PTC: A Fine Moat That the Market Has Stopped Discounting
Before the Gates
PTC sells subscription software to manufacturers. Engineers design products in its CAD tools (Creo, Onshape), manage the resulting data, design changes, and part structures across the product’s entire life in its PLM tools (Windchill, Codebeamer, Arena), and connect the shop floor through ServiceMax. Of roughly $2.74 billion in FY2025 revenue, about 95% is recurring, and ARR sits above $2.3 billion, growing around 10% a year. In March 2026 the company sold its non-core IoT businesses (Kepware, ThingWorx) for $523 million and refocused on the CAD-plus-PLM “digital thread.”
That is the business. Now the four gates.
Gate 1: Circle of Competence
The demand here is not mysterious. Manufacturers will still need to design products and manage product data in ten years; saying so is analysis, not a wager. Once a PLM system becomes the system of record for product data, replacement cycles stretch past a decade, and the competitive field — Dassault, Siemens, Autodesk — has been a stable oligopoly of three or four players for decades. Nothing about the industry’s pace of change threatens the predictability of the forecast.
This sits squarely inside our declared area of competence: software businesses whose lock-in structures can be observed in public data. Gate 1 passes.
Gate 2: Moat
The primary source is switching costs, supported by decades of accumulated domain IP. A customer whose product data lives in Windchill, integrated with Creo, Codebeamer, and ServiceMax, faces migration costs, downtime, and retraining running into the millions of dollars to leave. Morningstar reaches the same conclusion.
Evidence of pricing power: after the subscription transition completed (that 95% recurring figure), ARR is still growing 9–13% annually in constant currency — CAD +8%, PLM +9% in Q1’26 — without meaningful churn. Net expansion from the existing base at that rate means price increases and upsells are landing. Gross margin of 83.8% and net margin of 26.8% rank near the top of the industry.
The direction is slowly widening. Codebeamer is stacking additional lock-in next to Windchill in automotive and medical devices; the IoT divestiture concentrates capital on the moated core; and while PTC’s PLM market share of 9.6% trails Dassault’s 16.5%, ABI Research ranked it first in large-manufacturer PLM for 2024. Dassault and Siemens’ scale remains a permanent check.
On AI: design and product data are proprietary and buried deep in customer workflows, and AI design assistance — such as the NVIDIA integration into Creo and Windchill announced in July 2025 — works only on top of that data. When models are commoditized, what remains is data plus switching costs. AI-native CAD startups exist as an erosion scenario, but regulated manufacturers in aerospace, automotive, and medicine have little incentive to move their product data foundation to unproven tools. Moat erosion within ten years looks unlikely; if anything, AI strengthens the position.
Gate 2 passes.
Gate 3: Management
CEO Neil Barua took over in February 2024, previously running ServiceMax. His capital allocation record so far is disciplined: roughly $300 million of buybacks in FY’25 alongside debt repayment; a large step-up to $1.225–1.325 billion in FY’26 — expanded while the stock sat near its 52-week low, down 29% year-to-date, which is what buying below intrinsic value is supposed to look like; and the $523 million IoT divestiture immediately followed by a $375 million accelerated repurchase. There is an explicit stated principle: return roughly 50% of free cash flow when debt/EBITDA is under 3x.
One demerit, inherited: the prior CEO’s $1.46 billion ServiceMax acquisition in 2023 carried a disputed price. The current regime’s allocation, however, has been consistently shrinking the empire rather than building it.
On candor: the company publishes constant-currency ARR alongside reported figures, removing its own currency illusion, and cleanly separates one-time divestiture gains. No red flags. The retained-earnings test fails only on the recent drawdown — a quotational loss — while five-year ARR and double-digit FCF compounding support genuine value creation. Gate 3 passes.
Gate 4: Price
Owner earnings run about $825 million, consistent with FY2026 FCF guidance of ~$850 million — reasonable for a CapEx-light software model. Applying a 15x multiple to a business with ~10% ARR growth and 95% recurring revenue yields roughly $12.4 billion of intrinsic value, with a conservative floor of $9.9 billion at 12x — about $86 to $108 per share on ~115 million shares.
Note the reported PE of 11.76 is distorted: it includes a $463 million one-time gain from the IoT divestiture booked in FY2026 Q2. Normalized, the multiple is around 17x.
The judgment is made on owner earnings. At the thesis date, the market cap of $14.28 billion (~$124) stood at roughly a 15% premium to the upper estimate — no margin of safety. The 5.8% owner-earnings yield cleared the 10-year Treasury (~4.3%) but not by a “meaningful” amount. The target purchase price: 70% of intrinsic value, or a market cap of ~$8.7 billion, about $75 per share.
What followed is instructive. Q3 FY2026 results (July 29) validated the business rather than the price: constant-currency ARR of $2.448 billion, up 9.1%; net new ARR of $60 million above the high end of guidance; FCF guidance raised; a new $2 billion buyback; $525 million of opportunistic repurchases. Not one of our falsification triggers fired. The stock, sensibly, went up — $124 at the thesis date, $117.79 on July 22, then $132.46, $147.71, and $151.02 by August 10 as the good results were digested.
The owner-earnings yield fell in lockstep: 6.08%, 5.40%, 4.84%, 4.74%. Each snapshot confirmed the same quiet irony — the better the business performed, the farther the price moved from the $75 entry point. By August 10, the market cap of $17.41 billion sat above even the upper intrinsic value estimate, and the yield’s spread over the 10-year had thinned to a rounding error.
Gate 4 fails.
Verdict
Watch. Gates 1 through 3 pass cleanly: a durable switching-cost moat that AI reinforces rather than erodes, disciplined and candid management buying back stock at the lows, and an earnings base confirmed by a quarter in which ARR, FCF, and guidance all beat. But a good business is not a good investment at any price, and the market has repriced the quality faster than the intrinsic value can grow into it. The margin of safety has not merely failed to appear — good news has actively carried it away.
What would change the verdict: a return toward the target price of roughly $75 per share (~$8.7 billion market cap, a 30% discount to intrinsic value), or sustained evidence of a higher growth base — while watching for the specified warning signs: constant-currency ARR growth below 6% for two consecutive quarters, net churn, FCF guidance cuts, or buybacks giving way to a large acquisition.
This analysis is AI-generated, educational, and not investment advice. Figures may contain errors or be delayed. Disclaimer