Dassault Systèmes & NVIDIA: industrial AI platform for virtual twins

By Agustin Giovagnoli / February 7, 2026

A long‑term partnership between Dassault Systèmes and NVIDIA sets out to build a shared, science‑validated stack that companies can use to design, simulate, and operate high‑fidelity Virtual Twins across sectors from aerospace to consumer goods. Positioned as a shift from siloed point tools, the initiative centers on an industrial AI platform for virtual twins that acts as a system of record for mission‑critical design, manufacturing, and compliance workflows [1][2].

What’s being built: platform components and architecture

The collaboration weaves Dassault Systèmes’ 3DEXPERIENCE platform — including DELMIA Virtual Twin Factories and broader Virtual Twin technologies — with NVIDIA’s accelerated computing stack, Omniverse‑based physical AI libraries, Nemotron open models, and supporting AI infrastructure. The goal: a unified industrial stack that scales across engineering, manufacturing, biology, materials science, and consumer goods, enabling reliable, explainable results grounded in validated physics and engineering knowledge [1][2][3].

This 3DEXPERIENCE NVIDIA integration aims to support design, simulation, operation, and optimization at industrial scale, with NVIDIA framing the effort as advancing “physical AI” — AI grounded in the constraints of the real world. The integration moves beyond isolated AI applications to an architecture capable of autonomous, software‑defined production systems that improve efficiency and resource utilization [1][2][4].

industrial AI platform for virtual twins

At the core are science‑validated Industry World Models: high‑fidelity, physics‑ and engineering‑grounded representations of products, factories, and systems. These models form a system of record that underpins explainable industrial AI, replacing fragmented solutions with a consistent foundation for simulation, decision‑support, and continuous optimization. The approach targets compliant‑by‑design engineering, enabling reproducible, certifiable outcomes across domains where reliability and traceability are paramount [1][2][3].

Skilled virtual companions: Aura, Leo, Marie and workflow orchestration

Dassault Systèmes is introducing skilled virtual companions — agentic AI assistants embedded in 3DEXPERIENCE — that can interpret user intent, reason over Industry World Models, and orchestrate complex workflows. Examples include Aura for business, Leo for engineering, and Marie for scientific use cases. These assistants bring domain‑specific support to experts, integrating with Virtual Twin Factories and Omniverse physical AI libraries to accelerate decisions and execution across the product and production lifecycle [1][2][3].

Business use cases and early adopters

• Bel Group plans to model and optimize product formulations and packaging while meeting sustainability objectives using the shared platform and world models [1][2][4].
• The National Institute for Aviation Research (NIAR) is applying the architecture to create certifiable aircraft Virtual Twins while preserving data sovereignty — a significant requirement in regulated aerospace programs [1][2][5].

These cases illustrate how an industrial digital twin platform can unify R&D, engineering, and operations, linking validated models to measurable business outcomes [1][2][4][5].

Benefits for manufacturers and engineering organizations

By grounding AI in engineering‑accurate models and integrating compute, simulation, and workflow orchestration, the platform promises:

• Increased efficiency and resource optimization through autonomous, software‑defined production systems [1][2][4].
• Better explainability and reliability via science‑validated Industry World Models serving as a system of record [1][2][3].
• Support for sustainable product development and compliant‑by‑design engineering, including scenarios like packaging optimization and certification alignment [1][2][4][5].

Together, 3DEXPERIENCE, Virtual Twin Factories, and Omniverse‑enabled capabilities aim to reduce fragmentation and enable scalable, reproducible outcomes across the lifecycle [1][2][3].

Technical and organizational considerations

Organizations evaluating this architecture should consider:

• Data governance and sovereignty: Use cases like NIAR’s underscore the importance of preserving control over sensitive engineering data while building certifiable twins [1][2][5].
• Validation and explainability: Science‑validated Industry World Models are designed to anchor trustworthy, compliant‑by‑design processes across sectors [1][2][3].
• Cross‑functional adoption: The move from point tools to an integrated stack requires alignment across engineering, operations, and IT to fully leverage world models, Omniverse components, and skilled virtual companions AI [1][2][4].

How to evaluate and pilot this class of platform

• Define core objectives tied to operational or regulatory outcomes (e.g., sustainability metrics, certification steps, throughput gains) [1][2][4][5].
• Inventory data sources needed to build or extend Industry World Models for your domain [1][2][3].
• Start with a targeted workflow where agentic assistants can orchestrate tasks across design, simulation, and operations (e.g., formulation or packaging optimization) [1][2][4].
• Establish validation gates aligned to compliance or certification requirements and measure explainability and reproducibility [1][2][3][5].

For additional frameworks to structure your rollout, Explore AI tools and playbooks.

Outlook: what this partnership signals for industrial AI

This partnership signals a pivot toward integrated industrial AI architectures that harness physics‑based simulation, accelerated computing, and domain‑specific agents. As Industry World Models mature, sectors from aerospace to CPG can pursue more autonomous, software‑defined operations with clearer paths to sustainability and certification — all grounded in validated engineering knowledge. NVIDIA characterizes this as progress toward “physical AI,” and the joint roadmap positions Virtual Twins as the operational backbone of digital‑to‑physical transformation [1][2][3][4]. For the official announcement details, see the NVIDIA newsroom release NVIDIA announcement (external) [1].

Sources

[1] Dassault Systèmes and NVIDIA Partner to Build Industrial AI Platform Powering Virtual Twins
http://nvidianews.nvidia.com/news/dassault-systemes-nvidia-industrial-ai

[2] Dassault Systèmes and NVIDIA Partner to Build Industrial AI Platform Powering Virtual Twins
https://www.3ds.com/newsroom/press-releases/dassault-systemes-and-nvidia-partner-build-industrial-ai-platform-powering-virtual-twins

[3] Dassault & NVIDIA: A New Foundation for Industrial AI
https://manufacturingdigital.com/news/dassault-nvidia-a-new-foundation-for-industrial-ai

[4] Nvidia, Dassault Systèmes to Build Industrial AI Platform – AI Business
https://aibusiness.com/industrial-manufacturing/nvidia-dassault-build-industrial-ai-platform

[5] Dassault Systèmes, NVIDIA partner to build industrial AI platform powering virtual twins
https://www.robotics247.com/article/dassault-systemes-nvidia-partner-to-build-industrial-ai-platform-powering-virtual-twins