Mistral and Nvidia are working together on AI-native engineering workflows that run on Nvidia accelerated computing, Mistral said in a blog post on October 21. The aim is faster engineering simulation and closed-loop workflows that connect simulation, AI model development and engineering design.
The technical stack pairs Mistral Large 4 and Mistral's data generation and physics modeling pipelines with Nvidia's CUDA-X libraries and Nvidia Warp for acceleration. Nvidia Omniverse libraries cover GPU-accelerated 3D scene generation and SimReady asset creation.
Mistral and Nvidia also plan public benchmarks and open datasets for evaluating physics models, giving researchers and industry teams a shared basis for comparing accuracy and performance. The companies will build tools for creating and validating automated CAD-to-simulation workflows.
Key facts
- 01Mistral and Nvidia are building AI-native engineering workflows that run on Nvidia accelerated computing.
- 02The stack pairs Mistral Large 4 and Mistral's physics pipelines with Nvidia CUDA-X, Warp and Omniverse libraries.
- 03The companies plan public benchmarks and open datasets for physics models, plus agent benchmarks for CAE and EDA.
- 04Mistral Large 4 has about 1 trillion total parameters, 49 billion active, and was trained on 3,800 Grace Blackwell GPUs.
- 05Mistral's compute roadmap targets 1 GW of capacity by 2030.
A separate set of benchmarks targets agentic AI in computer-aided engineering (CAE) and electronic design automation (EDA). Those tests will measure whether agents can use engineering tools, complete multistep workflows and produce technically correct results, while tracking execution time and compute cost.
Mistral named aerospace, automotive and semiconductors as the target sectors. The post gives no release dates for the benchmarks or datasets.
Neil Ashton, Mistral's VP of Computational Engineering, tied the project to faster simulation cycles and to open benchmarks that validate agents on real CAE and EDA work.
“For too long, the pace of engineering has been set by the pace of simulation.”— Neil Ashton, VP of Computational Engineering, Mistral
Tim Costa, Nvidia's VP and GM of industrial and computational engineering, said the next wave of engineering breakthroughs will come from exploring ideas that were once too complex or time-consuming to test. He pointed to the combination of Mistral's open models and physics AI expertise with Nvidia's accelerated computing and shared industry benchmarks.
The two companies have history. They co-developed Mistral NeMo and are working through the Nvidia Nemotron Coalition on a frontier base model for open-weight development.
Mistral Large 4, nicknamed "Le Chonk," is in public preview. The multimodal model has about 1 trillion total parameters with 49 billion active, and Mistral trained it on 3,800 Nvidia Grace Blackwell GPUs in Mistral-operated data centers in Europe. Mistral said the weights are planned for release at the end of October.
Mistral is also bringing more Nvidia-powered infrastructure online for training, customization and inference. Its compute roadmap targets 1 GW of capacity by 2030.
Working on something we should cover, or seeing a story we missed? Send leads, documents, or feedback to hello@aichatdaily.com. For sensitive tips, see our secure tips page for Signal and PGP options.
Spotted an error? Email hello@aichatdaily.com with the URL and the issue, or read our full corrections policy.
