Skip to content

Dassault Systèmes SE

Company Overview

Business Overview

Dassault Systèmes was founded in 1981 as a spin-out of Dassault Aviation. It is headquartered in Vélizy-Villacoublay, France, and listed on Euronext Paris (ticker DSY, a CAC 40 constituent). It employs about 25,000 people in roughly 184 offices; around 41% work in R&D. Pascal Daloz has been CEO since January 2024 and, since February 21, 2026, Chairman & CEO (combined role). Long-time leader Bernard Charlès stepped down as Executive Chairman and left the Board on that date "for personal reasons," remaining available only in an advisory capacity. [Evidence] 2

Key financials:

  • 2025: revenue €6,235.8 M IFRS / €6,239.6 M non-IFRS (software €5,641.0 M IFRS), up 4% at constant currency but ~0% as reported (FX headwind). Recurring revenue grew 6% and subscriptions 11%. Non-IFRS operating margin 32.0%; non-IFRS diluted EPS €1.31. Reported February 11, 2026. [Evidence] 1
  • 2024 (directly verified): revenue €6.21 billion, software revenue up 6%, non-IFRS margin 31.9%, EPS €1.28. [Evidence] 3
  • 2026 guidance: growth of only 3–5%. This slowdown is deliberate — the company says it is spending 2026 as a "foundation year" for its Industrial-AI push. [Evidence] 1

Automotive Business

Automotive sits inside Dassault's Transportation & Mobility industry vertical, which spans car and light-truck OEMs, suppliers, trucks and buses, trains, mobility services, motorcycles and racing. The company does not publish automotive revenue separately. [Evidence — existence; thin on size] 4

Verified major automotive customers:

  • BMW Group — adopted 3DEXPERIENCE as its "future engineering platform," with around 17,000 users (announced February 2024). [Evidence] 20
  • Renault Group — 10,000+ users (2016) [Evidence], reported to have roughly doubled since [Inference]; a long-standing customer. 21
  • Jaguar Land Rover — 3DEXPERIENCE across all vehicle programs worldwide, 18,000+ users (May 2024). [Evidence] 23

Ford could not be verified as a Dassault customer from primary sources — Ford's history points toward the I-DEAS/Siemens tool lineage — so this report does not claim it. Tesla, initially unverifiable, is now confirmed as a CATIA/3DEXPERIENCE shop via Tesla's own job postings (established in this study's OEM research, 2026-08-02 — see the Tesla scope note). [Evidence of absence (Ford); Evidence (Tesla — updated 2026-08-02)]

Engineering Business

Dassault's core business is the engineering lifecycle itself: design (CATIA, SOLIDWORKS), simulation (SIMULIA, built on the well-known Abaqus solver), digital manufacturing (DELMIA), and product-lifecycle management or PLM (ENOVIA). These are increasingly sold as cloud "roles" on 3DEXPERIENCE. The company has started calling its PLM offering "IPLM" — "IP generation and management." [Evidence] 8

Role in Automotive Moulded Parts Workflow

Workflow Stage Relevant? Notes (product)
Requirements Yes ENOVIA; CATIA requirements/architecture apps; Aura orchestration
Industrial Design Yes CATIA surfacing / Generative Design; SOLIDWORKS
Concept Design Yes CATIA, SOLIDWORKS
CAD Modelling Core strength CATIA, SOLIDWORKS
Engineering Review Yes ENOVIA, 3DEXPERIENCE collaboration
Simulation Core strength SIMULIA (Abaqus); ML-accelerated for general physics, not moulding
DFM Partial — rules-based SOLIDWORKS DFMXpress (incl. injection-moulding rules); not AI
Tool Design Yes CATIA Mold & Tooling Designer (MTG)
Mould Flow Core strength — native solver SOLIDWORKS Plastics; SIMULIA Plastic Injection (INK/IME)
Prototype Yes SIMULIA, 3DEXPERIENCE
Validation Yes SIMULIA validation
Manufacturing Engineering Yes DELMIA
Production Release Yes DELMIA, ENOVIA

What makes Dassault structurally different: it covers the whole chain natively on one platform, and it owns the injection solver outright. But — as the next sections show — its AI layer has not yet reached the mould-flow or DFM stages. [Synthesis]

AI Strategy

Public AI Vision

Dassault's AI pitch rests on three pillars. [Inference, built on Evidence]

  1. Physics-grounded AI. Use its simulation engines to validate what the AI proposes, so the AI is "grounded, not hallucinating."
  2. European data sovereignty. Host everything on its own OUTSCALE cloud (which holds France's strict SecNumCloud security certification) and use the French AI company Mistral for language models — a "your data never leaves Europe" story.
  3. Proprietary data moat. Forty-plus years of customers' engineering data and IP locked into its platform.

It is explicitly not trying to compete on open, general-purpose AI models.

Investor Statements

  • CEO Pascal Daloz (February 2026): "In the Generative Economy, industry produces knowledge and know-how that generates objects: this is where true value lies." [Marketing] 9
  • Daloz, on the NVIDIA partnership: "When AI is grounded in science, physics and validated industrial knowledge, it becomes a force multiplier." [Marketing] 10
  • Business-model shift. Dassault is moving from per-seat licences toward value- and consumption-based pricing (new "units of knowledge, know-how and work"). 2026 is framed as the "foundation year," and growth guidance was deliberately lowered to 3–5% to fund the transition. [Evidence] 1

Engineering AI Strategy

Underneath the branding, Dassault's actual machine-learning work falls into three streams. [Synthesis]

  • Geometry ↔ CAD ML — e.g. inferring sketch constraints, or CAD features from freehand drawings and scans.
  • Surrogate simulation — train a neural network on a batch of physics simulations, then use the network to predict results near-instantly for new variants. (Also called ML-accelerated simulation.)
  • Semantic NLP and knowledge graphs — from its EXALEAD and Proxem acquisitions.

The large-language-model layer itself is bought in from Mistral, not built in-house. [Evidence] 6

Timeline of AI Evolution

Date Event
Jul 1, 2024 Mistral AI partnership; "LLM-as-a-Service" on OUTSCALE6
Feb 4, 2025 "3D UNIV+RSES" strategy reveal; PLM re-branded toward IPLM8
Feb 23–26, 2025 3DEXPERIENCE World — Aura unveiled as a SOLIDWORKS co-pilot
~Jul 2025 Aura goes live in SOLIDWORKS
Nov 26, 2025 Mistral partnership deepened (Le Chat Enterprise + AI Studio on OUTSCALE)7
Feb 3, 2026 NVIDIA partnership — "Industry World Models," Omniverse DSX10
Feb 11, 2026 Virtual Companions (Aura / Leo / Marie) announced; Leo & Marie "coming 2026"9
Jul 23, 2026 "AI-native agentic platform" expansion; all three companions live11

A misconception worth avoiding: Aura (2025) and the three-agent line-up (2026) are separate announcements, and Aura's own role changed between them — from a SOLIDWORKS design co-pilot in 2025 to a program/business-knowledge orchestrator in 2026. When citing "Aura," cite the dated source for the meaning you intend. [Evidence] 9

Products Relevant to Engineering

CATIA — the flagship CAD and systems tool. AI: Generative Design Engineering (topology optimization — software that "grows" a part shape to meet loads with minimal material), a Sketch Generative Constraint feature built on a graph neural network (the most concrete ML-architecture disclosure in the whole product line, though no accuracy figures are given), and the R2026x FD03 "Generative Experiences" release (July 2026) that turns text or scans into designs. Moulded parts: the Functional Molded Parts and Mold & Tooling Designer apps. [Evidence] 151417

SOLIDWORKS — the mid-market CAD product. AI: the Aura companion; 2026 brought generative drawings and auto-assembly; 2025 brought a Command Predictor and AI-assisted drawings. Moulded parts: SOLIDWORKS Plastics, sold in three tiers covering fill, pack, warp and cooling simulation. [Evidence] 18

SIMULIA — multiphysics simulation, built on Abaqus. AI: surrogate simulation — train a neural network on a subset of full simulations, then predict across the whole design space "in seconds." Moulded parts: the Plastic Injection roles (INK / IME) — the SIMPOE-derived native solver. [Evidence] 125

DELMIA / ENOVIA / 3DEXPERIENCE — manufacturing, PLM, and the unifying platform. Aura orchestrates across ENOVIA requirements and change management. [Synthesis]

AI Capabilities

1. Aura (Virtual Companion). An AI assistant. In its shipped 2025 form it is a SOLIDWORKS help-and-task chatbot trained on training guides and best-practice documents, private per user. In 2026 it was re-scoped as a program- and business-knowledge orchestrator. Limitation: the role keeps moving; the concrete, shipped thing is the 2025 chatbot. [Evidence] 9

2. Leo (engineering companion). Generates assembly structures, runs simulation studies, and assesses designs from text prompts. Demoed in February 2026 going from a prompt to a sketch, a parametric model, a 3D part and a fast surrogate stress check. Limitation: that was a staged demo; Leo was billed "coming mid-2026." [Evidence — staged] 9

3. Marie (science companion). Aimed at materials, chemistry and formulations. Limitation: described only in abstract terms — no demos, no benchmarks; "coming 2026." [Evidence — thin] 9

4. CATIA Sketch Generative Constraint. Automatically applies dimensional constraints to sketches using a graph neural network. Limitation: no accuracy data; the technical detail lives in a community blog post. [Evidence] 15

5. CATIA Generative Design. Topology optimization from a functional specification. Limitation: this is mature optimization technology wearing a generative-AI label27; the "push of a button" ease is unverified. [Evidence + Marketing] 13

6. CATIA R2026x FD03 "Generative Experiences" (July 28, 2026). Multimodal input — text, sketches, scans — generating parts, assemblies, rules and requirements. Limitation: freshly released; no independent validation yet. [Evidence — fresh] 14

7. SIMULIA surrogate simulation (MODSIM). A neural network trained on a subset of simulations predicts the rest of the design space. The one concrete number found: a quarter-million wing variants predicted at 91% stress accuracy — from a blog post in April 2021, about general structural analysis. Not injection moulding. [Evidence] 16

8. SOLIDWORKS 2026 generative drawings / auto-assembly + forum companion — generally available, late 2025. [Synthesis]

9. DFM = SOLIDWORKS DFMXpress. Deterministic rules-based manufacturability checks, including injection-moulding rules, shipping since 2008. Explicitly not AI. [Synthesis]

A product called "outcome-based simulation" was not found — the substance behind that phrase is the surrogate-simulation work above. Do not attribute the phrase to Dassault.

Engineering Workflow Contribution

Dassault can, in principle, carry an automotive moulded part end to end on one platform:

requirements in ENOVIA → surfacing and concept in CATIA → detailed CAD in CATIA/SOLIDWORKS → structural simulation in SIMULIA → mould-flow in SOLIDWORKS Plastics or SIMULIA Plastic Injection → tool design in CATIA Mold & Tooling → manufacturing engineering in DELMIA → release back through ENOVIA.

The companion layer (Aura, Leo, Marie) is being positioned to orchestrate across all of this; all three companions shipped only in mid-2026 (July), so they have little track record yet. As of August 2026, the AI with any real service history touches design authoring and general simulation — not the mould-flow or DFM stages. [Synthesis]

Public Customer Evidence

Case Studies

  • BMW Group — 3DEXPERIENCE as its "future engineering platform," ~17,000 users, vehicle virtual twin (February 2024). Solid deployment evidence, but no quantified outcome and nothing plastics-specific. [Evidence] 20
  • Renault Group — 10,000 users in 2016, growing past 20,000 with the "Renaulution" cloud move in 2021. Separately, a NETVIBES AI project optimized vehicle costs (February 2023) — but that is purchasing analytics, not moulded parts. [Evidence] 2122
  • Jaguar Land Rover — 3DEXPERIENCE across all vehicle programs worldwide, 18,000+ users (May 2024). The "save time, reduce cost" claims are unquantified. [Evidence + Marketing] 23
  • Sundaram Auto Components (TVS group) — an automotive injection-moulding supplier. Converting Class-A surfaces into manufacturing-ready solid models reportedly fell from 5–6 hours to 15–20 minutes per component using CATIA V5. This is the strongest quantified moulded-plastics number found — but note two things: it comes from a reseller (Altem), not from Dassault itself, and it is a CAD-productivity win, not AI. [Evidence — partner-reported] 24
  • TactoTek — injection-moulded structural electronics (IMSE) developed on 3DEXPERIENCE, used in automotive control panels (January 2018). A partnership announcement; no quantified outcome; now dated. [Evidence] 25

Conference Demonstrations

3DEXPERIENCE World 2025 (the Aura unveiling, February 2025) and the Virtual Companions demos (February 2026 — staged). [Evidence] 9

White Papers / Testimonials

Older marketing claims — "23 of 30 OEMs," "75% of new vehicles designed in CATIA" — trace back to a 2008 press release and must not be presented as current. [Synthesis — dated]

The gap. A public automotive story combining Dassault AI + moulded plastic parts + a quantified outcome does not appear to exist as of August 2026. [Evidence of absence]

Technical Architecture (Inferred)

What the sources establish

  • OUTSCALE — Dassault's majority-owned sovereign cloud (SecNumCloud-certified), described as "AI factories across three continents." [Evidence] 7
  • Mistral AI — supplies the foundation language models, as a service on OUTSCALE (July 2024; deepened November 2025 with Le Chat Enterprise and AI Studio). [Evidence] 67
  • NVIDIA — "Industry World Models," Omniverse DSX and Nemotron models (February 2026). [Evidence] 10
  • Knowledge graph — traces to the Proxem acquisition (2020), feeding the NETVIBES/EXALEAD "reusable industry knowledge graph." [Synthesis]

Putting it together

The stack reads, bottom to top: OUTSCALE (sovereign cloud) → Mistral language models plus NVIDIA world models → 3DEXPERIENCE (platform and virtual twins) → the Virtual Companions on top. [Synthesis]

Reading between the lines

The language-model layer looks bolted onto a semantic-search and PLM-data foundation that Dassault built through the 2010s (EXALEAD, NETVIBES, Proxem) — not a ground-up rebuild. How exactly the companions are grounded in a customer's engineering data is not publicly documented. One vocabulary note: "virtual twin" is Dassault's own term for what the industry calls a digital twin. [Inference]

AI Technologies

In one list: large language models (from Mistral — bought in, not owned); agentic AI (the Virtual Companions); generative design (topology optimization, the GNN sketch-constraint feature, multimodal generation); simulation AI (neural-network surrogates trained on Abaqus runs); virtual twins and "Industry World Models" (with NVIDIA); and a knowledge graph (EXALEAD/Proxem). [Evidence + Synthesis]

Research Publications

Patents

  • US11922573B2 (granted 2024) — a neural network infers solid CAD features from freehand drawings. [Evidence] 26
  • A method filed November 2024 — machine-learning reverse-engineering of parametric CAD from plain geometry. [Evidence]
  • A SIMULIA patent family on surrogate / hybrid twins — simulate, generate virtual data, train ML to predict behaviour from virtual data plus sensor readings. [Evidence]

Notes

The patents line up with the three internal AI streams from §3 (geometry↔CAD ML, surrogate simulation, semantic NLP). One contrast worth noting: the open-source CAD-ML building blocks of the wider field — UV-Net, AAGNet — come from Autodesk and academia — not Dassault. Dassault's comparable work is patented and closed. [Evidence + Synthesis]

Open Source

Near-nil for AI — reported plainly. [Evidence of absence] 19

  • The DassaultSystemes-Technology GitHub organisation has 4 repositories, all 3D-graphics related (a PBR shading model, a path-tracer, a glTF fork, the org site). No AI or ML.
  • There is no official Hugging Face organisation (the obvious URL returns 404), and no published models or datasets.
  • This is consistent with the strategy: AI is monetized as a proprietary sovereign-cloud service, and the language models come from Mistral rather than being open-released.

Engineering Service / Platform Mapping

CATIA (design/systems) · SOLIDWORKS (mid-market CAD + Plastics) · SIMULIA (simulation + Plastic Injection INK/IME) · DELMIA (manufacturing) · ENOVIA (PLM) · NETVIBES/EXALEAD (data intelligence) · OUTSCALE (cloud) — all federated on 3DEXPERIENCE. [Evidence]

Engineering Intelligence Stack Mapping

How Dassault maps onto the study's five-layer intelligence stack:

  1. Intent — requirements and architecture apps, plus Aura as orchestrator; the "generative economy" framing is precisely about capturing intent as knowledge. [Evidence + Marketing]
  2. Knowledge — the EXALEAD/Proxem knowledge graph over 40+ years of PLM data. Strong in principle, but the mechanics of how it grounds the AI are undisclosed. [Evidence + Inference]
  3. Reasoning — surrogate simulation and generative design exist for structural work and CAD; absent for mould-flow. [Evidence]
  4. Execution — CATIA/SOLIDWORKS/DELMIA authoring and manufacturing; Leo is intended to execute design and simulation tasks, but is still arriving. [Evidence — partial]
  5. Feedback — SIMULIA validation and virtual twins exist, but there is no evidence of a data-driven learning loop from real manufacturing outcomes (defects, warpage, scrap). [Evidence of absence]

Strengths

  • Owns a native injection-moulding solver (SIMPOE-derived) — rare among platform vendors. [Evidence] 5
  • End-to-end coverage on one platform, requirements through production. [Synthesis]
  • Deep, mature moulded-parts authoring: CATIA Functional Molded Parts and Mold & Tooling; SOLIDWORKS Plastics. [Evidence] 1718
  • A coherent, well-funded AI strategy with real differentiators: sovereignty, physics-grounding, proprietary data. [Evidence + Inference]

Weaknesses

  • The AI has not reached the moulding domain — no ML for fill or warp prediction, no AI-based DFM. [Evidence] 12
  • The companion story is very young — Leo and Marie, first shown in staged demos, shipped only in mid-2026 and have little track record. [Evidence] 911
  • Closed and cloud-locked — OUTSCALE plus Mistral, no open models, limited multi-CAD or on-premises openness. [Synthesis]
  • The core LLM capability is bought from Mistral, not owned — a dependency on Mistral and NVIDIA. [Evidence] 6

Current Gaps (largely manual today)

Interpreting mould-flow results and making DFM decisions remain human, rules-based work. Specifically: there is no AI surrogate for injection fill/warp, no data-driven prediction of defects, warpage or scrap, and no shipping product that lets a company search its own historical part library by geometry. [Evidence + Inference]

Future Direction

  • "3D UNIV+RSES" — a three-year R&D program; PLM becomes IPLM ("IP generation and management"); marketed as the "7th generation of world representation." [Evidence] 8
  • Value-based pricing for Industrial AI — 2026 is the "foundation year," with growth guidance deliberately lowered to fund it. [Evidence] 1
  • "Industry World Models" with NVIDIA as the science-grounded answer to generic language models. [Evidence] 10
  • My read of the roadmap: sovereign, physics-validated industrial AI; agentic companions "co-engineering" alongside humans; and ever-deeper lock-in on decades of proprietary customer data. [Inference]

Relevance to Automotive Moulded Parts

Capability Strength Notes
Plastic Part Design Strong CATIA Functional Molded Parts; SOLIDWORKS. "Behavior-driven" = rule-based know-how, not AI.
Surface Design (Class A) Strong CATIA surfacing (Sundaram case: 5–6 h → 15–20 min).
CAD Automation Strong, AI emerging Generative design, GNN sketch constraints, R2026x generative experiences.
Mould Flow Strongest — native solver SOLIDWORKS Plastics + SIMULIA INK/IME: weld lines, air traps, sink marks, clamp tonnage, over-moulding, two-shot. SIMPOE-derived.
Tool Design Strong CATIA Mold & Tooling: pull direction, parting line, undercuts, cooling and ejector layout.
DFM Partial — rules only DFMXpress; the gate-location adviser is a heuristic, not ML.
Manufacturing Engineering / Quality Strong DELMIA; SIMULIA validation.
Engineering Knowledge Reuse Emerging EXALEAD/NETVIBES knowledge graph; Aura.

Critical finding. There is no evidence of AI or ML applied specifically to injection-moulding simulation or plastic-part design. The moulding stack — the SIMPOE-derived solvers plus CATIA's moulding apps — is classical numerical simulation plus built-in engineering rules. Dassault's real ML lives in general structural/CFD simulation and CAD authoring. Tellingly, the SIMULIA AI/ML page lists aerospace, transportation and life-sciences examples — plastics and injection moulding are absent. [Evidence — strong negative] 12

Net. Dassault has the deepest native moulded-parts stack of any platform vendor — it owns the solver outright — but its AI layer has not yet reached the mould-flow/DFM domain. [Synthesis]

Key Takeaways

  1. Dassault is one of the few platform vendors that own an injection-moulding solver outright (SIMPOE, 2013) — and the only one whose solver is native to its own CAD platform. [Evidence] 5
  2. It offers genuine end-to-end moulded-parts coverage on one platform. [Synthesis]
  3. Its AI strategy is coherent and well-funded — sovereign and physics-grounded — but the companions (Leo and Marie) shipped only in mid-2026, with little track record so far. [Evidence] 911
  4. No AI reaches the moulding domain yet — the single most important point for this study. [Evidence] 12
  5. Its DFM is rules-based, not AI (DFMXpress). [Evidence]
  6. Its LLM capability is bought from Mistral, not owned. [Evidence] 6
  7. Its open-source AI footprint is near-nil — everything is proprietary and sovereign-cloud hosted. [Evidence] 19
  8. The strongest quantified moulded-plastics customer number (Sundaram/TVS) is CAD productivity, reported by a reseller — not AI. [Evidence] 24
  9. Clear white space — an open opportunity: AI for mould-flow/DFM, geometry retrieval, and defect prediction. [Inference]

References

Primary Sources — Dassault

  • Q4/FY2024 results, Feb 4 2025 — https://www.3ds.com/newsroom/press-releases/dassault-systemes-strong-q4-results-driven-new-business-acceleration-and-expanded-3dexperience-footprint
  • Q4/FY2025 results + 2026 guidance, Feb 2026 — https://investor.3ds.com/news-releases/news-release-details/dassault-systemes-q4-revenue-growth-1-solid-operating-margin-and/
  • Transportation & Mobility vertical — https://www.3ds.com/industries/transportation-mobility
  • SIMPOE acquisition, May 7 2013 — https://www.3ds.com/newsroom/press-releases/dassault-systemes-acquires-simpoe
  • Mistral partnership, Jul 1 2024 — https://www.3ds.com/newsroom/press-releases/dassault-systemes-and-mistral-ai-partner-offer-trusted-ai-powered-industry-grade-solutions-accelerate-generative-economy
  • Mistral deepened, Nov 26 2025 — https://www.3ds.com/newsroom/press-releases/new-era-sovereign-ai-dassault-systemes-and-mistral-ai-deepen-their-partnership
  • 3D UNIV+RSES, Feb 4 2025 — https://www.3ds.com/newsroom/press-releases/dassault-systemes-reveals-3d-universes-and-related-ai-based-services
  • Virtual Companions unveiled, Feb 11 2026 — https://www.3ds.com/newsroom/press-releases/dassault-systemes-unveils-new-way-working-industry-ai-powered-virtual-companions
  • NVIDIA partnership, Feb 3 2026 — https://www.3ds.com/newsroom/press-releases/dassault-systemes-and-nvidia-partner-build-industrial-ai-platform-powering-virtual-twins
  • Agentic platform expansion, Jul 23 2026 — https://www.3ds.com/newsroom/press-releases/dassault-systemes-expands-3dexperience-ai-native-agentic-platform-new-virtual-companion-skills-co-engineer-humans
  • SIMULIA AI & ML simulation — https://www.3ds.com/products/simulia/ai-and-machine-learning-simulation
  • CATIA AI-driven generative — https://www.3ds.com/products/catia/ai-driven-generative-experiences
  • CATIA R2026x FD03 Industrial AI, Jul 28 2026 — https://blog.3ds.com/brands/catia/industrial-ai-catia-r2026x-fd03/
  • CATIA Sketch Generative Constraint (GNN) — https://blog.3ds.com/brands/catia/unlocking-the-latest-ai-capabilities-for-engineering-design/
  • SIMULIA + ML, Apr 27 2021 — https://blog.3ds.com/brands/simulia/combining-simulation-machine-learning-product-development/
  • CATIA Mold & Tooling brochure (PDF) — https://www.3ds.com/fileadmin/PRODUCTS-SERVICES/CATIA/PDF/CATIA_Mold-and-Tooling_Brochure.pdf
  • SOLIDWORKS Plastics — https://www.solidworks.com/product/solidworks-plastics
  • GitHub — https://github.com/DassaultSystemes-Technology

Customer Evidence (primary + partner)

  • BMW, Feb 1 2024 — https://www.3ds.com/newsroom/press-releases/bmw-group-partners-dassault-systemes-bring-3dexperience-platform-its-future-engineering-platform
  • Renault, 10,000 users, Sep 29 2016 — https://www.3ds.com/newsroom/press-releases/dassault-systemes-3dexperience-platform-reaches-10000-users-renault
  • Renault NETVIBES AI, Feb 2 2023 — https://www.3ds.com/newsroom/press-releases/dassault-systemes-developed-new-data-science-solution-allow-renault-group-optimize-vehicle-costs
  • JLR, May 14 2024 — https://www.3ds.com/newsroom/press-releases/jlr-and-dassault-systemes-extend-partnership-deploying-3dexperience-platform-all-vehicle-programs-worldwide
  • Sundaram Auto (Altem), Feb 2023 — https://altem.com/class-b-surface-design-creation-and-design-for-manufacturing-using-catia-v5/
  • TactoTek, Jan 3 2018 — https://www.3ds.com/newsroom/press-releases/tactotekr-collaborates-dassault-systemes-bring-injection-molded-structural-electronics-imse-process-3dexperience-platform

Secondary Sources

DEVELOP3D (SOLIDWORKS Aura/agents, Feb 2025 & Feb 2026); engineering.com (Aura/Leo/ Marie); Digital Engineering 24/7; Verdict (patent US11922573B2); patents.justia.com/ assignee/dassault-systemes; Wikipedia (corporate facts); companiesmarketcap.com (market-cap snapshot).

Appendix A — Timeline

See §3. Key AI dates: Mistral (Jul 2024) → 3D UNIV+RSES (Feb 2025) → Aura unveil (Feb 2025) → Aura live (~Jul 2025) → Mistral deepened (Nov 2025) → NVIDIA (Feb 2026) → Virtual Companions (Feb 2026) → agentic-platform expansion (Jul 2026).

Appendix B — Glossary

  • 3DEXPERIENCE — Dassault's unifying cloud platform, on which all its brands run.
  • Solver — the numerical engine inside a simulation tool that computes the physics (e.g. how molten plastic fills a mould).
  • Surrogate simulation — a neural network trained on a batch of real simulations that then predicts results near-instantly for new design variants.
  • Sovereign cloud — cloud hosting guaranteed to stay under one jurisdiction's control; OUTSCALE is Dassault's, certified to France's SecNumCloud standard.
  • IPLM — Dassault's re-branding of PLM as "IP generation & management."
  • INK / IME — the SIMULIA Plastic Injection roles (design-side vs. advanced tooling). The codes are reseller-sourced; treat as approximate.
  • Virtual twin — Dassault's term for a digital twin.
  • Virtual Companions — the Aura / Leo / Marie AI agents.
  • DFM — design for manufacturability: checking a design against the rules that make it mouldable (draft angles, wall thickness, undercuts…).

Appendix C — Notes (thinnest-evidence areas to revisit)

  1. The claim that Mistral models power the companions is trade-press only; the companion press releases do not say it — state as reported-but-unconfirmed.
  2. Marie's and Aura's concrete inputs and outputs are abstract marketing (no demos).
  3. The INK/IME product codes and introduction years are reseller-sourced.
  4. Transportation & Mobility segment revenue is not public.
  5. Any AI applied to injection moulding specifically — a strong negative; the report's central honest point.
  6. RESOLVED 2026-08-02: FY2025 exact decimals and governance confirmed against the primary Feb 11, 2026 results release (€6,235.8 M IFRS / €6,239.6 M non-IFRS; non-IFRS margin 32.0%) and the Feb 21, 2026 governance release (Daloz Chairman & CEO; Charlès off the Board). §1 updated.

Executive Summary

  • Who they are. Dassault Systèmes is Europe's flagship engineering-software company. Its major products — CATIA, SOLIDWORKS, SIMULIA, DELMIA and ENOVIA — all run on one shared platform called 3DEXPERIENCE. Revenue in 2025 was €6.24 billion. [Evidence] 1
  • Why they matter for moulded parts. Dassault covers the whole moulded-parts chain — design, simulation, tooling, manufacturing — on its own platform. Most importantly, it is one of the few platform vendors owning an injection-moulding simulation engine (a "solver") outright — acquired with the company SIMPOE in 2013 — and the only one whose solver is native to its own CAD platform. That solver now ships as SOLIDWORKS Plastics and as SIMULIA's Plastic Injection roles. Most competitors have to partner with a third party for this. [Synthesis] 5
  • AI maturity: big ambitions, uneven delivery. The strategy is coherent — European "sovereign" AI (hosted on its own OUTSCALE cloud, using Mistral's language models, with NVIDIA for industrial world models) — and one AI assistant (Aura) has been live since 2025. The fuller agent line-up (Leo, Marie) shipped only in mid-2026, so it has little track record yet. [Evidence] 911610
  • The central honest finding. After a deliberate search, no public evidence was found of AI or machine learning applied specifically to injection-moulding simulation or plastic-part design. Dassault's real AI lives in general CAD authoring and general structural simulation. Its moulding tools are classical physics simulation plus rule-based engineering know-how. For this study, that gap is the most important thing to know about Dassault. [Evidence] 12

  1. Dassault Systèmes Q4/FY2025 results & 2026 guidance, Feb 11 2026 — https://investor.3ds.com/news-releases/news-release-details/dassault-systemes-q4-revenue-growth-1-solid-operating-margin-and/; full financial tables via GlobeNewswire — https://www.globenewswire.com/news-release/2026/02/11/3235907/0/en/Dassault-Syst%C3%A8mes-Q4-revenue-growth-of-1-with-solid-operating-margin-and-EPS-expansion-Initiating-2026-revenue-guidance-3-5-growth.html ↩↩↩↩↩

  2. "Dassault Systèmes' CEO Pascal Daloz becomes also Chairman of the Board of Directors," Feb 21 2026 — https://www.3ds.com/newsroom/press-releases/dassault-systemes-ceo-pascal-daloz-becomes-also-chairman-board-directors-dassault-systemes ↩

  3. Dassault Systèmes Q4/FY2024 results, Feb 4 2025 — https://www.3ds.com/newsroom/press-releases/dassault-systemes-strong-q4-results-driven-new-business-acceleration-and-expanded-3dexperience-footprint ↩

  4. Transportation & Mobility industry page — https://www.3ds.com/industries/transportation-mobility ↩

  5. SIMPOE acquisition press release, May 7 2013 — https://www.3ds.com/newsroom/press-releases/dassault-systemes-acquires-simpoe ↩↩↩↩

  6. Dassault × Mistral AI partnership, Jul 1 2024 — https://www.3ds.com/newsroom/press-releases/dassault-systemes-and-mistral-ai-partner-offer-trusted-ai-powered-industry-grade-solutions-accelerate-generative-economy ↩↩↩↩↩↩

  7. Mistral partnership deepened, Nov 26 2025 — https://www.3ds.com/newsroom/press-releases/new-era-sovereign-ai-dassault-systemes-and-mistral-ai-deepen-their-partnership ↩↩↩

  8. "3D UNIV+RSES" reveal, Feb 4 2025 — https://www.3ds.com/newsroom/press-releases/dassault-systemes-reveals-3d-universes-and-related-ai-based-services ↩↩↩

  9. Virtual Companions (Aura/Leo/Marie) unveiled, Feb 11 2026 — https://www.3ds.com/newsroom/press-releases/dassault-systemes-unveils-new-way-working-industry-ai-powered-virtual-companions ↩↩↩↩↩↩↩↩↩↩

  10. Dassault × NVIDIA partnership, Feb 3 2026 — https://www.3ds.com/newsroom/press-releases/dassault-systemes-and-nvidia-partner-build-industrial-ai-platform-powering-virtual-twins ↩↩↩↩↩

  11. AI-native agentic platform expansion, Jul 23 2026 — https://www.3ds.com/newsroom/press-releases/dassault-systemes-expands-3dexperience-ai-native-agentic-platform-new-virtual-companion-skills-co-engineer-humans ↩↩↩↩

  12. SIMULIA "AI & Machine Learning Simulation" page (note: no plastics/injection-moulding examples) — https://www.3ds.com/products/simulia/ai-and-machine-learning-simulation ↩↩↩↩↩

  13. CATIA AI-driven generative experiences — https://www.3ds.com/products/catia/ai-driven-generative-experiences ↩

  14. CATIA R2026x FD03 "Industrial AI" blog, Jul 28 2026 — https://blog.3ds.com/brands/catia/industrial-ai-catia-r2026x-fd03/ ↩↩

  15. CATIA Sketch Generative Constraint (GNN) blog — https://blog.3ds.com/brands/catia/unlocking-the-latest-ai-capabilities-for-engineering-design/ ↩↩

  16. "Combining Simulation & Machine Learning" blog, Apr 27 2021 — https://blog.3ds.com/brands/simulia/combining-simulation-machine-learning-product-development/ ↩

  17. CATIA Mold & Tooling brochure (PDF) — https://www.3ds.com/fileadmin/PRODUCTS-SERVICES/CATIA/PDF/CATIA_Mold-and-Tooling_Brochure.pdf ↩↩

  18. SOLIDWORKS Plastics product page — https://www.solidworks.com/product/solidworks-plastics ↩↩

  19. DassaultSystemes-Technology on GitHub — https://github.com/DassaultSystemes-Technology ↩↩

  20. BMW Group × Dassault press release, Feb 1 2024 — https://www.3ds.com/newsroom/press-releases/bmw-group-partners-dassault-systemes-bring-3dexperience-platform-its-future-engineering-platform ↩↩

  21. Renault reaches 10,000 3DEXPERIENCE users, Sep 29 2016 — https://www.3ds.com/newsroom/press-releases/dassault-systemes-3dexperience-platform-reaches-10000-users-renault ↩↩

  22. Renault NETVIBES data-science (cost optimization), Feb 2 2023 — https://www.3ds.com/newsroom/press-releases/dassault-systemes-developed-new-data-science-solution-allow-renault-group-optimize-vehicle-costs ↩

  23. JLR deploys 3DEXPERIENCE across all vehicle programs, May 14 2024 — https://www.3ds.com/newsroom/press-releases/jlr-and-dassault-systemes-extend-partnership-deploying-3dexperience-platform-all-vehicle-programs-worldwide ↩↩

  24. Sundaram Auto Components case study (Altem, reseller-reported), Feb 2023 — https://altem.com/class-b-surface-design-creation-and-design-for-manufacturing-using-catia-v5/ (Sundaram Auto Components / TVS group — Chennai, India) ↩↩

  25. TactoTek IMSE on 3DEXPERIENCE, Jan 3 2018 — https://www.3ds.com/newsroom/press-releases/tactotekr-collaborates-dassault-systemes-bring-injection-molded-structural-electronics-imse-process-3dexperience-platform ↩

  26. US11922573B2 (via Verdict / patents.justia.com) — https://patents.justia.com/assignee/dassault-systemes ↩

  27. What "optimization technology" means and why it is not generative AI — see the concept note What is Optimization technology: topology optimization computes shapes from physics equations (finite-element analysis in an iterative material-removal loop); it learns nothing from data, so it is classical optimization, not AI. ↩