Autodesk: A Deep Value Investment Analysis

Fact: What does the cost structure look like? Autodesk's cost structure is very typical. FY2026 total revenue was 7.206 billion dollars, cost was 650 million dollars, gross profit was 6.556 billion dollars, and GAAP gross margin was about 91.0%. The main expenses are sales and marketing (2.373 billion dollars) and R&D (1.643 billion dollars), showing this is not an asset-intensive business but a knowledge platform built on R&D, sales, channel, and customer success. FY2026 capital expenditure was only 43 million dollars, a capex intensity of just about 0.6%.

Fact: Does it depend on a few customers, suppliers, channels, policies, or key people? Historically it relied heavily on distributors, but that dependence is falling. The revenue share contributed by TD Synnex dropped from 39% in FY2024 and 33% in FY2025 to 14% in FY2026, while direct sales rose from 42% in FY2025 to 63% in FY2026. The company explicitly states its business no longer "materially depends" on TD Synnex, which is healthier than in earlier years. On the other side, at the product level it still relies heavily on a few core solutions, and the company itself lists "a substantial portion of revenue coming from a small number of solutions, including the AutoCAD family and Collections" as a risk factor.

Inference: Is this business simple, transparent, and easy to understand? The bar to understand it is not low, but if you are familiar with enterprise software and professional design tools, the business logic is clear: must-have software for professional use cases plus high switching costs plus subscription pricing plus a measure of cloud and platform expansion. The complexity is not in the business model but in the key accounting conventions, especially free cash flow, capitalization of contract acquisition costs, and the effect of channel and billing model changes on cash-flow timing. The business model itself is clear; the accounting presentation is more complex than the business model. Business understandability score: 4.5/5.

Opinion: If the stock market closed for five years, would I hold this business? Yes to owning the business, but that does not mean buying the stock at any price. Even in the 2020 pandemic environment the company delivered FY2021 revenue of 3.790 billion dollars, up 16% year over year, with operating cash flow of 1.437 billion dollars, showing it is not a fragile software company. In FY2026, facing high interest rates, geopolitical conflict, and economic uncertainty, revenue still grew 18% with free cash flow of 2.409 billion dollars, again showing strong resilience.

Industry judgment: growth or mature? The industry is not one-dimensional. Core CAD and design tools are relatively mature, but architectural digitization, construction cloud, manufacturing collaboration, industrial software moving to the cloud, and the AI overlay bring new structural growth. Autodesk itself puts its future emphasis on cloud, platform, and AI, and in FY2026 reorganized to tilt resources toward these directions. My judgment: the core business is mature and solid, while the incremental businesses are still in a growth phase.

Is long-term industry demand stable, and is it easily disrupted? Long-term demand is stable, because architecture, manufacturing, engineering, and media content production will not disappear. But tool form factors will change, especially with cloud-native CAD, AI automation, collaboration platforms, and the reworking of industry workflows. In its risk factors, Autodesk repeatedly flags technology change, intensifying competition, privacy and cybersecurity regulation, export controls, and government procurement requirements. My judgment: industry demand is stable, but the technology paradigm faces medium-to-high intensity evolution.

Main competitors and industry position. In manufacturing and product design, PTC and Dassault Systèmes are important rivals. In infrastructure and engineering software, Bentley is one of the direct comparables. As a broader software-quality comparison, Adobe's subscription model and cash-flow structure are instructive. Autodesk holds a strong position in architecture, design, and cross-workflow integration, but it is not a monopolist without rivals.

Opinion: a good company in a good industry. Industrial, construction, and design software is a classic good industry, and Autodesk is indeed a good company within it. It is not an excellent company in a bad industry. Industry attractiveness score: 4/5.

Moat and Management

Moat judgment. I break Autodesk's moat into six layers.

The first layer is high switching costs. The evidence is not a marketing slogan but operating results: recurring revenue at 97%, NR3 above 100% to 110%, RPO of 8.3 billion dollars, and revenue spanning multiple professional industries, all showing customers do not churn or switch easily. For architects, engineers, construction teams, and manufacturing design teams, replacing a core design-workflow tool means retraining, file compatibility, process redesign, rebuilding collaboration, and potential project risk, costs that usually exceed the software license itself. This is Autodesk's most important moat.

The second layer is scale advantage. The company posted FY2026 revenue of 7.206 billion dollars, GAAP gross profit of 6.556 billion dollars, and R&D expense of 1.643 billion dollars, covering architecture, manufacturing, and media and entertainment ecosystems. R&D and sales investment of this scale is not easily replicated by a small company. In professional software especially, scale brings broader product lines, more vertical solutions, stronger brand trust, and a larger partner network.

The third layer is brand and embedded industry standards. Here I am deliberately restrained: I do not define it as the consumer-style "brand premium" but more as a professional tool brand plus historical inertia plus organizational training assets. The company describes itself as a global leader in 3D design, engineering, and entertainment technology solutions, and its revenue scale and product distribution show it is not a marginal player in multiple segments. This is more of a soft moat rooted in professional habit and industry standards. This is an inference, not something the company quantifies directly.

The fourth layer is data and platform capability. For now this is more of a potential moat than a proven one. In the FY2026 annual report and the FY2026 Q4 earnings discussion, management emphasized shifting resources toward cloud, platform, and AI, and claimed it holds specialized data, context, and expertise. My judgment: if Autodesk can truly turn design data, construction data, manufacturing data, and cloud workflows into effective fuel for AI, the moat could widen. But as of now this is more a future option than a certain advantage already realized in the financials.

The fifth layer is channel and customer data advantage, which is being reshaped. FY2026 direct sales rose to 63%, meaning the company increasingly controls customer contact, pricing, renewals, and product usage feedback directly. This is a plus for the long-term moat, though in the short run it brings execution risk from the transaction model switch, billing cadence changes, and organizational adjustment.

The sixth layer is cost advantage and network effects. I score these two lower. Autodesk's cost structure is excellent, but that does not mean it is "cheapest" for customers; it sells high value, not low price. Network effects are not its core moat either; it has the weak network effects of a product, data, and interface ecosystem rather than the strong network effects of a social platform.

Overall judgment: a strong moat, but not impregnable. I see Autodesk's moat as broadly stable with room to widen in places. The core switching-cost and scale moats are thick; the brand and industry embedding are stable; the data and AI moat is still forming; cost advantage and network effects are not the main engine. For a competitor to fully replicate Autodesk's position across workflows like AutoCAD, AEC, and Fusion usually takes years of product refinement, industry validation, and ecosystem building, not a single year of throwing money at the problem. Moat strength score: 4/5.

Pricing power, inflation resistance, and recession resistance. Autodesk has some pricing power, but not unconstrained pricing power. The evidence is strong renewal and expansion performance, NR3 still above 100% to 110%, and revenue and RPO continuing to grow through a macro-uncertain environment. Its performance in 2020-2021 and FY2026 shows it can keep growing and generating cash even during economic downturns. My view: it has a moderate and sustained ability to raise price and value, not the kind of business that can push through double-digit price hikes that customers simply accept.

Is management trustworthy? This is where Autodesk most needs a discount right now. On the positive side, the board and management have set mandatory ownership rules, with the CEO required to hold stock worth six times base salary and other executives three times; the company has adopted a clawback policy; a substantial portion of executive long-term incentives uses performance shares, and FY2027 PSU metrics are shifting toward NNARR and profit targets more tied to recurring growth. These show an institutional long-term orientation. The company also disclosed that in FY2026 it engaged with the shareholder holding more than half of its shares.

On the negative side, actual insider ownership is not high. Per the 2026 proxy statement, the CEO held beneficial ownership of about 115,200 shares, and all directors and officers together held 294,000 shares, each less than 1% of shares outstanding. More importantly, the 2024 audit-committee internal investigation into the company's free-cash-flow and non-GAAP operating-margin practices, even though the SEC and USAO closed the matter in August 2025, was acknowledged by the company to be time-consuming and costly, and it triggered shareholder litigation and potential additional costs. For a long-term owner, a regulatory closure does not mean the governance flaw equals zero.

Is capital allocation rational? On reinvestment, the company is putting more resources into cloud, platform, and AI, which is reasonable in direction. On the balance sheet, the company had not drawn on its 1.5 billion dollar revolving credit at FY2026 year-end, held cash and marketable securities totaling nearly 2.973 billion dollars, and carried negative net debt, a sound financial position. The biggest issue is the combination of buybacks and stock-based compensation: FY2026 buybacks were 1.402 billion dollars, FY2025 buybacks were 852 million dollars, and FY2024 buybacks were 795 million dollars, but stock-based compensation expense over the same periods ran as high as 788 million, 683 million, and 703 million dollars. Share count is declining, but not quickly. This means a substantial part of the buybacks is using real cash to offset dilution from stock-based compensation rather than purely creating excess value for shareholders when the stock is undervalued.

My score for management and capital allocation: 3/5. Not because operating ability is weak, but because the "disclosure and governance caution" deduction is real, actual insider ownership is low, and the marginal efficiency of buybacks is not pretty enough.

Financial Quality and Owner Earnings

First, the core financial profile. Over the past several years Autodesk has demonstrated three important things: first, revenue keeps expanding; second, gross margin holds around 90%; third, capital expenditure is very low and cash-generating power is strong. Unlike many software companies that grow ever more cash-hungry, it is the type whose growth requires little physical capital investment.

Table note: FY2023 operating cash flow is a value back-solved from the net change in cash on the cash-flow statement, the investing and financing cash flows, and the foreign-exchange effect, and is therefore an estimate. FY2023 capital expenditure could not be fully confirmed in this public retrieval, so free cash flow is marked "Data needed." All other figures come from the company's original annual report / 10-K disclosures.

Revenue growth rate. From FY2021 to FY2026, revenue grew from 3.790 billion dollars to 7.206 billion dollars, a 5-year compound rate of about 13% to 14%. Looking at the longer FY2019 to FY2026 stretch, the compound rate approaches 16%. This is not an explosive SaaS growth rate, but for a mature professional software leader it is quite good.

Margin trend. Gross margin has held steady at about 90% to 91% over the long run, the typical mark of an excellent software business. Operating margin rose from about 10% in FY2020 to about 22% in FY2025/FY2026, showing scale effects and product-mix upgrades at work. FY2026 net income was roughly flat with FY2025, mainly because of higher taxes and restructuring charges rather than any failure of the operating model.

Operating cash flow and free cash flow. FY2026 operating cash flow was 2.452 billion dollars and free cash flow was 2.409 billion dollars, both at highs. FY2025 figures were 1.607 billion and 1.567 billion dollars; FY2024 figures were 1.313 billion and 1.282 billion dollars. The company's FY2027 guidance calls for free cash flow of 2.7 billion to 2.8 billion dollars and capital expenditure of about 50 million dollars, signaling management expects cash generation to step up further.

Free-cash-flow conversion and earnings quality. Comparing free cash flow directly with net income, FY2026 FCF/net income was about 2.14 times, FY2025 about 1.41 times, and FY2024 about 1.42 times. On the surface this is a company whose cash flow exceeds its profit, which is indeed a strength. But two points need emphasis: first, the company's heavy stock-based compensation is added back within operating cash flow; second, capitalization of contract acquisition costs and the billing-model switch affect working capital and cash-flow timing. So its profit is broadly real cash profit, but you cannot mechanically treat reported FCF as 100% distributable cash.

Return on capital. The company's ROE looks extremely high; a rough FY2026 calculation on average shareholders' equity approaches 40%, and FY2025 is even higher. But caution is needed here, because years of large-scale buybacks have depressed book equity, mechanically inflating ROE. More meaningful is to look at operating quality: high gross margin, low capex, net cash and low leverage, continued buybacks, and strong renewals all show the underlying economic return is genuinely very high, but book ROE is not the best single metric to use here.

Balance sheet and leverage. At FY2026 year-end the company held cash and cash equivalents of 2.249 billion dollars and short- and long-term marketable securities totaling 724 million dollars, for about 2.973 billion dollars in liquid financial assets. Net long-term notes were 2.483 billion dollars, and the company disclosed that its revolving credit was undrawn as of March 3, 2026. On this basis Autodesk is effectively in a slight net-cash position. FY2026 interest expense was 80 million dollars, and EBIT/interest coverage was about 20 times, so financial risk is low.

Receivables, payables, deferred revenue, and working capital. In FY2026 accounts receivable rose from 1.008 billion to 1.439 billion dollars, deferred revenue rose from 4.128 billion to 4.693 billion dollars, and other current assets rose from 588 million to 906 million dollars. Management explains that part of the change in receivables and working capital comes from fourth-quarter billing seasonality and the new transaction model and annual billing arrangements for multi-year contracts. My read: cash flow is still strong, but you have to admit working capital and accounting conventions increasingly deserve careful reading of the footnotes rather than only the headline FCF.

Are there signs of fraud or aggressive accounting? I will not lightly conclude "fraud." The known facts are: the company did draw an internal investigation over its free-cash-flow and non-GAAP operating-margin practices; the SEC and USAO closed the matter in August 2025; but the company still flags related litigation and future investigation risk. My judgment: there is not enough evidence to support a "financial fraud" conclusion, but there is enough evidence to support that the company's disclosure and non-GAAP presentation once had problems worth watching. For a conservative investor, that alone is enough to raise the valuation bar.

Owner earnings analysis. To estimate Buffett-style owner earnings, the key is not reported FCF but whether to treat SBC as a real cost.

My conservative approach is this: start from FY2026 reported free cash flow of 2.409 billion dollars and subtract FY2026 stock-based compensation of 788 million dollars. The reasoning is simple: SBC is not a current cash outflow, but it dilutes owners unless the company keeps spending cash on buybacks to offset it, and that buyback is itself a real cost. This yields FY2026 conservative owner earnings of about 1.621 billion dollars. On roughly 2.12 billion FY2026 year-end shares, that is about 7.6 dollars per share; at the current 243.63 dollar share price, that corresponds to about 32 times conservative owner earnings.

The broader cash rule is more optimistic: if you treat FY2026 free cash flow of 2.409 billion dollars directly as distributable cash, that is about 11.3 dollars per share, with the current price at about 21.6 times FCF. This is the core of the Autodesk valuation debate: if you accept that reported FCF is owner earnings, it looks merely a bit expensive; if you insist on treating SBC as a real cost, it is clearly not cheap. I personally lean toward the former as a reference and the latter as the conservative basis for decisions.

Intrinsic Value and Margin of Safety

Method one: discounted owner earnings. What follows is not fact but my valuation assumptions. I use conservative parameters because the user's risk preference is "balanced, leaning conservative," and because Autodesk's current valuation premium is high.

Note: The conservative scenario corresponds to "using FY2026 conservative owner earnings as the base, with AI as a supplement and no disruptive upside assumed." The neutral scenario corresponds to "the core business continuing to grow high single digits to low double digits, with a small margin improvement." The optimistic scenario more fully counts the long-term upside from cloud, platform, and AI, and from Construction and Fusion. The current 243.63 dollars sits roughly in the lower half of my optimistic scenario, above my neutral value range.

Method two: relative valuation. On how the market prices it, Autodesk is not cheap. The company's current market capitalization is about 52.38 billion dollars, with a TTM P/E of about 47.3 times. On FY2026 figures, EV/EBITDA is roughly 29 times and P/FCF about 21.7 times. By comparison, PTC currently trades at a P/E of about 14.2 times; Bentley currently trades at a P/E of about 38.3 times, with 2025 free cash flow of about 520 million dollars, an FCF yield of about 4.8% on its market cap, and recurring-revenue net dollar retention of 109%; Adobe currently trades at a P/E of about 14.8 times, with FY2025 revenue of 23.769 billion dollars, net income of 7.130 billion dollars, operating cash flow of 10.031 billion dollars, and RPO of 22.52 billion dollars. Business structures differ across peers, so you cannot compare mechanically, but at least one clear conclusion stands: Autodesk is not the cheapest among comparable software assets.

Method three: asset or liquidation value. Autodesk is not a fit for the asset method as a primary valuation framework. At FY2026 year-end, cash and marketable securities were about 2.973 billion dollars and net notes were roughly negative, that is, close to net cash. But goodwill was 4.295 billion dollars and intangible assets were 467 million dollars, a high share of assets. On a book basis, total shareholders' equity is only 3.045 billion dollars, and once you strip out goodwill and intangibles, the tangible cost protection is not strong. This means that if the business model were broken, liquidation value offers weak protection to shareholders; what really supports the valuation is ongoing cash flow, not an asset floor.

My intrinsic value range. Conservative intrinsic value range: 150-170 dollars. Fair intrinsic value range: 180-220 dollars. Optimistic intrinsic value range: 240-300 dollars. At the current 243.63 dollar share price, Autodesk is roughly: clearly premium to conservative value, at a small-to-moderate premium to fair value, and close to fair to slightly undervalued against optimistic value. This is not the odds structure traditional value investors prefer.

Ideal buy, acceptable holding, and clearly overvalued ranges. Applying a 25% to 30% margin-of-safety requirement, I put the ideal buy price range at 140-180 dollars. If you already own a high-quality asset and your tax cost is high, I see the acceptable holding price range roughly at 180-240 dollars. If the price stays above 260 dollars for a long time, I would tend to view that as a clearly overvalued range, unless the company proves that AI and platformization can deliver owner earnings materially above my current optimistic assumptions.

Margin of safety judgment. My answer: the current margin of safety is not sufficient. The most fragile valuation assumption is not "will revenue grow tomorrow" but "can owner earnings grow high single digits to low double digits over the next decade while the market keeps awarding a high multiple." If growth comes in below expectations, margins slip a little, or the multiple falls from above 30 times conservative owner earnings back to the low 20s, the return would be marked down meaningfully. Autodesk is very likely a textbook case of "a good company, but not a good price right now."

Risks, the Bear Case, and Comparison with Other Opportunities

The most important risk is not short-term volatility but permanent loss of capital. The first category is competition and technology substitution risk. CAD, BIM, PLM, construction cloud, and manufacturing collaboration are not uncontested. Cloud-native tools, AI-native design workflows, and vertical industry software could all erode part of Autodesk's value chain. If AI commoditizes part of the capability of professional tools, and Autodesk cannot turn its data and workflow advantages into higher added value, then the high valuation paid today becomes a problem.

The second category is governance and disclosure credibility risk. The 2024 internal investigation shows that, at least on metric presentation and convention governance, the company is not the "always reassuring" type. Although the SEC and USAO have closed the case, shareholder litigation, additional costs, potential future follow-up questions, and an investor trust discount on management are not zero. For a high-valuation company, a governance discount is often amplified by the market.

The third category is overvaluation risk. Reported FCF looks strong, but if you treat stock-based compensation as an economic cost, true owner earnings are not as cheap as the stock chart makes them look. FY2026 SBC was 788 million dollars, equal to about 10.9% of revenue. If the company keeps offsetting dilution with large buybacks in coming years, and the buyback price is not low, then growth in intrinsic value per share will not necessarily be as pretty as headline FCF.

The fourth category is end-market and cycle risk. Autodesk is software, but the architecture, engineering, and manufacturing industries its end customers operate in are still affected by interest rates, the property cycle, industrial capex, and government budgets. The company's risk factors directly mention high interest rates, inflation, trade wars, geopolitical conflict, regulatory uncertainty, and government procurement processes.

The fifth category is business-model convention risk. The new transaction model and annual billing arrangements for multi-year contracts change the cadence of billings, RPO, cash flow, and working capital, and the surface result can look stronger or weaker. The company itself cautions that such changes will continue to affect FY2027 billings and cash collection cadence. For investors who look only at headline cash flow, this is a source of misjudgment.

The strongest bear case. If I were short, I would put it this way: Autodesk is of course a good company, but the market already knows it is a good company. Buying it now essentially means paying all at once for "AI will widen the moat, Construction and Fusion will keep accelerating, management will not have another disclosure problem, and SBC will not erode true returns." If any two of these fail to hold, the current valuation is not cheap. What the bear sees is usually not a business collapse but "too expensive a valuation plus true owner earnings not as high as they look plus a governance flaw that warrants a discount."

What facts would make me admit I was wrong? If the following facts emerge, I would require myself to rewrite the investment thesis: First, NR3 falling below 100% and staying there for multiple quarters or years, signaling deteriorating renewal and expansion momentum. Second, RPO growth clearly stalling and recurring revenue share declining, signaling weakening subscription quality. Third, SBC running around 10% of revenue over the long run while buybacks cannot keep shrinking share count, signaling true owner earnings are overstated. Fourth, growth in Construction Cloud, Fusion, or the platform and AI businesses slowing while the core AutoCAD family also stalls. Fifth, another major disclosure or internal-control problem at the governance level.

Comparison with other opportunities. Against a broad index, I cannot say buying Autodesk is clearly better than buying the index right now. The reason is simple: it is a high-quality but high-valuation single stock with low error tolerance, while the index offers diversification and does not require betting on a single execution team. Against the risk-free rate, as of May 19, 2026, the U.S. 10-year Treasury yield was about 4.67% and the Aaa corporate bond yield about 5.72%. Autodesk's yield on FY2026 reported free cash flow is about 4.6%, roughly on par with the 10-year Treasury. On a conservative owner-earnings basis, the yield of about 3.1% is even below Treasuries and high-grade bonds. For conservative capital, this shows the risk compensation for buying at the current price is not thick.

Against comparable software opportunities, Bentley's recurring-plus-FCF structure is attractive, but its scale and platform breadth trail Autodesk and it carries 2.1 times net leverage. Adobe is larger, with stronger cash flow and a lower current P/E, but a different arena. PTC is an important alternative to watch in manufacturing and PLM, with a current P/E far below Autodesk's. My conclusion: Autodesk is a high-quality asset, but not the best-odds asset in the software sector right now. If a portfolio could hold only five assets, at "the current price" I would not rank it among the top.

Open questions and limitations. In this public retrieval, the full capital expenditure and free-cash-flow split for FY2023 could not be fully verified. In addition, the latest detailed annual financial conventions for PTC and Dassault are not fully developed here, so the peer relative-valuation section leans toward a "directional comparison" rather than exhaustive peer work. One more point: as of May 21, 2026, Autodesk's Q1 FY27 results had not yet been released. The investor relations page shows the Q1 FY27 earnings call is scheduled for May 28, 2026, so the most recent quarter's public data is not yet in this report.

Investment Checklist and Final Conclusion

Investment Checklist

Table note: This table is a synthesized judgment based on the facts, assumptions, and inferences above, not the company's original disclosure. Supporting data is referenced in the preceding sections.

【Final Rating】 Watch

【One-Sentence Investment Thesis】 Autodesk is a high-quality, high-stickiness, strong-cash-flow professional software business, but the current price pays more for future optimistic expectations than for a certain margin of safety.

【Core Bull Case】 First, revenue is highly recurring, with FY2026 recurring revenue at 97% and RPO of 8.3 billion dollars, giving extremely strong cash-flow visibility. Second, gross margin is about 90% to 91% over the long run and capital expenditure is very low, a superb asset-light software model. Third, switching costs are high, and customer workflows, training, and collaborative integration are hard to replace easily. Fourth, growth businesses such as AECO and Construction and Fusion are still expanding, with FY2026 revenue up 18% year over year. Fifth, the balance sheet is sound, close to net cash, with strong resistance to the cycle.

【Core Bear Case】 First, the current valuation is not cheap, and on a conservative owner-earnings basis the margin of safety is insufficient. Second, stock-based compensation is large, and a substantial part of buybacks merely offsets dilution. Third, the 2024 investigation weakened the credibility of management and disclosure conventions. Fourth, the AI and platformization upside is not yet sufficiently validated by financial results, but the market is already pricing it in. Fifth, while end industries are stable over the long run, they remain subject to architecture and manufacturing capex cycles.

【Key Assumptions】 For the investment to hold, the following conditions must be met: recurring revenue share stays high; NR3 stays above 100%; Construction, Fusion, and platform businesses keep contributing growth above the company average; SBC does not continue to worsen; governance and disclosure problems do not recur; and owner earnings can grow high single digits to low double digits over the next 10 years.

【Fair Buy Price】 My range is 140-180 dollars. The basis: this range roughly corresponds to my conservative-to-neutral intrinsic value range with a further 10% to 30% discount, providing more realistic protection against slowing growth, multiple compression, the true cost of SBC, and a governance discount.

【Target Holding Period】 Suitable for 10 years or more. This is not a stock for playing quarterly swings; if the buy thesis holds, the real value realization should come from industry digitization, platformization, and product-line penetration, not the next earnings season.

【Expected Annualized Return】 Near the current 243.63 dollar price, my subjective estimate is: Conservative scenario 3% to 5%; Neutral scenario 7% to 9%; Optimistic scenario 10% to 13%. The reason I do not give higher is that the current entry point is not cheap, and a substantial part of future return depends on whether results can cover the valuation premium.

【Maximum Loss Risk】 If growth slows and the market compresses the valuation from its current high back to a more ordinary software-company level, even without a business collapse the stock could see a mid-term drawdown of 35% to 50%. Add technology substitution, further damage to governance credibility, or a clear weakening of industry demand, and a permanent capital loss scenario of 60% or more cannot be ruled out, because the company's tangible asset protection is not strong.

【Tracking Metrics】 Going forward I will keep tracking: recurring revenue share, NR3, RPO and current RPO, AECO and Fusion/Construction growth, the gap between GAAP and non-GAAP operating margin, the SBC/revenue ratio, changes in total share count, operating cash flow and free cash flow, changes in deferred revenue and prepaid/contract acquisition costs, and any new regulatory, litigation, or internal-control disclosures.

【Signals That Trigger Re-evaluation】 If the following occur, the thesis must be re-examined: NR3 weakening continuously below 100%; RPO growth slowing markedly while revenue growth declines in tandem; SBC staying elevated for the long run while share count stops falling; AI and platform investment rising but no improvement in profit or retention; another major disclosure, investigation, or internal-control problem; an end-industry cycle turning weak and causing large customers to cut budgets.

【Final Recommendation】 Soberly put, Autodesk is very likely a quality company worth watching, and even owning, for the long term. But for an investor who is balanced and leaning conservative with a holding period of more than 10 years, "a good company" does not automatically equal "buy now." I would place it on a high-priority watch list and wait for two things: first, the price returning to a range with a real margin of safety; second, management proving over the next several quarters and years that the AI and platform transformation brings genuine growth in per-share owner earnings rather than a more complex narrative and a more expensive valuation. At the current price, my stance is not to reject the business but to refuse to pay up grudgingly.

This report is based on public information and does not constitute investment advice. Markets carry risk; invest with caution.