Mistral and NVIDIA collaborate to integrate AI models into engineering workflows

The effort builds on earlier work such as Mistral Large 4, a multimodal model with roughly 1 trillion total parameters and 49 billion active parameters, trained on 3,800 NVIDIA Grace Blackwell GPUs in Europe. Weights are slated for release at the end of October. Mistral also plans to add NVIDIA‑powered AI infrastructure targeting 1 GW of capacity by 2030. Quotes from Neil Ashton, VP of Computational Engineering at Mistral, and Tim Costa, vice president and general manager of industrial and computational engineering at NVIDIA, emphasize faster simulation cycles and shared benchmarks to validate AI agents in real engineering workflows.