Physics AI research that’s shaping the industry.

Mistral AI published a blog post titled "Physics AI research that's shaping the industry," describing its focus on foundational Physics AI for industries including aerospace, automotive, semiconductors, and energy.

The post states that the acquisition of Emmi AI highlighted Mistral's commitment to pushing the state of the art in AI research and enterprise solutions for industrial engineering. Emmi's work, now part of Mistral, is described as dedicated to enabling engineers to build next-generation products faster and to secure continuous performance gains in operations at scale for customers.

The post lists published breakthroughs the work rests on. "Going with the Speed of Sound: Pushing Neural Surrogates into Highly-turbulent Transonic Regimes" presents a new dataset of CFD simulations for 3D wings in the transonic regime, comprising volumetric and surface-level fields for around 30,000 samples with unique geometry and inflow conditions. The post says existing aerospace datasets predominantly focus on 2D airfoils, neglecting these critical 3D phenomena.

"Fluid Intelligence: A Forward Look on AI Foundation Models in Computational Fluid Dynamics" bridges the machine learning and CFD communities by deconstructing industrial-scale CFD simulations into core components.

"AB-UPT for Automotive and Aerospace Applications" adds two new datasets to empirically evaluated use-cases of AB-UPT, combining high-quality data generation with state-of-the-art neural surrogates.

"GyroSwin: 5D Surrogates for Gyrokinetic Plasma Turbulence Simulations" addresses understanding plasma turbulence, which the post says significantly impairs plasma confinement and is vital for next-generation reactor design.

Other listed items include AB-UPT, an Anchored-Branched Universal Physics Transformer for aerodynamics CFD that handles raw geometry without remeshing at 9M surface and 140M volume cells on a single GPU; NeuralDEM, described as the first end-to-end deep learning surrogate for large-scale multi-physics processes enabling real-time simulation of industrial processes such as fluidised bed reactors; and UPT, a Universal Physics Transformer framework for scaling neural operators across spatio-temporal problems, supporting grid and particle simulations.