NVIDIA outlines an AI-defined semiconductor fab workflow
NVIDIA described an “AI-defined fab” workflow combining accelerated simulation, physics-based AI, AI agents and digital twins across semiconductor manufacturing. The report cited vendor- and partner-reported speedups of up to 10x for COMSOL simulations, 10x for Applied Materials’ Ginestra platform, 17x for key Siemens Calibre workloads using cuLitho and 35x for Applied Materials’ ACE+ topography simulations. Lam Research intends to integrate NanoVDB into SEMulator3D, with the addition expected to accelerate selected wafer-etch modeling by up to 30x. Applied Materials also uses NVIDIA Omniverse libraries for fab-level digital twins, while Advantest uses NeMo Switchyard agents to automate model selection. NVIDIA said engineers remain responsible for validation and final decisions; the performance claims were not independently verified in the supplied reports.
The workflow connects simulation, AI models, agents and digital twins across chip manufacturing, allowing engineers to test more alternatives virtually and inform production decisions with production data. The reported performance figures remain vendor and partner claims rather than independently validated results.
Demand for advanced chips is rising. The processes to build them are getting more complex. And the margin for error is shrinking.
For engineers, understanding how a process change will behave, how a material will interact with a new step, how a defect maps back to a root cause are critical, challenging steps in semiconductor manufacturing — and these analyses can take days. Physical experiments on a fab line are slow and expensive. A wrong call on yield or process quality has real consequences.
AI is transforming semiconductor manufacturing, impacting what engineers can do — and how fast.
Accelerated computing shortens simulation time, allowing engineers to evaluate more alternatives before committing to physical trials.
Physics-based AI helps engineers predict process behavior without running a full simulation every time.
AI agents handle sequences of engineering tasks that once consumed hours.
#NVIDIA#COMSOL#Applied Materials#Siemens#Lam Research#Advantest#NVIDIA cuDSS#NVIDIA cuLitho#NVIDIA PhysicsNeMo#NVIDIA Omniverse#NVIDIA NeMo Switchyard#Ginestra