Dassault Systèmes: A Zen Horizon Framework Deep-Dive Research Report

Hunting ten-year five-baggers among great growth stocks — pressing the upside question: "Can it get much bigger?"

• QHow high is its market ceiling? Is it expanding an existing pie, or creating an entirely new market?5/10

  A

  Bottom line: Dassault's market ceiling is high quality, but not especially high. The vast majority of its revenue comes from expanding a mature existing industrial software pie with high single-digit to low double-digit growth: PLM/CAD, simulation CAE, and life-sciences software together add up to only several tens of billions of dollars. It is not creating a wholly new market. For a company already at €6.2 billion in revenue and an oligopoly leader in every submarket, the Baillie Gifford-style “5x over the next decade” space does not hold up under Q1. Dassault looks more like the largest incumbent in a slowly widening body of water than a company standing at the start of a new exponential track. The “virtual twin / AI factory / Virtual Twin” story does carry the possibility of opening a new pie, but so far, including the 2026-02 “largest in 25 years” NVIDIA partnership, there is no quantifiable revenue. It should be treated as an option, not as the ceiling.

  Start by sizing the pie. Dassault's 3 main arenas are all classic mature oligopoly markets. Simulation CAE is expected to grow from $12.28 billion in 2025 to $19.96 billion in 2030, a CAGR of about 10.2%, consistent with the report's $12.3 billion to $20.0 billion framing and the most respectable growth of the 3 areas. PLM software is expected to grow from about $28.0 billion in 2024 to around $43.5 billion in 2030, a CAGR of about 7.7%. The life-sciences clinical software submarkets start from smaller bases: clinical trial management systems (CTMS) were only about $2.44 billion in 2025 and about $4.89 billion in 2030, while clinical data management systems (CDMS) were about $6.35 billion in 2025. Roughly adding these areas, including mainstream 3D/CAD where SOLIDWORKS sits, Dassault's addressable software pool is only in the several tens of billions of dollars, with overall annual growth in the 7%–10% range. In other words, the pie itself is only thickening at a high single-digit to low double-digit rate. That is the fundamental backdrop behind revenue growth falling from about +12% in 2022 to FY2025 +4% (Q4 only +1%), with 2026 guidance of just +3–5%. The market has not collapsed, but the ceiling is visible, and Dassault is already the player closest to it.

  So how much “5x room” is left? Do the honest arithmetic. Dassault's FY2025 software-related revenue was about €6.2 billion, a little over $7.0 billion. Its 3 main submarkets in 2030 together are only in the one-to-two-hundred-billion-dollar range, and that pool must be split among oligopolists including Siemens (PLM #1, having just paid about $10.0 billion for Altair), Synopsys / Ansys in simulation, PTC, Autodesk, and Veeva. Even if Dassault holds or slightly expands share in each line, “expanding the existing pie + taking share” points closer to industry-level high single-digit compounding than to the Baillie Gifford demand for “5x in 10 years.” The answer to this question is not in Q1 (ceiling); it can only be sought in Q2/Q3/Q9: ARPU uplift from subscription migration, pricing power, and a second curve scaling up. Whether those levers are large enough is not answered by this question.

  The only variable that could turn “expanding the pie” into “creating a new market” is Dassault's bet that Virtual Twin spills over from “software for design and manufacturing” into an operating foundation for healthcare, cities, and full industrial lifecycles, layered with the 2026-02-03 official “largest in 25 years” NVIDIA partnership: connecting Virtual Twin to Omniverse/CUDA-X, while NVIDIA in turn uses Dassault's MBSE to design gigawatt-scale AI factories. If this narrative works, in theory it maps to a “physical-world AI simulation” pie far larger than today's CAE+PLM. But we have to be honest: so far there is no quantifiable revenue contribution or segment guidance, and the company has not provided a corresponding TAM reset. It is an upside option worth tracking over the long term, not a number that can be written into today's ceiling. Overall, Q1's answer is: the pie is high quality, the structure is excellent, and Dassault is the share leader, but the market is mature, growth is high single digit, and the ceiling is “steadily high” rather than “excitingly high.” The true blue-sky case depends entirely on whether virtual twins can rise from industrial software into a new market, and that moment has not arrived.

  Jun 4, 2026
• QCan its revenue at least double over the next 5 years? Will growth mainly be driven by volume, price, or new businesses?3/10

  A

  Bottom line: no. Doubling revenue in 5 years requires about a 14.9% compound growth rate (1.149⁵≈2.0), while Dassault Systèmes is currently at only +4% constant currency, with 2026 guidance of +3–5%, and actual Q1 2026 at +3% cc. The simplest arithmetic under the current trajectory is this: at 4% compound growth, revenue after 5 years would be about €6.24 billion ×1.04⁵≈€7.59 billion, only +22%. Even using the top end of guidance at 5% compound growth, the 5-year increase is only about +28%. Getting from here to “doubling” would require growth to jump from 4% to 15% and stay there for 5 consecutive years. That is fantasy even within the company's own 7–9% medium-term target, especially since Goldman Sachs has already cut its 2026–2030 medium-term growth forecast from about 7% to about 5%, below the lower end of company targets. Frankly, this is a clear failing dimension for Dassault in the Baillie Gifford 10 questions: measured by a 5-year revenue double, it is far out of reach.

  Breaking down the sources of growth makes clear what “doubling” lacks. Current growth is mainly driven by price/conversion, not volume or new businesses. Within software revenue, subscriptions and support were €4,624.1M (+6%), of which pure subscription was +11%. That looks bright, but it mainly reflects the SaaS transition moving existing perpetual-license customers into subscriptions, combined with price increases and cloud expansion. It is a billing-model switch + unit-price uplift for the same customer base, not a net expansion in market size or seat count. The mirror image is that traditional licenses and other revenue were €1,020.7M, −6%, with upstream licenses at one point down −13% in Q3 alone; this is the other half being cannibalized by subscriptions. Netting the two, true “volume” contributions, meaning net new customers and installed-seat expansion, are thin. This is clearest in segment data: Mainstream, the largest piece, grew only +2%; Industrial Innovation +6% and at one point only +1% in Q1 2026; Life Sciences was even −2%. When high subscription growth comes more from structural migration than incremental demand, it can support 6% ARR, but it cannot support the 15% needed for a double.

  What would need to happen for a “double”? Roughly, some combination of the following: the drag from traditional licenses would need to be fully cleared, taking the roughly €1.0 billion shrinking line at −6% to zero decline; the stalled Mainstream and Life Sciences lines would need to reaccelerate to double digits; and a new growth curve capable of contributing several billions of euros would need to emerge. The most imaginative candidate is the February 2026 AI partnership with NVIDIA, combining generative AI with industrial “digital twins.” There are also real bright spots in cloud: 3DEXPERIENCE Cloud was +30%, but from too small a base to change the whole company. The problem is that this AI curve currently has no quantifiable revenue at all; it is pure option value. In other words, a 5-year double is not a destination reachable by “pressing the accelerator on the current engine.” It would require a narrative-level reshaping (see Q5/Q10). Measured honestly by Baillie Gifford's search for 5x stocks over the next decade, Dassault is a high-quality company with stable cash flow and a deep moat, but it does not currently have a growth structure capable of doubling revenue in 5 years. That is one of the core reasons it is rated “Watch,” not “Buy.”

  Jun 4, 2026
• QFive years from now, what will take over as the next growth engine? Does this “second curve” exist today?3/10

  A

  Bottom line: Dassault Systèmes does not have a “realized second curve” today. Life Sciences (Medidata), which was supposed to take over, has not accelerated; it is bleeding at −2% (Q1 2026 fell further to −3%, down to €259 million). The genuinely exciting AI / virtual twin narrative is currently a zero-revenue “option,” not a “curve.” On the question Baillie Gifford cares about most, namely “who takes over after the existing PLM/CAD core matures and slows (Industrial Innovation +6%, Mainstream SOLIDWORKS only +2%),” Dassault's answer is weak at this point: it has a roadmap and imagination, but it lacks a new growth engine that is already running and can be discounted.

  Interrogate the 3 candidates one by one, separating “curves” from “options”:

  ① Life Sciences / Medidata: it was supposed to be the second curve; now it is a bleeding wound. Dassault spent about $5.8 billion to acquire Medidata in 2019, intending to turn clinical trial data management into a third growth pillar beyond CAD/PLM. But the segment's FY2025 revenue was €1.081 billion, 19% of the total, and fell −2%, making it the only negative-growth segment among the 3 major software divisions. Q1 2026 was still −3%, which management attributed to CRO headwinds and weaker 2025 orders. Behind this is Veeva's continuing encroachment: in August 2025, IQVIA and Veeva fully settled and IQVIA joined Veeva's CRO clinical data partner program, using Veeva EDC to run trials. In a market growing at an 11% compound rate, Dassault is losing share. Honestly, this is no longer a “curve that has not yet started”; it is a downward-facing curve. Management hopes an enterprise end-to-end platform can escape trial-volume volatility, but so far that is only a promise, not an inflection.

  ② AI / virtual twin / NVIDIA partnership: an option with huge imagination, but zero revenue today. On February 3, 2026, Dassault and NVIDIA announced a long-term strategic partnership to build industrial AI platforms: NVIDIA will use Dassault's MBSE to design its own AI factories, starting with the Rubin platform and integrated into the Omniverse DSX Blueprint; SIMULIA physical twins will connect to CUDA-X; BIOVIA will connect to BioNeMo for molecular discovery. The story is genuinely attractive, but Baillie Gifford's honesty standard requires clarity: this is a zero-financial-disclosure call option, not a curve. It has no quantifiable current or near-term revenue guidance and contributes nothing to FY2026's +3–5%. Whether, when, and at what scale it turns into Dassault's own subscription and cloud revenue, with cloud at 25% of software and +8%, cannot be discounted today. It is an “why the market has not yet realized it” upside idea, but it has not landed as revenue as a successor engine.

  ③ Virtual Companions / agentic AI: still at roadmap stage. Aura (business) is already live on the 3DEXPERIENCE platform, while Leo (engineering, referencing da Vinci) and Marie (science, honoring Curie) will launch during 2026. Embedding agentic AI into the core platform is a logical evolution and may be the most plausible future monetization path, but today it is a product roadmap, not a profit center. There is no standalone revenue line, much less a handoff.

  Overall judgment, tied to Q3: Dassault's “second curve” is weak right now. Put the 3 candidates in their honest places: Life Sciences is a declining old curve, NVIDIA/AI is an unmonetized option, and Virtual Companions is still on the roadmap. In other words, as the PLM/CAD core slows, there is no visible new engine today that is already being discounted and can carry the load. For a high-quality stalled company rated “Watch,” this is exactly the weakness Q3 should record: the second curve exists as an “option portfolio,” but not yet as a “realized growth engine”. Whether it becomes real depends on 2 independent signals: the NVIDIA partnership turning into revenue, and Medidata stopping the bleed and stabilizing. Until then, the second-curve narrative should not receive a high growth score.

  Jun 4, 2026
• QWhat is its core competitive advantage? Will this moat widen or narrow over the next 3 to 5 years?6/10

  A

  Bottom line: Dassault's moat is real and very deep, but its shape is “one deep central pile, surrounded by shallow edges.” The core, CATIA in high-end aerospace/auto design, is almost unshakable and should deepen over the next 3 to 5 years. But the edges, simulation, life sciences, and general PLM, are being squeezed by 3 forces at once. Net judgment: the core is deepening, the edges are under pressure, and the overall moat is likely “not narrowing, but also hard to widen.” The platform must keep pulling customers upward to lock them in.

  First, define the moat. It is stacked from 4 interlocking pieces. First, extremely deep switching costs: since the 1980s, CATIA has been embedded in the design workflows of Boeing, Mercedes-Benz (since 1982), BMW, Honda, Toyota, Dassault's own Rafale fighter, and Falcon business jets. All geometry and engineering data for an aircraft or a car, decades of design standards, and upstream/downstream supply-chain collaboration all live in this system. Moving away is not “changing software”; it is “rebuilding the entire engineering data asset,” which is nearly impossible in practice. Second, platform lock-in: 3DEXPERIENCE unifies design, simulation, manufacturing, and data on one foundation; the deeper customers use it, the more tightly they are locked in. Third, revenue stickiness: under the report's framing, FY2025 recurring revenue already accounted for 82% of software revenue, up from 80% the prior year, a natural result of the installed base. Fourth, scale: a network and ecosystem of >370,000 customers, plus ~22–24% of revenue poured into R&D and a 32.0% non-IFRS operating margin, among the best in the software peer group, give it the resources to keep defending the castle. These 4 pieces remain a textbook wide moat in the core battlefield, high-end CAD for aerospace and auto OEMs, with no near-term threat visible.

  But if we are honest about direction, the edges are leaking, and pressure is coming on 3 fronts simultaneously. Simulation: Siemens acquired Altair in March 2025 for about $10.0 billion, folding HyperMesh, OptiStruct and others into Simcenter, directly strengthening the portfolio against Dassault's SIMULIA. Synopsys also completed the approximately $35.0 billion acquisition of Ansys in July 2025. The whole simulation landscape has been reshaped by 2 huge deals, turning Dassault in this area from a “challenger” into a player squeezed by 2 larger rivals. Life Sciences: Dassault's Medidata remains a leading player in eClinical/clinical data, but Veeva already holds about 80% share in life-sciences CRM and is expanding through Vault into clinical territory. It is clearly the attacker, not the defender. General PLM: this signal is more direct. In ABI Research's July 2025 PLM assessment for large discrete manufacturers, Siemens ranked first, PTC second, and Dassault only third. It remains in the “leaders” quadrant, but it is no longer the top name. In MES assessments closer to manufacturing execution, Dassault has even fallen out of the leaders quadrant. The final longer-term concern is the business model itself: if AI-native design/simulation tools mature, they may eventually erode the old logic of charging by seat. That is still unclear, but it cannot be ignored.

  Summarize the direction in one sentence: there is no sign that the moat is narrowing, but it is also hard to say it is widening. The central pile is going deeper, while the edges are being eroded in simulation, life sciences, and general PLM by bigger or more specialized competitors. FY2025's full-year +4% cc, with Q4 revenue up only 1%, shows exactly that this wide and deep moat is currently enough to protect the base, but not enough to translate into accelerating growth. It protects cash-flow stability, not the next decade's expansion that Baillie Gifford most wants to see. For a company rated “Watch,” that is the crux: the moat is genuinely wide, but “wide moat ≠ high growth,” and Dassault has not yet crossed that gap.

  Jun 4, 2026
• QIf its core business is disrupted, does it have the DNA to reinvent itself? How does it treat mistakes and bad news?5/10

  A

  Bottom line: Dassault Systèmes has historically had a real and unusually strong “reinvention DNA,” but it is currently stuck at an awkward point. The DNA proves that it could turn in the past; it has not yet turned through this AI disruption cycle. More concerning, its recent handling of bad news carries a clear suspicion of “highlighting the good and soft-pedaling the bad.” A veteran leader suddenly stepping away at the worst performance point is a governance signal that should not be brushed aside. For the Baillie Gifford question, “does it have the ability to reinvent if the core is disrupted?”, the honest answer is: strong DNA, but current execution is unproven; on attitude, the negatives outweigh the positives.

  Start with reinvention DNA. The evidence is solid. Over more than 40 years, Dassault repeatedly disrupted itself: it was spun out of Dassault Aviation in 1981 to build CATIA; it acquired SolidWorks in 1997 to extend CAD from high-end aerospace into mid-market mechanical design; it acquired Abaqus in 2005 to build the SIMULIA simulation line; it took back the PLM direct-sales channel from IBM in 2008; it bet on 3DEXPERIENCE as a platform in 2012; and it acquired Medidata for about $5.8 billion in 2019 to step into life sciences while also starting the subscription transition. This was not a stress reaction after being hit. It was the “6 generations of industrial transformation” style of active iteration recorded in the report. An incumbent leader that can repeatedly disrupt its own installed-base charging model has rare reinvention capacity. In this AI cycle, it is not sleeping either: in February 2026, it entered a long-term strategic partnership with NVIDIA, connecting 3DEXPERIENCE virtual twins with Omniverse, BioNeMo and others to build “physically verifiable industrial world models” and “virtual companions” on agentic platforms. So, does it have the ability to turn? History says yes. It has written the turnaround script 5 or 6 times. But Baillie Gifford does not pay for nostalgia; the question is whether it can do it this time. This time's key risk is endogenous to AI itself: AI-native design/simulation tools may over the long term bypass the old “per-seat charging” model. Dassault may use AI to reinvent the platform, but AI may also pull away the foundation of its monetization. This layer of reinvention, in the report's judgment, has “not yet been realized.” There are partnership announcements, but no verifiable new growth curve.

  Then look at the more delicate question of “how it treats mistakes and bad news.” Here the problems are clear. Growth deceleration is a fact, not an accounting framing issue: Q1 2025 license revenue fell −10.5% YoY, even though guidance had called for positive growth; Q3 2025 licenses were −13%, and full-year revenue-growth guidance was cut from 6%–8% to 4%–6%, matching the report's anchor of FY2025 revenue at only €6.24 billion, +4% cc. The issue is not just the number; it is the narrative. Management repeatedly described Medidata's slowdown as “temporary,” but Life Sciences has dragged for multiple quarters and CRO headwinds have not gone away. When “temporary” is still being used in the second year without being disproved, it starts to look more like reassurance than disclosure. This is exactly the “highlighting good news, soft-pedaling bad news” smell that Baillie Gifford would penalize. The sharpest governance signal is this: Bernard Charlès, who had led the company for about 27 years and had only become executive chairman in 2024, suddenly resigned as chairman and board director on 2026-02-21 for “personal reasons,” just 10 days after the weak Q4 results on 2026-02-11, when the share price fell about 18%. Daloz then took both the chairman and CEO roles. The announcement framed this as a “carefully prepared transition,” but the sequence of “worst results → veteran exits 10 days later → one person holds both roles” creates tension between the calm wording and the abrupt timing. Either it is a coincidence, or the governance layer chose a polished narrative over transparent explanation in the face of bad news. Either way, for a growth company that should be frank about errors, this is a negative and further weakens already diminished board checks and balances.

  Overall judgment, tied to Q5: Dassault Systèmes has textbook reinvention DNA: 5 or 6 proactive pivots over 40 years, and it has already stepped onto the bridge in this AI cycle through the NVIDIA partnership. So the answer to “does it have the DNA to reinvent if the core is disrupted?” is “yes.” But whether it can turn this time is not yet proven, and AI is both its new story and a potential gravedigger for its per-seat monetization base. What really drags down this score is its attitude toward bad news: repeated polishing of Medidata's “temporary” weakness during successive guidance cuts, plus a veteran's abrupt exit at the worst performance point and a concentration of power, show governance blemishes around “highlighting good news and soft-pedaling bad news.” For long-term growth investors betting on “5x in 10 years,” reinvention DNA is necessary, and Dassault passes that test. But candor and governance transparency are prerequisites for converting that DNA into the next curve, and those are exactly where the question marks are today. The DNA earns credit; the attitude needs watching.

  Jun 4, 2026
• QDoes management, especially the founder, have a long-term horizon and deep alignment with the company? Is it willing to sacrifice current profits for the next 5 to 10 years?6/10

  A

  Net judgment: management passes the “alignment” test, and the anchor is solid, but it only matches half of the Baillie Gifford-preferred pattern. The alignment comes from the family's lock on control through double voting rights, not broad economic alignment among all managers. Evidence of long-term orientation is fairly strong: about 25 years of consecutive dividends, R&D consistently around ~22–24%, and continued investment despite weak demand. But its willingness to aggressively reinvest by depressing current profits for the next 10 years is limited. It looks more like a stable cash cow with high profitability (32.0% non-IFRS operating margin) plus meaningful dividends and buybacks than the kind of company Baillie Gifford prefers, one that sacrifices current gross margin for long-term share. Add recent governance flaws: the founder abruptly left at a weak-performance point, and the new leader now combines chairman + CEO, weakening checks and balances. Q6 is “neutral to positive, but with reservations.”

  ① Interest alignment: very strong control, limited economic alignment. The Dassault family's holding platform GIMD holds about 40.1% of capital and about 53.8% of voting rights (shares registered for more than 2 years receive double voting rights). Together with Charles Edelstenne and Bernard Charlès, GIMD's concert parties can control about 63% of voting rights. This means the family has absolute control at the voting-rights level (>50%), but only about 40% at the capital level, not absolute ownership. The remaining roughly 60% of capital is held by public and institutional investors, but corresponds to less than half of the voting rights. For Baillie Gifford, this cuts both ways. The benefit is firm control and insulation from short-termist capital pressure, which supports long-cycle investment. The cost is that this is family control, not the kind of skin in the game where management has its net worth tied to the company and shares the full upside and downside with all shareholders. Minority shareholder voice is structurally diluted. The report also notes that “minority shareholder influence is limited,” consistent with this.

  ② Long-term horizon: fairly strong evidence. The company has paid dividends for about 25 consecutive years and has not cut them in recent years (FY2025 proposed dividend of €0.27/share, to be voted on at the May 20 shareholder meeting). R&D has long been ~22–24% of revenue, and it has continued to invest through weak European auto/manufacturing demand, using most annual cash flow for R&D, acquisitions (Contentserv €189 million, increased Centric stake €252 million), and buybacks. In February 2026, around the €18 price level, it net repurchased about €340 million of treasury shares, about 3.5 million shares. That is management using real cash near its internal valuation anchor, a positive sign for “long-termism + confidence in intrinsic value.”

  ③ Willingness to sacrifice current profits for the long term: conservative, with a new governance variable. This is where it diverges from the Baillie Gifford pattern. A 32% high operating margin plus continued dividends/buybacks is the financial profile of a “defensive cash cow,” not an aggressive growth company voluntarily lowering current profits and pouring every euro into reinvestment for 5 to 10 years out. Growth has already slowed to only +4% cc in FY2025, Q4 +1%, with 2026 guidance of 3–5%, and the company has chosen to protect margins rather than trade margin for growth. A governance flaw must also be counted honestly: founder Bernard Charlès, who had led the company for about 27 years, resigned immediately as executive chairman and director on 2026-02-21 for “personal reasons”, retaining only an AI strategic adviser role. This came at a sensitive time, about 10 days after weak results were disclosed on February 11, which is poor timing. His successor Pascal Daloz has since combined chairman and CEO roles, weakening independent board oversight of management. Overall: interest alignment in the control-rights sense and long-term horizon are both present, enough to help Dassault avoid short-termist traps. But the aggressiveness of “sacrificing the present for the long term” is insufficient, and the founder's weak-point exit plus role concentration introduce recent flaws in continuity and checks and balances. Q6 is not a strong support for a “5x stock”; it is a solid but not ambitious passing item.

  Jun 4, 2026
• QIf it disappeared tomorrow, how much would customers miss it? Is its growth sustainable and not dependent on harming society or exploiting regulation?6/10

  A

  Bottom line: Q7 is one of the rare “high-score on both sides” dimensions for Dassault Systèmes under the Baillie Gifford framework. Its indispensability “if it disappeared tomorrow” is almost full-score, and the sustainability of its growth model is clearly above average. But the internal tension must be stated honestly: extreme stickiness comes precisely from the product becoming infrastructure, and infrastructure software is naturally sticky but slow. Indispensable ≠ high growth. This is one of the root reasons its rating is only “Watch,” not “Buy.”

  Start with indispensability. If CATIA / 3DEXPERIENCE disappeared tomorrow, aerospace and auto OEMs would not merely “miss” it; their core R&D workflows would seize up. CATIA was the design tool for the fully digital design of the Boeing 777, Dassault Falcon business jets, and the Rafale fighter, and it has long been core CAD for Mercedes-Benz, BMW, Honda, Toyota and other carmakers. Historically, 7 out of every 10 new aircraft and 4 out of every 10 new cars were designed with its software. The essence of this stickiness is data and process lock-in: decades of design data, parametric models, and engineer expertise for a product line are all tied to CATIA / ENOVIA, and the expertise customers build around CATIA itself makes them almost unwilling to consider migrating. The same logic applies in life sciences: Medidata is a partner to 19 of the top 20 pharmaceutical companies globally, has supported more than 38,000 clinical trials, and holds one of the industry's largest clinical datasets; moving away means rebuilding the regulatory compliance chain. The report's FY2025 recurring revenue at 82% of software revenue and more than 370,000 customers is the financial projection of this “once installed into process, it cannot be pulled out” reality. A near-full score on this dimension is justified.

  Now consider whether the growth model is clean, sustainable, and not reliant on harming society or exploiting regulation. The answer is strongly positive, and even a positive example of data sovereignty. Dassault's growth relies on subscription migration (FY2025 subscription revenue +11%, recurring mix up another 2 percentage points YoY), which smooths revenue rather than harvesting one-off license pulses. It uses OUTSCALE sovereign cloud as a differentiator, and in its long-term NVIDIA cooperation it explicitly wrote localized AI factories across 3 continents, protection of customer data privacy, intellectual property, and sovereignty into the agreement. In the European regulatory context, this is positive differentiation by protecting customers rather than arbitraging regulation. Add policy-order resilience from defense/aerospace customers such as Rafale, and its growth logic neither depends on harming users nor on regulatory loopholes. The cleanliness is high.

  But the honest ending must highlight the tension. This near-perfect indispensability is a product of “industrial infrastructure” status, and the curse of infrastructure software is precisely sticky but slow. The same lock-in keeps customers from leaving, but it also ties Dassault's fate to customer capex cycles. When auto and aerospace R&D spending tightens and licenses shrink, FY2025 total revenue at only +4% cc and Q4 as low as +1% is the evidence. For Baillie Gifford's demand for “5x over the next decade,” Q7 proves that Dassault “deserves long-term holding because of moat depth,” but it does not by itself prove “growth is good enough.” It answers “will it disappear?”, not “can it accelerate?” This question deserves a high score, but readers should not mistake high stickiness for high growth. That is the key dividing line that keeps Dassault at “Watch,” not in the 5x candidate pool.

  Jun 4, 2026
• QWhat are the unit economics of this business (gross margin, incremental returns)? Do they improve or deteriorate with scale? Where does the money it earns go?8/10

  A

  Bottom line: unit economics are Dassault's strongest dimension. This is a “software franchise” level business, but the “gets better as it gets bigger” compounding engine has largely flattened. It is more “excellent and stable” than “excellent and expanding.” What Baillie Gifford truly wants is a compounding machine with great unit economics that continue improving as scale expands. Dassault satisfies the first half; on the second, the inflection point is past.

  ① Unit economics: excellent, earning software-franchise money. FY2025 revenue was €6.24 billion, +4% cc, with a 32.0% non-IFRS operating margin (IFRS about 21.7%). Over the past 10 years, that margin has stayed in a narrow high range of 31.9–32.8%, unusually stable for anything tied to industrial cycles. Incremental economics are also attractive: recurring revenue is already 82% of software revenue (80% in 2024), capex/revenue is <4%, and the business is hardly capital intensive. FY2025 operating cash flow was about €1.63 billion, with conversion to IFRS net income (diluted EPS €0.90, net income around €1.2 billion) of >100%. It also had about €1.53 billion net cash on the balance sheet. CATIA/SOLIDWORKS/DELMIA are deeply embedded in customer R&D and production workflows, with very high switching costs. That is the root of its long-term “franchise rent.”

  ② Do economics improve or deteriorate with scale? Honestly: marginal improvement is converging. This is the key tension with the Baillie Gifford framework. Growth has slipped from double digits to 4%. Although FY2025 margin still rose 40 basis points at constant currency, that mainly came from cost and expense discipline, not continued operating leverage from revenue scaling. With only 4% top-line growth, there is limited room to spread costs and push margins much above 32%; 2026 guidance is only 3–5% growth. Inside the business model, it is still “robbing Peter to pay Paul”: pure subscription +11% while traditional licenses shrink and Life Sciences is still -2%. Subscription transition improves revenue quality, but has not lifted overall growth. In other words, unit economics themselves have not deteriorated, but the “scale → better” curve has flattened. After reaching the top, it is moving sideways rather than continuing upward.

  ③ Where does the money go? Reinvestment intensity is high, but returns skew toward defense. Cash mainly flows to 4 places: R&D, which has long been about 22–24% of revenue (FY2025 around €1.5 billion), the core spending that maintains the moat; acquisitions (the roughly $5.8 billion Medidata acquisition in 2019 remains the largest, with ContentServ added in 2025); about 25 years of consecutive dividends (FY2025 €0.27/share); and ongoing buybacks (full-year 2025 net treasury-share repurchases of about €340 million, plus another 3.5 million shares bought around €18 in February 2026). The issue is that high reinvestment is producing 4% growth, so marginal capital productivity is falling. Direct returns to shareholders are also thin: dividend yield only about 1.4%, and PB remains high. Net judgment: unit economics are first-class and the cash machine is solid, but the growth-compounding story of “larger scale earns increasingly more” is nearing its end. From a Baillie Gifford lens, this is a classic “high quality but stalled” case. The quality is good enough for the watchlist, but it is not the kind of 10-year 5x compounder whose scale self-reinforces growth.

  Jun 4, 2026
• QWhat conditions must hold for it to rise 5x in 10 years? Are those conditions realistic? What expectations are implied in today's share price?3/10

  A

  Bottom line: for Dassault, a 5x over 10 years is a low-probability scenario that requires 4 difficult reversals to happen at once, not the base case. Lay out the numbers: going from a €25B market cap to about €125B requires about 17.5% annualized returns for 10 consecutive years. A company whose growth has slid from ~+12% in 2022 to +4% in 2025, whose own 2026 guidance is 3–5%, and whose Goldman Sachs medium-term 2026–2030 forecast has been cut to about 5%, cannot deliver 17.5% shareholder returns through earnings compounding alone. Revenue growth, AI monetization, valuation rerating, and cyclical recovery must all arrive together. That is a long way from today's 4% reality.

  Break down the necessary conditions and realism one by one. ① Revenue growth must reaccelerate sharply: to support a 5x, revenue needs to return to double digits or close to the company's own 7–9% target, rather than today's 4%. But the drag is structural: Industrial Innovation Q4 only +1%, license contraction, and weak European manufacturing and auto demand. This is not something 1 or 2 quarters can fix, so realism is low. ② The AI/NVIDIA partnership must move from roadmap to real quantifiable revenue: the “largest in 25 years” partnership announced on February 3, 2026 is still a platform architecture and vision, with no consolidated revenue contribution. Becoming a second growth engine would take years, so realism remains unproven and it cannot solve the near term. ③ Valuation must not compress, and may even need to rerate upward from ~15× forward PE: current forward PE [~14×](https://stockanalysis.com/quote/epa/DSY/statistics/) is near the low end of the past 10 years (about ~36× at the end of 2024, historical peak ~55×). A 5x script usually needs a “Davis double play” of valuation and earnings, but with 4% growth and no confirmed stabilization, the market is more likely to maintain a low multiple than proactively pay a premium, so realism is low to moderate. ④ Life Sciences must stop bleeding + auto/manufacturing must recover: Medidata is dragging 2026 performance, and Life Sciences is bleeding at −2%. This area must first stop declining before contributing growth. All 4 conditions are hard; requiring them to happen at the same time creates a very low joint probability. This is Dassault's clearest weakness under the Baillie Gifford framework. Honestly, with the 4% current trajectory and a valuation already around 15×, a 10-year 5x is not the base expectation.

  What today's share price implies is the one upside asymmetry worth highlighting. At €20.16 and forward PE ~15×, the price implies a bearish assumption that “low-speed growth persists for a long time and the AI option is basically not counted.” The market is valuing the NVIDIA partnership's potential monetization near zero and anchoring permanent growth at 4–5%. This is the mirror image of peers rerated to the top of their valuation bands by AI narratives: others have optimism fully in the price; Dassault has pessimism almost fully in the price. Honestly, the 4-part reversal needed for a 5x is hard to assemble, and Dassault is more likely to be a high-quality, reasonable but range-bound name than a 5x growth stock. But the low starting valuation means the downside has support from the €18 buyback price and cash flow, while any positive surprise on even 1 line, such as growth stabilizing or AI showing initial revenue, could trigger asymmetric valuation repair. This is not a 5x story. It is simply that under “low expectations + low valuation,” upside odds look better than downside odds.

  Jun 4, 2026
• QWhy has the market not realized all this yet? Does it not understand, does it look down on it, or does it not look far enough? What would become the “narrative inflection point”?4/10

  A

  Bottom line: Dassault is not a “great company the market does not understand.” It is a two-sided name that is “cheap for reasons, while the AI option is indeed counted at zero.” “Not understanding” basically does not hold. “Looking down on it” does hold, and the market is mostly right. Only “not looking far enough” is the real and differentiated upside option here. These 3 layers must be separated to avoid mistaking “low valuation” for “the market is wrong.”

  First, reject “not understanding.” Dassault has extensive sell-side coverage: about 21 institutions follow it, with a consensus “Hold.” On 2026-03-04, Goldman Sachs downgraded the rating from Buy to Neutral and cut the target price from €29 to €20, explicitly saying the 3–5% 2026 guidance was “realistic,” while visibility on growth reacceleration was limited. This is not an awareness gap. It is a group of people seeing the situation clearly and pricing it calmly. So the real question here was never “does the market understand?” It is “what is the market bearish on, and what is it missing?”

  “Looking down on it” holds, and the market is mostly right. This must be said honestly. The market is putting a forward PE of ~15× on Dassault's 4% growth (FY2025 revenue +4%, Q4 only +1%, 2026 guidance 3–5%), versus a historical normal range of 35–45× and now near the lowest point in 10 years. In essence, the valuation model is fully pricing in “low-speed perpetuity.” This is not emotional mispricing. Dassault has missed growth expectations for 2 consecutive years (2024, 2025), Q4 license revenue fell −7% YoY to €358M, Life Sciences is bleeding, and European auto and manufacturing demand is weak. Goldman used this to cut medium-term 2026–30 growth expectations from ~7% to ~5%, below the company's own 7–9% target. Discounting an industrial software company with stalled growth, a rising subscription mix but no overall stabilization, is largely “right.” It is cheap for a reason.

  The truly differentiated piece is “not looking far enough,” and this is the mirror image of Infineon/Delta. All sit within the “industrial × AI” narrative. Infineon has been rerated to the top of its valuation band because AI data-center power-chip demand is visible, with AI revenue already quantifiable in the financials; Delta is similar. Dassault is the opposite: it is treated as a “mature, slowing, cheap” software stock. Its 2026-02-03 “largest in 25 years” partnership with NVIDIA (Virtual Twin × Omniverse/CUDA-X, while NVIDIA in turn uses Dassault MBSE to design gigawatt-scale AI factories) is almost counted at zero in a price a little above €20 and forward PE of 15×. The key difference is “quantifiable.” Infineon's AI is revenue already in the financials; Dassault's AI is still on the roadmap, with no quantifiable revenue at all. The market refuses to pay for an option without revenue numbers. That logic is coherent, but it also leaves all long-term upside outside the price.

  Net judgment and narrative inflection point. The honest answer to Q10 is therefore: Dassault is neither simply “a great company underestimated by the market,” nor should it be lazily labeled a “value trap.” It is a name where “the current price fully embeds bad news, but prices the good option at zero.” Downside has 15× PE and steady cash flow as support; upside depends entirely on the still-unrealized AI option. The narrative inflection point will not come from valuation itself. It needs an inflection signal that can be booked in the financials: any quarter where constant-currency growth stabilizes back to 5–6%, or Life Sciences turns positive from −2%, or the NVIDIA partnership discloses its first quantifiable revenue contribution. If any 1 of the 3 happens, the market's “low-speed perpetuity” pricing premise will be broken, and the ignored AI option will begin to enter the price. Until then, it is cheap for a reason, and the AI option is indeed undervalued; both are true at the same time.

  Jun 4, 2026