TAIPEI – Nvidia revealed at Computex that Taiwan Semiconductor Manufacturing Company (TSMC) is now one of its customers for accelerated computing.

On top of a yearly $150 billion investment in its Taiwanese operations, Nvidia has grown its footprint in the nation by providing the chip giant with its Nvidia CUDA-X libraries and AI models to advance semiconductor design and manufacturing. Specifically, TSMC is using CuLitho (computational lithography), CuEST(computational electronic-structure simulation), CuML (computational machine learning), and H200 GPUs for GPU-accelerated scheduling to help manage complex fab constraints and streamline production paths.

Nvidia also said TSMC is using its Metropolis and the TAO Toolkit to improve automated defect inspection with vision AI, including detection of nanometer-scale defects, while reducing repeated labeling and retraining. In parallel, the Taiwanese giant is exploring Nvidia Omniverse libraries to build FabTwin, a virtual fab environment for testing layouts and simulation workflows before physical implementation.

"TSMC, one of our great partners and the world leader in semiconductors, is now a customer," said Rev Lebaredian, vice president of physical AI simulation at Nvidia.

"CuEst speeds up chemistry simulation by 50 times, letting them discover new materials in hours instead of days. Preparing one advanced chip design can consume 30 million CPU hours; CuLitho cuts that workload from 40,000 CPU systems to 350 H100 systems, reducing cycle times by nearly 50% and slashing costs on the fab floor."

The reveal sees the pair come full circle, with TSMC having long served as Nvidia's lead foundry partner. The vast majority of its GPUs are manufactured by the Taiwanese firm across sites in its home country and Arizona in the U.S.