Simcenter PhysicsAI 2026.1 adds diffusion-based concept generation, SPH results and multi-GPU training from STAR-CCM+

Siemens releases Simcenter PhysicsAI 2026.1, its geometric deep learning tool that trains surrogates on past simulation results. The release adds PhysicsAI Generate, an experimental diffusion-based generator of 3D design concepts from performance and dimensional targets, with optional conditioning on 3D point clouds. Models can now learn from SPH particles and point-mass elements, predict eigenfrequencies for modal analysis and predict stress and strain directly on elements. Siemens reports training up to 5× faster with up to 50% less peak memory than 2026.0, and the PhysicsAI add-on in Simcenter STAR-CCM+ gains multi-GPU training.

Share on LinkedIn, X or Bluesky, or copy the link.