Dassault Systèmes and Nvidia announce a long-term partnership to build an industrial AI platform designed to strengthen digital twins and develop “Industry World Models.”
Their joint vision is to make AI a core component of engineering, manufacturing, and research—well beyond mere proofs of concept. It extends a collaboration of more than 25 years between the two groups, which began around the 3D modeling software CATIA on GPUs and gradually broadened to accelerated physical simulation.
The stated ambition is to define a shared industrial architecture for mission-critical AI, anchored in physics, technical constraints, and industrial know-how rather than general-purpose data.
A technological foundation that combines Virtual Twin and Omniverse
Dassault Systèmes contributes its 3DEXPERIENCE platform and its Virtual Twin technologies, spanning design with CATIA, manufacturing with DELMIA, and systems engineering. Nvidia brings its AI infrastructure—including GPUs, CUDA and RTX, its open Nemotron models, its accelerated software libraries, and its Omniverse platform dedicated to physical simulation and 3D collaboration.
The two companies discuss the concept of “physical AI,” an artificial intelligence capable of understanding and reasoning about the physical world by drawing on scientifically validated models and domain constraints. Omniverse’s physical AI libraries will be integrated into the Delmia digital twins to enable autonomous, software-defined production systems.
Industry World Models and Virtual Assistants
Industry World Models, sector-specific reference models that fuse digital twins, operational data, and AI models, are intended to serve as the foundation for the design, simulation, and control of systems across various sectors: aerospace, automotive, life sciences, robotics, or materials.
On the 3DEXPERIENCE platform, these Industry World Models will feed Virtual Companions, AI agents embedded in industry tools and capable of providing contextual recommendations. Based on Nvidia’s Nemotron models and Dassault’s domain models, these assistants are designed to help engineers, researchers, and operators explore scenarios, optimize designs, or adjust production parameters in real time.
AI Factories on Three Continents
The partnership includes an infrastructure component with the deployment of AI factories on three continents via Outscale, Dassault Systèmes’ cloud. These centers will be equipped with Nvidia technologies to train and run the AI models used by the digital twins, while meeting data sovereignty, intellectual property protection, and regulatory compliance requirements.
For its part, Nvidia will use Dassault’s modeling and systems engineering tools to design its own AI factories, including those based on the forthcoming Rubin platform, drawing on the Omniverse DSX Blueprint architecture. This reciprocity illustrates an approach in which each side applies the other’s models and tools to its own infrastructure.
Several companies are already presented as early adopters of this convergence between Virtual Twin and accelerated AI: Lucid Motors, Bel, the OMRON Group, and the National Institute for Aviation Research. In the automotive sector, the objective is to accelerate the move from concept to production while improving the predictive accuracy of vehicle and drivetrain simulations.
