Nvidia is pushing artificial intelligence further into semiconductor factories with a new software library and expanded partnerships across the chip industry. The company says the tools could help engineers simulate materials, improve production yields and increase the supply of advanced chips.
cuPotential models atomic behavior
Nvidia introduced cuPotential, a CUDA-X library that developers can use to build AI agents for simulating how atoms and molecules interact. These calculations help researchers study semiconductor materials and manufacturing processes before changes are made on a production line.
Nvidia said early-access testing made already GPU-optimized workflows another two to three times faster. Synopsys is among the early adopters using cuPotential for materials simulations, according to Nvidia.
Chipmakers adopt factory AI
The announcement also covers wider use of Nvidia software by Applied Materials, Cadence, KLA, Lam Research, Siemens and Synopsys. Their work spans chip design, factory simulation, inspection and production testing.
Nvidia describes the result as an “AI-defined fab,” meaning a semiconductor factory where AI helps engineers optimize processes, solve production issues and improve efficiency. The company said Lam Research testing showed up to 30 times faster chipmaking simulation with NanoVDB, an open-source library for working with three-dimensional data. Nvidia NPP tools are also being used to accelerate chip inspection at KLA.
Early access limits availability
The effort affects companies that design and manufacture processors, memory and other semiconductor components. It could also influence the broader technology market because advanced chips are used in AI systems, smartphones, cars and data centers.
Commercial availability is still limited. Nvidia says cuPotential is in early access, while NanoVDB is available as open source within OpenVDB. Lam Research’s NanoVDB-accelerated SEMulator3D capability is targeted for release by the end of 2026. Nvidia is showcasing the technology during SEMICON West 2026, which runs from October 13 through October 15.
References
NVIDIA. (2026, October 14). NVIDIA and semiconductor leaders advance AI across chip design and manufacturing for AI-defined fabs. NVIDIA Newsroom. https://nvidianews.nvidia.com/news/nvidia-and-semiconductor-leaders-advance-ai-across-chip-design-and-manufacturing-for-ai-defined-labs
NVIDIA. (n.d.). NVIDIA at SEMICON West 2026. https://www.nvidia.com/en-us/events/semicon-west/