Dassault Systèmes SA and Nvidia Corp. today announced a long-term strategic partnership aimed at creating a shared industrial artificial intelligence platform that combines science-based virtual twins with large-scale AI infrastructure to support applications across multiple industries.

The companies said the partnership is designed to move industrial AI beyond point products by grounding AI systems in physics, engineering constraints and validated industrial knowledge. The result, they said, will be “industry world models” that can be used as systems of record for designing, simulating and operating complex products and production systems.

At the core of the collaboration is the integration of Dassault Systèmes’ virtual twin technologies with Nvidia’s accelerated computing platforms, AI models and software libraries. Virtual twins are digital models that mirror real-world counterparts, using real-time data for simulations, predictions, and performance optimization.

The combined architecture is intended to support applications spanning biology, materials science, engineering and manufacturing, while also enabling a new class of AI-driven “virtual companions” embedded in Dassault Systèmes’ 3DExperience platform, used to create virtual twins.

The announcement was made at 3DExperience World in Houston, Dassault Systèmes’ annual user event for design and engineering professionals.

The partnership builds on nearly 25 years of collaboration between the companies, beginning with Dassault’s Catia design software on Nvidia graphics processors and later expanding into graphics processing unit-accelerated physics simulation using Nvidia’s Cuda and RTX technologies.

The new agreement reflects what both companies describe as a shift toward “physical AI,” where AI systems are designed to understand and reason about the physical world rather than operate solely on text or images.

AI factories

Dassault Systèmes said it will deploy “AI factories” through its Outscale cloud brand, using Nvidia infrastructure on three continents. They allow customers to train and operate AI models while maintaining data sovereignty, intellectual property protection and regulatory compliance.

Nvidia, for its part, is adopting Dassault’s model-based systems engineering tools to design its own AI factories, starting with the flagship Nvidia Rubin platform. The same approach will be used in Nvidia’s Omniverse DSX Blueprint for large-scale AI factory deployment.

The companies outlined several industry-specific use cases enabled by the platform. In life sciences, Nvidia’s BioNeMo platform will be combined with Dassault’s Biovia scientific models for drug discovery and materials research. In engineering, Dassault’s Simulia-based virtual twins will use Nvidia Cuda-X libraries to provide near real-time performance predictions for designers who are not simulation specialists.

In manufacturing, Nvidia Omniverse physical AI libraries will be integrated with Dassault’s Delmia virtual twin of production systems to support autonomous, software-defined factories.

A central element of the partnership are AI-powered “virtual companions” that act as domain-specific assistants for business users, engineers and scientists. The agents are designed to understand user intent, reason using industry world models and orchestrate actions across workflows.

Executives from both companies emphasized that the partnership is intended to define a long-term framework for industrial AI rather than a single product launch. By combining validated virtual representations of the physical world with scalable AI infrastructure, they argue, the platform could make advanced simulation and AI-driven decision-making accessible to a much broader set of professionals.

Image: SiliconANGLE/Microsoft Designer