Dassault Systèmes and Nvidia build industrial AI platform

Dassault Systèmes and Nvidia announced a long-term strategic partnership on February 3, 2026. The goal is a joint industrial architecture for "mission-critical artificial intelligence" - AI that not only delivers pretty dashboards but serves as a central building block for the development, simulation, and operation of complex systems.

At its core, it is about connecting two worlds: Dassault Systèmes' virtual twin technologies and Nvidia's AI infrastructure, open models, and accelerated software libraries. From this, "science-validated industry world models" are to be created - validated world models that position industrial AI as a system of record and not as a point solution.

Dassault Systèmes sets the bar high: instead of AI that only generates or predicts, AI should "understand the real world" - with a scientific and physical foundation. Pascal Daloz, CEO of Dassault Systèmes, sums it up in a key passage: "We are entering an era where artificial intelligence not only predicts or generates but understands the real world. When AI is anchored in science, physics, and validated industrial knowledge, it becomes an amplifier of human creativity. Together with Nvidia, we are building industry world models that combine virtual twins and accelerated computing to help industry design, simulate, and operate complex systems in biology, materials science, engineering, and manufacturing with confidence. This partnership creates a new foundation for industrial AI - one that is trustworthy from the ground up and can scale innovation in the generative economy."

Nvidia also strategically classifies the cooperation - and positions "physical AI" as the next development step. Jensen Huang, founder and CEO of Nvidia, puts it this way: "Physical AI is the next frontier of artificial intelligence - anchored in the laws of the physical world. Together with Dassault Systèmes, we combine decades of industrial leadership with Nvidia's AI and Omniverse platforms to transform how millions of researchers, designers, and engineers build the world's largest industries."

The subtext is clear: anyone who wants to accelerate industrial value creation with AI in the future must seamlessly integrate computing power, simulation models, data spaces, and engineering workflows. This is exactly where both partners come in.

Platform instead of point solution: why the industry is currently grappling with architecture

Industrial companies know the pattern: AI use cases start quickly - and often end in a zoo of individual applications, proprietary data pipelines, and hard-to-maintain models. What is missing is a robust architecture that can be scaled, validated, and integrated into existing engineering processes.

Dassault Systèmes and Nvidia address this gap with a platform approach that explicitly aims to be "deployable at scale." Particularly interesting is the focus on "science-validated world models" - models that not only work statistically but are intended to be linked with physics, materials science, and validated industrial knowledge.

This is more than word cosmetics: In manufacturing and mechanical engineering, it's not the most creative AI response that matters, but the demonstrably correct one. When virtual twins are increasingly supplemented by AI in the future, robust mechanisms for validation, traceability, and compliance are needed - especially in regulated industries.

Another component: so-called "skilled virtual companions" on the agent-based 3DExperience platform. These are AI assistants that do not chat in a vacuum but are embedded in the context of product data, simulation models, and manufacturing logic.

Outscale, AI factories, and Omniverse: Infrastructure becomes a power factor

Technically, it becomes exciting where both companies understand the platform not just as a software collaboration, but as an infrastructure project.

Dassault Systèmes, with its cloud brand Outscale, is focusing on building AI factories - embedded in a "sustainable and sovereign cloud strategy." These Outscale AI factories are to utilize the latest Nvidia AI infrastructure on three continents. At the same time, Dassault Systèmes promises classic industrial criteria that often appear too late in AI debates: data protection, protection of intellectual property, and sovereignty of customer data.

In parallel, Nvidia is adopting engineering methodology from Dassault Systèmes in reverse: Nvidia plans to use model-based systems engineering (MBSE) to design AI factories - starting with the Nvidia Rubin platform and integrated into the Nvidia Omniverse DSX blueprint for large-scale AI factory rollouts.

This is a remarkable role reversal: the AI infrastructure specialist is relying on systems engineering discipline - and Dassault Systèmes is linking its virtual twin world more closely to GPU-accelerated AI stacks.

Where the partnership specifically begins: from molecules to production lines

The announced architecture is intended not only to impact traditional mechanical engineering but explicitly to open up new possibilities in multiple domains - from the laboratory to the production line.

Biology and materials research are to be accelerated through the combination of Nvidia BioNeMo and Biovia models. The goal is to identify new molecules and next-generation materials more quickly.

Design and engineering are to benefit from AI-supported simulations: Simulia-based virtual twin physics, accelerated by Nvidia Cuda-X libraries and AI physics libraries, are to deliver results faster and more precisely - a clear lever for development times and variant management.

In manufacturing, the integration of Nvidia Omniverse Physical-AI libraries into Delmia models aims at autonomous, software-defined production systems. In short, the digital twin should not only document but increasingly control.

Finally, virtual companions as AI assistants on the agent-based 3DExperience platform are intended to bridge the gap between model, data space, and operational decision-making - with "trusted, actionable intelligence" instead of generic text modules.

User voices: Food, automation, automotive, aviation

Such platform approaches only gain substance when real industry workflows benefit, as shown by statements from different industries.

From the consumer goods and food industry comes a clear scaling argument. Cécile Béliot, CEO of the Bel Group, emphasizes computing power and sustainability goals: "Bel Group is shaping a sustainable food future - through responsible recipes and packaging. Through the collaboration of Nvidia and Dassault Systèmes, we gain the computing power to model and optimize our products on a large scale - accelerating innovation while fulfilling our sustainability commitments."

Omron addresses the increasing complexity of modern production and the transition to more autonomous systems. Motohiro Yamanishi, President of Industrial Automation at Omron, puts it clearly: "To manage the growing complexity of modern manufacturing, the industry must evolve towards fully autonomous and digitally validated production systems. By combining Nvidia's physical AI frameworks with Dassault Systèmes’ virtual twin factory and Omron's automation technologies, manufacturers can move from design to implementation with more confidence and speed."

Industrial Metaverse Conference 2026 - Where the industry of tomorrow is created

At the "Industrial Metaverse Conference" on February 10 and 11, 2026, in Munich, decision-makers, technologists, and thought leaders of the industry will meet to showcase tangible paths into the industrial metaverse - from digital twins and physical AI to collaborative value creation. Get to know the minds and concepts shaping this technological evolution.

Stefan Weinzierl

Lucid, on the other hand, focuses on speed, iteration, and multiphysics simulation - precisely the factors that determine competitiveness in automotive development. Vivek Attaluri, vice president of vehicle engineering at Lucid, says: "Lucid's award-winning engineering and technology continue to set new standards in the automotive industry, and Dassault Systèmes remains a central partner that enables us to stay at the forefront of vehicle and powertrain development. Agility, speed of innovation, and rapid iteration are at the core of our work methods, and our exploration of multiphysics digital twin simulation models - powered by Nvidia's open-source, physics-informed AI models - has the potential to help our teams move from concept to production faster than ever before, without compromising predictive accuracy. We look forward to further collaboration and to using these new tools to support Lucid's future innovations."

And aviation research also addresses a typical industry pain point: compliance, certification, and data sovereignty. Shawn Ehrstein, director of emerging technologies and CAD/CAM at NIAR (Wichita State University), highlights the benefits of virtual companions: "NIAR empowers the next generation of aviation. From digitizing assets to design and manufacturing development as well as their validation, virtual twin technology brings unprecedented capabilities and efficiency. Dassault Systèmes’ virtual companions for engineering - based on the agent-based 3DEXPERIENCE platform and utilizing Nvidia Nemotron open models - accelerate the 'by-design' compliant synthesis of aircraft virtual twins. Using the platform to adapt the virtual twin to demonstrate compliance reduces certification effort while preserving information sovereignty."

What this means for the industry

The partnership is a signal: industrial AI is becoming an infrastructure issue. Those who want to use virtual twins not only for visualization but as an operational control and decision-making model will need computing power, validation, data sovereignty, and engineering integration all in one.

For mechanical engineering and manufacturing, this specifically means: the debate is shifting from "Which AI app are we testing?" to "Which architecture can we justify - technically, regulatorily, and economically?" And yes: this is less glamorous than the next chatbot hype. But that's precisely why it might work this time.

With material from Nvidia