TSMC deploys NVIDIA AI across its semiconductor manufacturing operations
NVIDIA announced on Sunday that TSMC is deploying its accelerated computing and artificial intelligence technologies throughout the semiconductor design and manufacturing cycle, marking a deepening of a partnership that spans nearly three decades between the chip designer and the world's leading foundry. The announcement was made at GTC Taipei on June 1 and covers a broad range of applications from lithography to factory scheduling.
TSMC is applying NVIDIA cuLitho, a GPU-accelerated computational lithography library, to achieve improvements of 20 to 50 percent in cost efficiency or turnaround time compared to CPU-based approaches at equivalent total cost of ownership. For transistor and process simulation, the foundry uses NVIDIA cuEST, which delivers an average 50x acceleration of chemical simulations used in semiconductor materials design. TSMC also uses NVIDIA's cuML machine learning library to accelerate large-scale analysis, synthesizing hundreds of thousands of process parameters across thousands of manufacturing steps to reduce process variation. GPU-accelerated scheduling computation via CUDA on NVIDIA H200 GPUs has improved fab productivity by helping TSMC manage complex operational constraints and streamline production flows.
Beyond computational acceleration, TSMC uses the NVIDIA Metropolis platform and the NVIDIA TAO Toolkit to improve automated defect classification through visual AI, enhancing nanoscale defect detection while reducing the need for repeated relabeling and retraining as manufacturing conditions evolve. The foundry is also exploring NVIDIA Omniverse to build what it calls FabTwin, a virtual manufacturing environment designed to evaluate equipment layout and simulation workflows. By testing design scenarios digitally before physical implementation, TSMC aims to identify potential constraints earlier and accelerate capital decision-making.
NVIDIA chief executive Jensen Huang stated that TSMC is integrating NVIDIA AI and accelerated computing at the heart of its fabs to tackle some of the most complex design and manufacturing challenges in the world. TSMC president and chief executive C.C. Wei said the collaboration strengthens the company's technology leadership and manufacturing excellence in service of future products and customer success. The announcement came during a busy week of activity in Taipei, where Huang delivered a keynote address covering topics ranging from the full production ramp of the Vera Rubin AI platform to a redesigned PC lineup and advances in physical AI.
-
19:48
-
08:30
-
19:28
-
17:15
-
15:45
-
13:15
-
12:30
-
11:02
-
08:42