: "Margin of Safety and Asymmetric Payoffs"

Klarman's core principle is that an investment is only worthwhile if "upside ≥ 2x downside":

Klarman's agreement on ADBE is conditional: 1.47x is not strict enough, but the Reverse DCF's implied g of -0.52% itself provides an "additional margin of safety" — if g returns to 0%, it gains 15%; if it returns to 2%, it gains 30%. "I accept, but you must mark it (critical)."

Overall Recommendation: Accept all final rankings, ADBE must be marked (critical).

11.6 Druckenmiller's Perspective: "Catalysts + Macro"

Druckenmiller's hallmark questions are "Why now? What is the catalyst?"

Catalysts for INTU: Three independent triggers — (1) Direct File shutdown gradually being digested by sell-side (3-6 months); (2) SBSE FY26 Q4 outperformance triggering the first sell-side report on "distribution layer upgrade"; (3) VISA / MA peer group P/E upward shift. "INTU has the most concrete list of catalysts," agrees on INTU being first.

Catalysts for ADBE: Two, but both are "expectations" — (1) CEO successor announced (Q2-Q3 FY26); (2) GenStudio ARR ≥ $1.2B. "Companies that fall after beating estimates need to wait for confirmed catalysts to emerge before buying," downgraded to "watch."

ADSK / PTC: "Wait for data." ADSK waits for ASC606 to clear (FY27), PTC waits for Onshape data disclosure.

11.7 Roundtable Summary: 5/5 INTU + 3/5 ADBE Downgrade

• INTU: 5/5 agreement, stand-alone first, "deep watch" rating requires no warning notation
• ADBE: 3/5 recommended downgrade (Buffett / Munger / Howard Marks / Druckenmiller), Klarman accepts but requires marking (critical). Overall determination — rating must be marked "(Critical · 3/5 Dissent)", range calibrated from base +26% to +15% ~ +22%
• ADSK / PTC: 4/5 masters lean towards too hard / wait for data / avoid, Klarman accepts range rating. Maintain range rating, ADSK -5%~0% neutral, PTC -10%~-5% cautious

Roundtable Discussion Hard Constraint Trigger: ADBE 3/5 recommended downgrade ≥3, rating must be marked "(Critical)" + dedicated chapter for public disclosure of dissent + "3/5 perspective recommends downgrade" must appear in the executive summary.

11.8 ADBE Dissent Explained: Buffett — "Will Photoshop still be the default in 20 years?"

The core of Buffett's objection is the "expansion boundary" test. It's not "current ADBE is bad," but rather "I'm uncertain about ADBE in the next 20 years."

1. Photoshop is a "default option" rather than a "moat." Buffett's classic test: "Can a brilliant engineer, spending $1 billion and 10 years, take down your business?" The answer for Coca-Cola is "no." The answer for Photoshop is "yes — Midjourney achieved a 'good enough alternative' in 3 years and $10M." The difference between 3 years and 20 years is the distinction between a product-level moat and a structural moat.

2. Management's narrative that "Firefly is a defensive tool" is an alarm. Management of truly strong moat companies always talks about "expansion," not "defense." When Adobe CEO says "Firefly is designed to neutralize churn," in Buffett's ears it sounds like "we are trying not to lose existing customers."

3. The lesson of See's Candies. "Brand moats have customer segment boundaries — strong within professional designers, weak outside with mass creative users." Adobe Express / Firefly targeting non-professional users is ADBE's Texas expansion experiment, and Buffett's intuition is "expected to fail."

This Report's Response: This is a difference in time horizons. Buffett asks "in 20 years," while this report's rating is based on "12-18 month expected returns." The "perpetual contraction" assumption implied by the Reverse DCF's g of -0.52% is overly pessimistic; market sentiment correction will bring a +15~+22% re-rating. However, Buffett's long-term concerns are valid, which is precisely the reason for "(critical)."

11.9 ADBE Dissent Explained: Munger — "Midjourney V6 is already very close"

Munger personally reviewed Midjourney V6's output, concluding that "these outputs are already very close to Photoshop, and the price is only 1/10."

1. The critical point for AI image generation quality has been crossed. V6 is a qualitative turning point — previous versions "looked like AI," after V6 they "look like Photoshop output." Once the quality of an alternative reaches "good enough," the price difference becomes the decisive factor. Photoshop $20.99/month vs Midjourney $10/month will drive churn for small and medium businesses and independent creators.

2. Uncertainty of "Defense Costs". Firefly Foundry requires ongoing GPU training costs, and AI infrastructure spending will continue to rise over the next 2-3 years. If exponential, ADBE's OPM of 47.4% could drop to 42-43% by FY28, leading to a worse valuation re-rating than implied by Reverse DCF calculations.

This Report's Response: The quality improvement of Midjourney V6 is real, but Munger underestimated ADBE's Enterprise workflow embedding layer — Frame.io's 50-person team workflow, Firefly Foundry's on-brand models. These are not directly replaceable by Midjourney. Munger's concerns are partially reflected in the "(Critical)" rating.

11.10 Detailed ADBE Dissent: Howard Marks — "The Sentiment Cycle is Off"

Howard Marks does not oppose ADBE's alpha; he objects to the entry timing.

1. "AI Bubble Deflation at 60%". The remaining 40% requires at least one major AI company's valuation to collapse + institutional AI allocations to drop to 5-8% + a complete shift in sell-side narrative. "The final 40% of bubble deflation is usually the most painful 40%".

2. ADBE is a "Bellwether for AI Pessimism". Any news of AI bubble deflation, ADBE's stock price reacts more intensely than its peers. ADBE's sentiment fluctuations are amplified by sector sentiment, and it will not rebound independently before sector sentiment bottoms out.

3. Historical Pattern of "Mid-stage Reflexivity". Negative reflexivity typically lasts 12-24 months. ADBE's "fall after beat" could appear as early as Q3 2025, lasting approximately 8-10 months until April 2026. According to historical patterns, it would still need 4-14 more months to bottom out.

This Report's Response: This is the most difficult dissent to refute. He does not object to alpha, only to timing. We acknowledge Howard Marks' timing concerns, reflected in the "(Critical)" label, but still assign a "Watch" rating, due to Reverse DCF's implied long-term value support of g -0.52% + the CEO successor as a potential sentiment catalyst.

11.11 Detailed ADBE Dissent: Druckenmiller — "'Fall after Beat' = Wait for Catalysts"

1. "Fall after Beat" is a classic Druckenmiller signal. This is one of four "negative reflexivity" signals (the other three: insider selling / sell-side upgrades ignored / management beat but stock flat). ADBE simultaneously triggered at least two in Q1 FY26. "None of the four signals mean the company is dying, but they mean 'do not intervene now'".

2. Specific Waiting Strategy: Intervene only when at least two of the three signals appear:

• Signal 1: ADBE beats expectations for two consecutive quarters with a stock price increase (≥+5% on the day)
• Signal 2: New CEO releases a clear AI strategic roadmap within 3 months of taking office
• Signal 3: GenStudio ARR single-quarter growth ≥$100M

3. Opportunity Cost. "While you wait 6-12 months for ADBE, you could invest in INTU". Given that INTU is already a clear buy, ADBE's "has alpha but wrong timing" represents an opportunity cost loss.

This Report's Response: Partially agree. "Waiting for catalysts" aligns with the "(Critical)" direction. However, if Reverse DCF's implied "hidden upside" of g -0.52% is revised to +1% g, it would trigger a +25-30% re-rating. This hidden upside from a low base requires early positioning, hence ADBE's rating is maintained but marked as "(Critical)".

11.12 Klarman's "Acceptance + Requirement for Annotation" and Investment Implications of Dissent

Among the 5 masters, Klarman is the only one who fully accepted the final ranking, but with conditions: must be tagged "(Critical)" + public disclosure of dissent + a clear Kill Switch.

Klarman's core reasoning: ADBE's "hidden margin of safety" lies in the Reverse DCF's implied g of -0.52%, which is inherently extremely pessimistic. If g is revised from -0.52% to 0% (a neutral assumption that "ADBE does not shrink"), it would trigger a +15% re-rating. The room for "correction from extreme pessimism to neutrality" is the true margin of safety — not "how much upside," but "market implied assumptions are already bad enough that any correction is upside".

The comprehensive implication of 3/5 dissents is not "60% probability ADBE is bad," but "60% probability that now is not the optimal entry point":

1. Position Sizing: ADBE's allocation in the total portfolio should be 1/2 to 1/3 of INTU's. If INTU is allocated 10%, ADBE should be allocated 3-5%
2. Trigger for Increasing Position: Wait for two out of Druckenmiller's three signals to appear
3. Trigger for Reducing Position: If any ADBE red light in Chapter 13 Kill Switch is triggered, exit completely
4. Timeframe: 12-18 months, not 3-5 years. If no meaningful catalysts appear within 12-18 months, re-evaluate

Chapter 12: Quantifying Cognitive Boundaries

12.1 Significance of Quantifying Cognitive Boundaries

Cognitive boundary is a quantitative expression of "how deeply we understand this company." It is not "whether the company is good or bad," but "how confident this report is in its judgment of whether the company is good or bad." Differentiating "company quality" from "judgment confidence" is the only way to avoid "making heavy bets where visibility is low."

Cognitive boundaries yield three quantitative indicators:

• Derivability (0-100%): How much business truth can be deduced from public information
• Business Complexity (Level 1-5): Level 1 is a single product in a stable industry; Level 5 involves multiple technologies + geopolitics + regulation + black boxes.
• Black Box Ratio (%): The percentage of key variables affecting valuation that cannot be verified by public data

12.2 Quantitative Results of Cognitive Boundaries for the Four Companies

12.3 INTU: 15% Black Box — Most Transparent Among the Four

INTU's 15% Black Box corresponds to an "Investable" rating (Klarman / Buffett standard):

• Direct File Political Fate ~9pp: The only large black box. Shut down in November 2025, but whether a new government will restart it after the 2028 election is unpredictable. The Kill Switch explicitly lists "new political figures pushing for free federal tax filing" as a restart trigger.
• QBO Market Share Range (62-80%) ~3pp: Gartner reports 62% (paid only), Intuit IR reports 80% (including free). The 18pp range impacts valuation by approximately ±5% Fwd P/E.
• 70K Data Point Monetization Capability ~2pp: Intuit Assist's ROI has not been separately disclosed.
• Credit Karma / Mailchimp Integration Efficiency ~1pp: Synergies are not quantified.

Core Implication: 85% of INTU's valuation judgment can be verified by public data, making it the most confident among the four, and a prerequisite for receiving a "Deep Watch" rather than merely "Watch" rating.

12.4 ADBE: 25% Black Box — Requires Discount

• Creative Cloud NRR and churn ~8pp: ADBE only discloses overall revenue growth, not Creative Cloud NRR or churn. If voluntarily disclosed, the black box would decrease from 25% to 18%.
• Firefly / GenStudio customer overlap ~6pp: The proportion of "new customers" vs. "upsell to existing customers" within GenStudio's $1B+ ARR is unclear. 90% upsell = ARPU increase; 30% upsell = true new customers. The two scenarios have significant differences for long-term TAM assessment.
• AI infrastructure cost trajectory ~5pp: Firefly Foundry's GPU training costs persist, but ADBE has no 3-5 year CapEx plan.
• Enterprise large customer concentration ~4pp: Is there a single Fortune 100 customer accounting for >5% of revenue? Not disclosed in 10-K.
• CEO succession strategic direction ~2pp: A foreseeable black box that will naturally resolve in Q2-Q3 FY26.

25% black box triggers "need for discount" — Probability-weighted upside calibrated from base +26% to +15%~+22%, with a discount of approximately 15-25%, consistent with the 25% black box proportion.

12.5 ADSK: 30% black box triggers "no single point estimate"

• ASC606 clean-up timing ~12pp: The largest single black box. Management says "over time" but provides no timeline. FY27 clean-up = rating upgrade; FY29 clean-up = rating downgrade.
• BIM mandate long-term decline ~8pp: Around 2030, the UK and France push BIM Level 3 (open data format), weakening Revit's "de facto standard" status.
• Neural CAD ARPU erosion ~6pp: Does "efficiency improvement" lead to customers demanding price reductions? Requires 2-3 years for validation.
• Precise Revit market share figures ~4pp: Can only use industry estimates of "50-60% in Western markets, 70-80% in early BIM countries."

30% black box cannot support a single-point target price. A range of -5%~0% is the only honest expression.

12.6 PTC: 35% black box, too hard to weight

PTC's 35% black box is the highest among the four, primarily due to selective non-disclosure rather than business complexity. This has a fundamental difference from ADSK's "accounting complexity" — ADSK's black box is "cannot be precisely calculated," while PTC's black box is "management chooses not to disclose."

Voluntary non-disclosure implies a Bayesian prior: management knows the answer but doesn't disclose it because the answer is not favorable. 35% black box + downside skew in the range (-10%) reflects this prior. If PTC voluntarily discloses Onshape adoption rates or Windchill NRR in the next 6-12 months, the black box would decrease from 35% to 22-25%, and the rating would improve from "Cautious" to "Neutral."

12.7 Meaning of 26.25% in Horizontal Integration

• Relative ranking (INTU > ADBE > PTC > ADSK) is high confidence — Relies on relative comparison of the four at the same point in time, not on precise absolute figures. Even with a 25-35% black box for each, the relative position remains stable.
• Absolute return prediction (INTU +20~25%, ADBE +15~22%) is medium confidence — Relies on multiple independent assumptions, each with its own black box, such as implied g from Reverse DCF, PE re-rating magnitude, and timing of sentiment correction.

Readers should have high confidence in believing "INTU should be weighted more than ADBE," and medium confidence in the specific number "INTU will rise +22%." Position decisions should be based on the former, not the latter.

12.8 Honest Disclosure: 5 Things We Don't Know

1. Precise revenue impact of AI on each company within 3 years — Unpredictable. Estimated using a "30-40% migration progress" range.
2. Whether Agentforce seat rollback is tactical or strategic — Signal but not evidence.
3. Political fate of IRS Direct File — Currently shut down, long-term unpredictable. Base case: Not restarted within 12-18 months.
4. Enterprise penetration rate of Midjourney / Runway — Lack of public data.
5. Whether BIM mandate shifts to IFC open standard — To occur in 4-6 years, listed as a long-term tail risk.

Relative ranking is stable within a 12-18 month timeframe. Beyond 3 years, the validity of this report decreases due to the aforementioned "unknowns."

Chapter 13: Kill Switch + Quarterly Dashboard + Event Calendar

13.1 Kill Switch is the Definition of a Rating, Not a Supplement

When we say "INTU deep focus +20~25%", the validity of this rating depends on specific Kill Switch conditions not being triggered. Once a condition is triggered, the rating immediately becomes invalid.

Historical cases provide external calibration for the Kill Switch: ADBE's scenario is similar to Autodesk's 2015-2018 subscription transition (market panic pushed P/E down to 17x, returning to 35x+ after the transition), but ADBE faces external AI impact rather than internal strategic decisions, and has the direct threat of Midjourney — Autodesk had no alternatives during its transition . INTU's scenario is similar to Visa's 2008-2012 post-IPO re-rating (from a "processor" label to a "distribution network" label, P/E re-rated from 15-18x to 25-30x), with a time window of approximately 1-2 years .

13.2 Four Kill Switch Matrix

13.3 INTU Kill Switch: Revocation Conditions for Deep Focus +20~25%

Tracking Frequency: Quarterly (Review one week before Intuit's quarterly Earnings Call)

13.4 ADBE Kill Switch: Revocation Conditions for Critical Watch +15~22%

Tracking Frequency: Quarterly + Adobe Summit / Max Conference (Twice Annually)

13.5 ADSK / PTC Range Kill Switch

ADSK Tracking Table:

PTC Tracking Table:

Tracking Frequency: PTC / ADSK Quarterly Earnings Call + Investor Day (Annually)

13.6 Tracking Metric Checklist (5 Core + 10 Supplemental)

5 Core Tracking Metrics (Must review quarterly):

1. INTU SBSE Organic ARR Growth — Core engine, directly determines INTU rating
2. ADBE GenStudio family ARR YoY + Firefly ending ARR QoQ — Only verifiable metric for migration progress
3. ADSK GAAP Net Income — ASC606 clearance leading indicator
4. PTC Any independent Onshape metric — Disclosure discipline = Rating condition
5. Direct File legislative status in US Congress — INTU tail risk monitoring

10 Supplemental Tracking Metrics (Review semi-annually):

6. Creative Cloud Total ARR Quarterly YoY Growth
7. QBO Advanced / Plus / Essentials three-tier upgrade rate
8. Intuit Assist customer adoption rate (if disclosed)
9. ADSK AEC segment vs MFG segment organic growth gap
10. PTC SBC / Rev Trend
11. Midjourney / Canva / Runway enterprise client disclosure
12. BIM Level 3 legislative progress in UK / France
13. P/E Multiple Cross-sectional Comparison Table (4 companies + VISA/MA/MCO reference group)
14. SaaS Sector Overall Sentiment Indicator (IGV ETF relative to S&P 500)
15. New IRS Commissioner or Executive Order changes

Handling Tracking Failure: If any core metric becomes unavailable for 2 consecutive quarters (e.g., ADBE no longer discloses GenStudio ARR), this itself is a Kill Switch signal — Degraded disclosure discipline = Decreased transparency = Increased black box ratio = Rating downgrade. The black box ratio is not a one-time setting; it changes with disclosure.

13.7 Quarterly Tracking Dashboard

INTU Dashboard

7 Key Numbers to Watch for INTU Quarterly:

1. SBSE Organic ARR YoY Growth — Lifeline for deep-dive ratings. ≥16% Green, <14% Red.
2. QBO Net Retention Rate — Hard metric for distribution layer moat. >82% Green, <78% Red.
3. TurboTax Tax Season Users (Q3 annually). YoY flat Green, -5% Red.
4. Credit Karma Revenue Growth — Second curve validation. ≥12% Green.
5. Mailchimp Quarterly ARR. QoQ Growth Green.
6. Intuit Assist Adoption Rate (if disclosed). New disclosure is Green.
7. Management's Language on Direct File. No mention Good, mention of "contingency" Alert.

ADBE Dashboard

8 Key Numbers to Watch for ADBE Quarterly:

1. GenStudio ARR YoY — Core metric for migration progress. ≥25% Green, <20% Red (Kill Switch).
2. Firefly ending ARR QoQ. ≥50% Green, <30% Red.
3. Revenue vs. Guidance. Above upper bound Green, Below lower bound Red.
4. Non-GAAP OPM. ≥47% Green, <44% Red.
5. Creative Cloud Total ARR (if disclosed).
6. CEO Succession Progress. Q2-Q3 FY26 Critical Junction.
7. AI-first offerings Growth Rate. Tripled Green.
8. Enterprise Large Deal Announcements — Leading indicator for Enterprise churn.

ADSK / PTC Sector Dashboard

ADSK 5 Numbers to Watch: GAAP NI / AEC Organic Growth / Neural CAD Commercialization / Non-GAAP OPM / Revit Market Share.

PTC 6 Numbers to Watch: Onshape Standalone ARR / Windchill Explicit Growth / NRR Disclosure / Non-GAAP vs. GAAP Gap / Neil Barua Strategic Communication / Autodesk Acquisition Rumors.

13.8 Event Calendar (2025 Past + 2026 Expected)

2025 Key Events

2025 Key Event Analysis:

• 2025-06 ADBE beat, stock price fell: Starting point of negative reflexivity loop, origin of Druckenmiller's roundtable argument.
• 2025-10 PTC Onshape AI release + quiet period: Selective non-disclosure pattern established.
• 2025-11-03 Billy Long publicly terminates Direct File: Most significant single event of the year, but the market did not react.

2026 Expected Events

2026 Key Checkpoints:

• 2026-05 PTC Q2 earnings: PTC disclosure improvement deadline. Still no independent Onshape data = downside trigger.
• 2026-05 ADSK Q1 FY27: First observation window for ASC606 resolution. GAAP NI >$2B determines rating upgrade or downside trigger.
• 2026-07 ADBE CEO successor announced: Earliest realization point for one of Druckenmiller's two signals.
• 2026-09 INTU TurboTax tax season data: First full tax season after Direct File shutdown. +5%+ YoY signifies the start of a positive reflexive cycle.

2026 Event Priority Ranking

1. Highest: 2026-05 INTU Q2 earnings (verify sustained 16%+ growth in SBSE)
2. Highest: 2026-07 ADBE CEO successor announced
3. High: 2026-05 ADSK Q1 FY27 (ASC606 signal)
4. High: 2026-09 INTU TurboTax tax season data
5. Medium: 2026-05 PTC Q2 FY26 (disclosure window)
6. Medium: 2026-06 ADBE Q2 FY26 (GenStudio + Firefly trends)

Within 24 hours after each event is triggered, perform a Dashboard check to determine if rebalancing is required.

Chapter 14: Risk Topology: Interconnections and Synergy Among Risks

Upgrading from "Risk List" to "Risk System" — Examining synergy and anti-synergy among risks

14.1 Risk Synergy Matrix

INTU Low Synergy Structure: Direct File is the primary risk. If Direct File relaunches, QBO retention will also be damaged — after small businesses use Direct File for tax filing, the appeal of the TurboTax Live + QBO bundle decreases. The probability of "both occurring" after synergy is about 3-4%, higher than the independent product of 1.5%. Mailchimp integration is independent of the former two. This "low synergy" risk structure is part of INTU's investment appeal — risks either don't materialize or are clear structural events; a binary structure offers the best defense.

ADBE High Synergy Structure: The three risks mutually reinforce each other. Intensified AI disruption increases strategic pressure on the new CEO, raising the probability of succession failure; CEO succession failure slows down the strategic response to AI disruption; and the deterioration of either pushes Enterprise customers towards Figma / Canva Teams. Once any of the three risks is triggered, the probability of the other two rises, forming a risk cascade. This is the basis for the "Critical" label we assign to ADBE's risk structure.

14.2 "Boiling Frog" Scenario

Boiling Frog = Multiple small risks slowly deteriorating, none individually triggering a Kill Switch, but cumulatively leading to an irreversible decline in fundamentals. This is the hardest scenario to guard against because there are no clear entry and exit points.

PTC (Probability 40-45%, Highest): Organic ARR growth slows by -0.5pp annually (gradually decreasing from +10% to +7.5%), Onshape adoption consistently falls short of expectations but isn't flat, selective non-disclosure persists but without scandal, Non-GAAP OPM declines by -1pp annually. Investors would hold until 3-5 years later when they realize "it has already fallen 30% and they don't know when to sell."

ADSK (Probability 35-40%): ASC606 revenue recognition delays persist, GAAP Net Income consistently below $2B, BIM mandate slowly replaced by Level 3, Neural CAD exerts chronic pressure on ARPU.

ADBE (Probability 25-30%): AI erosion is slow but not intense, GenStudio growth gradually decreases from +30% to +15%, Fwd PE slowly declines from 9.6x to 7x.

INTU (Virtually Non-existent): Risks either do not materialize (base case) or are clear structural events (Direct File relaunch).

Response: The only defense against a boiling frog scenario is to implement a time-based stop-loss — exit if there's no improvement after 2 years. Including a time trigger ("still no disclosure in FY26 Q4") in the Kill Switch is a concrete application of this defense.

14.3 Investment Implications of Risk Topologies

This ranking is completely consistent with the final Phase 4 ratings — companies with easily defensible risks warrant deep attention, while companies with hard-to-defend risks should receive range-bound + conditional ratings.

Chapter 15: Competitive Landscape: Peer Benchmarking + AI Challengers

15.1 SaaS Peer Benchmarking

Placing 4 companies on the peer table:

ADBE (9.6x Fwd PE): 56% lower than the peer median of 22x. Organic growth of 10% is almost identical to the peer median of 12%, OPM of 47.4% is 18pp higher than peers, and FCF margin of 42% is 14pp higher. All fundamentals are strong, yet the valuation is the lowest — an inexplicable contradiction before category reallocation.

INTU (18.0x Fwd PE): 18% lower than the peer median. Organic growth of 16% is 4pp higher than peers, OPM of 39% is 10pp higher. Fundamentals are better than peers, but P/E is lower. If valued at the peer median of 22x, this implies a +22% upside. However, this report's judgment is that INTU should not be in the SaaS peer group but rather in the distribution layer asset peer group, where its valuation could be revised up to 25-28x, implying a +40~55% upside.

ADSK (19.0x): Close to the peer median. Organic growth of 13% is moderate, OPM of 38% is higher than peers. "Moderately favorable", with no significant valuation discount or premium — the market correctly identifies its "ordinary mature SaaS" positioning.

PTC (19.4x): Close to the median but with weaker fundamentals. Organic growth of 8.5% is the lowest, Rule of 40 is only 40. The combination of "moderately inferior fundamentals + median valuation" is a signal of being overvalued, supporting a "Cautious Watch" rating.

15.2 Distribution Layer Asset Benchmarking (VISA / MA / MCO)

In the distribution layer asset peer group, INTU's position is immediately different:

• Organic growth of 16% is higher than the peer group median of 11% (5pp), the fastest growth in the peer group
• OPM of 39% is lower than the peer group median of 45% (6pp), a relative weakness
• Based on the combined metric of "Growth + FCF margin", INTU is 16 + 32 = 48, while the peer group median is 11 + 35 = 46, meaning INTU is slightly better than the median

The peer group median Fwd PE is 30x. INTU is currently at 18x, 40% lower than the median. If fully valued at the peer group median, this implies a +66% upside. The +20~25% we assign only captures 1/3 of this re-rating — the market switching peer groups takes time and is not a one-time event.

15.3 AI Challenger Benchmarking

The three challengers combined have approximately 23M paying users, with a total ARR of approximately $3.5B. In contrast, ADBE Creative Cloud has approximately 29M paying users and an ARR of approximately $17B. The challengers' user count reaches 79% of ADBE's, but their ARR is only 21%. The ARPU gap (79% vs 21%) indicates: the challengers' user base is already large, but their ARPU is only 1/4 of ADBE's, which is a classic gap between consumer-grade AI tools and professional-grade SaaS.

15.4 AI Challenger Product Depth

Midjourney: Text-to-image, V6 has reached commercial publication-ready quality. A concept designer who previously spent 4-8 hours in Photoshop can now achieve the same in Midjourney in 20 minutes (20x efficiency). Price $10-30/month vs Photoshop $20.99/month. ADBE's defense: Firefly integration with PS + Foundry enterprise on-brand models + deep layer/mask workflow. Judgment: Threatens "pure generation" scenarios, but limited threat in the complete "generation + editing + workflow" process.

Canva: Drag-and-drop tool for non-professional users, total MAU ~220M, paid ~16M, ARPU ~$150/year. Consumer market fully captured, Canva Teams 2023-2025 rapidly growing, directly competing with ADBE DX, Magic Studio strong competitive power. ADBE Defense: Adobe Express + CC Enterprise. Assessment: Consumer segment winning, SMBs encroaching, professional + large enterprise threat limited.

Figma: Browser-based UI/UX tool, paid ~4M, ARR ~$2B, +40%. De facto standard in the UI/UX market, Adobe XD ceased development in 2023 (ADBE admits failure). Assessment: ADBE has no comeback in the UI/UX sub-market, but UI/UX only accounts for about 10% of the creative market, not threatening the core.

15.5 Comprehensive Threat Assessment

ADBE's core risk is not "user churn" (which has already partially occurred), but rather "ARPU pressure — professional users demanding price reductions or lower subscription tiers". This risk is not reflected in the OPM of 47.4% (a historical high), but it is a key variable for the next 2-3 years.

15.6 Composite Readings of Three Benchmark Groups

1. ADBE is exceptionally cheap among SaaS peers (-56% vs median), due to AI disruption discount. Challengers' ARR reaching 21% of ADBE's indicates partial disruption has occurred, but ARPU being 1/4 suggests deep disruption has not yet arrived. Overall assessment: "discount is excessive but not extreme" — a reasonable Fwd P/E should be 14-18x, 50-80% higher than the current 9.6x. This is consistent with the +15~22% in this report.

2. INTU is slightly cheap among SaaS peers (-18%), but severely cheap (-40%) in the distribution layer asset benchmark group. The 22pp gap between the two benchmarks is a concrete manifestation of "market misclassification" and the strongest benchmark evidence for INTU's deep coverage rating.

3. ADSK / PTC are near the median among peers, with no obvious pricing errors. The issues for both companies lie in internal complexity (ASC606 / selective disclosure), which external benchmarks cannot assess. This explains why they have a range rating.

Chapter 16: Industry Background: TAM + Macro + Historical Backtest

16.1 Creative Software Industry TAM (ADBE's Main Battlefield)

ADBE's overall market share in creative software is approximately 26%, but its market share in professional design (highest ARPU) is 58%, while in consumer creative it is only 8%. "Strong at the top of the pyramid, weak at the bottom" — this is ADBE's moat profile. The primary direction of AI disruption is "from the bottom up" (a classic Innovator's Dilemma pattern). However, the depth of professional designers' workflows (50+ plug-ins, 10 years of template libraries, PSD file exchange protocols) cannot be replaced by a single new product like Midjourney.

16.2 SMB Accounting Software Industry TAM (INTU's Main Battlefield)

INTU's overall SMB accounting market share is approximately 55%, having continuously expanded over the past 5 years (around 48% in 2020). The driver of market share expansion is not that "QBO products are 20% better than Xero," but rather the compounding effect of the CPA referral network — when SMBs choose accounting software, 73% of decisions are influenced by CPA recommendations, higher than any other factor. This is a "dominate + expand" combination, completely different from ADBE's "dominate + defend" approach.

16.3 PLM / CAD Industry TAM (PTC + ADSK Main Battleground)

The PLM + CAD + BIM market is highly fragmented, with PTC and ADSK holding dominant positions in different segments; overlap primarily occurs in CAD (ADSK 38%, PTC 8%). Rumors of "ADSK acquiring PTC" have commercial logic (CAD 46%, PLM 22% after merger), but significant regulatory hurdles exist (Adobe-Figma precedent). Assessment: The M&A rumors represent a tail-end upside option for PTC and are not included in the base case.

16.4 Consumer Tax + SMB Finance TAM (INTU's Second Battleground)

INTU's dominant position in consumer tax (75%) is strongest, but this segment's TAM is experiencing negative growth (-1%/year). INTU's "second growth curve" is not in consumer tax, but in SMB Finance/Payments (+14%) + Credit/Wealth Management (+10%). The market viewed INTU as a "TurboTax company" from 2023-2025; after 2026, it should transition to a "QBO + adjacent products" framework. The speed of this transition will determine the pace of valuation re-rating.

16.5 Macro Environment: Interest Rates + Exchange Rates + Policy

The current Federal Funds Rate is approximately 4.5% (April 2026), which is moderately high. If the Fed cuts interest rates to 3.5% in the second half of 2026, the revaluation boost for ADBE would be slightly greater than for INTU, but INTU's distribution layer premium transition is independent of interest rates.

INTU focuses on the US SMB + consumer tax market (only 10% international), making it insensitive to USD exchange rates. Its localized moat naturally hedges against currency risk.

Policy Risk: INTU is the lowest (Direct File resolved), PTC is the highest (ITAR + dual data protection). Key policy concerns: ADBE's AI copyright disputes + EU DMA; ADSK's BIM Level 3 open data requirements + antitrust (55% market share in the AEC market).

The overall macro environment ranking is fully consistent with the rating ranking in this report — cross-validation from independent dimensions.

16.6 Historical Backtest: Performance of 4 Companies, 2021-2026

Observation 1: INTU's 5-year CAGR of 9.1% is the highest, also exceeding the IGV ETF's 4.6%. Over the past 5 years, INTU has been a "sector-outperforming" stock — demonstrating continuous compounding, not merely waiting for low valuation.

Observation 2: ADBE's CAGR of 4.1% lags INTU by 5pp, primarily due to 2025-2026 (AI disruption anxiety). Our "watch critical" rating for ADBE is a bet that this underperformance is emotional, not structural.

Observation 3: PTC's CAGR of -1.7% is the only negative, consistent with "cautious watch" — selective non-disclosure + the unresolved Onshape option indeed led to long-term underperformance.

Methodology Backtest: If, in 2021, a portfolio of 50% INTU + 30% ADBE + 10% ADSK + 10% PTC were constructed using Perspective One (Product Layer vs. Distribution Layer) categorization, the CAGR would be +5.97% vs IGV's +4.6%, an annual outperformance of +1.37pp, and a cumulative 5-year outperformance of approximately +7pp.

Limitations: Short sample period (5 years), survivor bias (all 4 companies are large-cap), hindsight bias. The +1.37pp annual outperformance is not an expected future return, but merely evidence that the methodology was effective historically.

16.7 PE Historical Trend

Source: Bloomberg 2026-04-09.

ADBE fell from 45x in 2021 to 9.6x in 2026, a contraction of -79% — extremely rare in SaaS history, comparable only to Zoom (-88%) and Peloton (bankruptcy-level). However, ADBE's fundamentals (OPM 47.4%, #1 in the SaaS industry) are nothing like Zoom's. The PE contraction is an emotional overreaction, not a reflection of fundamentals.

INTU fell from 35x to 18x, a contraction of -49%. The 18x bottom is not justified by distribution layer asset comparables (28-31x), implying potential for PE expansion.

If AI disruption anxiety completely dissipates, ADBE's upside is +150%. This report's base case assumes AI anxiety only subsides by 30-40%, corresponding to +22%.

Chapter 17: Common Rebuttals & FAQ

17.1 Rebuttals to "INTU as the Independent Leader"

Rebuttal 1: "The distribution layer moat has existed for 10 years, why is it only now worth +25%?"

Over the past 10 years, INTU's PE was between 22-35x (average 26x), reflecting the distribution layer premium. The compression from 25x to 18x between 2024-2026 was mainly due to Direct File anxiety. Direct File has been resolved, but the PE has not recovered, indicating mispricing. The +20~25% in this report refers to PE returning from 18x to 21x, not exceeding its historical average.

Rebuttal 2: "Can SBSE business growth of 16% be sustained?"

There is a risk of decline, which is the Kill Switch trigger point (SBSE ARR <10%). If triggered, we exit. Phase 4's asymmetrical odds of 2x already factor in this risk — a downside of -10~-15% is much smaller than an upside of +25~+30%. A decline to 12-13% would not cause the rating to fail; it would only adjust the probability-weighted Upside from +25% to +15%.

Rebuttal 3: "Does a distribution layer Fwd PE of 28-31x require 'undisruptable' hard evidence?"

Direct File was the "most powerful force to disrupt INTU," and it has been shut down. The event in November 2025 is a once-in-a-decade structural positive. If even Direct File could not disrupt INTU, other AI tools would find it even harder. This is the value of a natural experiment — it's not theory, it's fact.

Rebuttal 4: "Are 5 masters agreeing indicative of groupthink?"

5 位大师用完全不同的方法论 (巴菲特护城河 / 芒格反向 / Howard Marks 周期 / Klarman 赔率 / Druckenmiller 反身性). 他们对 ADBE 的意见是 3/5 反对, 说明方法论差异存在. 对 INTU 的 5/5 一致是因为 INTU 同时满足 5 种标准 — 真正的多方法验证.

17.2 Rebuttal to "ADBE Watch (Critical)"

Rebuttal 1: "9.6x is already so cheap, is (Critical) too conservative?"

Cheapness is a factor, not the entire picture. Klarman's 2x odds requirement (ADBE's 1.47x is insufficient) + Druckenmiller's reflexivity warning + Howard Marks' timing thesis, three independent signals all point to waiting. Cheap is not safe.

Rebuttal 2: "GenStudio +30% / Firefly +75% QoQ are both positive, why is it still Critical?"

The data is genuinely positive, but the data and the narrative are disconnected. The market's 9.6x implies a narrative of 'AI perpetually defeating ADBE', and good data should have triggered a re-rating but did not. Market sentiment has entered a phase where 'any good news is ignored' (negative reflexivity). Breaking this requires a structural catalyst (CEO successor), not an accumulation of good data.

17.3 Rebuttal to "ADSK / PTC Range"

Rebuttal 1: "ADSK Owner PE 161x is extreme, but Non-GAAP 19x is reasonable, why use Owner PE?"

We simultaneously look at Non-GAAP (19x neutral) + Owner PE (161x anomalous) + GAAP (45.9x neutral to high). Three inconsistent anchor points = inability to make a precise judgment. The range of -5%~0% is an honest expression of uncertainty.

Rebuttal 2: "PTC's lower bound of -10% is too pessimistic, shouldn't Onshape's success warrant optimism?"

Onshape's success probability is assigned ~10-15%, corresponding to an upper bound of +13%. The probability is low, and after weighting, it remains a base case of -7.5%. Selective non-disclosure is a negative Bayesian prior; 10-15% is already leaning optimistic.

Chapter 18: Investment Execution Checklist

18.1 Three Reader Types

Type A — Active Institutional Investors: Manage large sums of capital, conduct in-depth research on individual stocks, and are capable of internalizing all 19 chapters of analysis and independently verifying them. Execute INTU + ADBE ratings.

Type B — Professional Individual Investors: Manage their own capital, have limited time but possess fundamental analysis capabilities. Only execute INTU ratings (highest conviction target), with ADBE as an observation position.

Type C — Passive ETF Investors: Do not make single-stock selections, gain SaaS sector exposure through IGV (iShares Expanded Tech-Software ETF) or WCLD (WisdomTree Cloud Computing ETF). The value of this report for Type C lies in understanding the structural impact of AI on the SaaS sector, assisting with ETF timing.

18.2 Execution Framework: INTU (Type A + B)

18.3 Execution Framework: ADBE (Type A Only)

18.4 Do Not Execute: ADSK / PTC

Range rating = Do not initiate trades. Existing holders manage reductions based on Kill Switch signals (see Chapter 13); non-holders should not enter.

18.5 Investment Portfolio Construction Principles

Core Value of the Kill Switch: The Kill Switch is a tool to "limit downside", not a promise to "guarantee upside". When the Kill Switch is not triggered, positions operate according to valuation targets; when triggered, disciplined stop-losses limit single-trade losses. Portfolio-level expected returns depend on the weighted probability of triggers and the win rate of each target — this is precisely the logic behind the synergy between the Chapter 9 valuation system and the Chapter 13 Kill Switch Registry.

Chapter 19: Finale: The Left Hand and Right Hand of AI

19.1 The Picture

The same AI has two hands.

The Left Hand — Impacts the product layer. AI creates "good enough alternative tools" (Midjourney replacing Photoshop, Neural CAD compressing Revit's billable hours, open-source PLM threatening Windchill's data monopoly), leading to gradual customer attrition. ADBE / ADSK / PTC are all impacted by this left hand. The speed differs (ADBE fast, ADSK slow, PTC moderate), but the direction is the same — the moat is being linearly diluted.

The Right Hand — Elevates the distribution layer. AI enables CPAs to serve more clients at a lower cost, and CPAs still recommend QuickBooks. Will AI create better accounting calculations? Useless — clients hire CPAs not for calculations, but for "someone to be responsible when problems arise". AI lowers CPAs' service costs, which in turn strengthens the economic logic for 46,000 CPAs to recommend QuickBooks. INTU is elevated by this right hand.

The Real Explanation for the 2x PE Span (9.6x → 19.4x): The market knows AI is affecting SaaS, but it conflates the "left hand's strike" and the "right hand's lift" within the same "AI disruption discount". Those being struck and those being lifted are receiving roughly similar pricing treatment — this is the root of mispricing.

19.2 Final Ranking of the Four Companies

18.3 What to Watch For + What Valuation Language to Use Going Forward

Stop ranking these four companies using Rule of 40 / Fwd PE. Focus on two questions:

1. "Is AI impacting this company with its left hand or its right hand?" Left hand (product substitution) = linear moat erosion, use Reverse DCF to measure the degree of decline; Right hand (distribution enhancement) = moat expansion, refer to VISA / Mastercard / Moody's (PE 28-31x) valuation.

2. "Does the market's implied perpetual growth rate align with the impact of AI?" ADBE's implied g = -0.52% (perpetual contraction), but AI defense data exceeded all expectations — the left hand hit but didn't kill, yet the market is still pricing it as "killed." INTU's implied g = +4.0%, consistent with the right-hand logic, but the reference group is incorrect — using the VISA/MA reference group, the P/E should be 25-28x instead of 18x.

Valuation Language: For "left-hand" companies (ADBE/ADSK/PTC), use Reverse DCF to measure "how much decline the market has priced in"; actual data refuting this decline is alpha. For "right-hand" companies (INTU), switch the reference group to distribution layer assets (VISA/MA/MCO); current P/E significantly below the reference group is alpha.

© 2026 Investment Research Agent. All rights reserved.