AI speeds simulation and process testing in chip fabs

AI and accelerated computing are helping semiconductor engineers investigate process changes, predict material behavior and identify manufacturing problems faster. Shorter simulations let teams explore more alternatives before committing to expensive physical experiments, while engineers retain responsibility for production decisions.

COMSOL is using NVIDIA cuDSS to accelerate multiphysics simulations by up to 10x. Applied Materials’ Ginestra platform runs up to 10x faster than CPU-only approaches with the same library, while its ACE+ platform delivers topography simulations up to 35x faster using NVIDIA PhysicsNeMo. Siemens is integrating NVIDIA cuLitho into Calibre lithography software, delivering up to 17x acceleration on key simulation workloads.

Physics-based AI can also avoid the need to run a full simulation for every prediction. Lam Research demonstrated plasma-behavior predictions in milliseconds using PhysicsNeMo, compared with days for a full simulation. Advantest is using NVIDIA NeMo Switchyard agents to automate AI model selection on the test floor, reducing response times from hours to minutes.

Applied Materials is building physically accurate factory digital twins with NVIDIA Omniverse libraries to simulate material movement, identify bottlenecks and evaluate layouts before deployment. Connecting production data, predictive models and virtual testing could help engineers apply lessons from manufacturing cycles sooner and assess changes across entire fabs.