“NVIDIA and TSMC Advance AI in Semiconductor Fabs”

NVIDIA and TSMC have announced an expansion of their partnership to integrate AI and accelerated computing directly into semiconductor manufacturing processes. This collaboration aims to enhance chip production speed, increase yields, and optimize fab operations. The move underscores the growing importance of AI in advanced fabs as chip manufacturing becomes more intricate at smaller process nodes.

The partnership, revealed at NVIDIA GTC Taipei, encompasses various semiconductor workflows, including computational lithography, transistor simulation, defect inspection, and factory scheduling. This development signifies a shift where AI is no longer confined to data centers and cloud applications but is now being leveraged by semiconductor manufacturers to streamline physical production processes, potentially reducing costs and expediting the introduction of advanced chips in automotive, industrial, and edge AI markets.

For eeNews Europe readers, this collaboration signifies a significant step towards integrating AI into semiconductor manufacturing. By utilizing NVIDIA CUDA-X libraries and GPU-accelerated tools, TSMC is enhancing simulation and process optimization within its fabs. Technologies like NVIDIA cuLitho and cuEST are improving lithography and electronic structure simulations, respectively, leading to cost savings and faster processing times.

Moreover, TSMC is employing NVIDIA cuML for process control, enabling the analysis of vast amounts of manufacturing parameters to enhance production consistency. The use of CUDA-powered computation on NVIDIA H200 GPUs is also boosting fab scheduling and productivity, aiding TSMC in managing complex production constraints in advanced manufacturing facilities.

Jensen Huang, founder and CEO of NVIDIA, emphasized the significance of this collaboration, stating, “TSMC is bringing NVIDIA AI and accelerated computing into the fab itself, tackling some of the world’s most complex design and manufacturing challenges with simulation, optimization, and AI to improve speed, efficiency, and yield for the next generation of chips.”

Vision AI for defect inspection

In addition to process optimization, TSMC is utilizing the NVIDIA Metropolis platform and TAO Toolkit to enhance automated defect inspection. By leveraging vision AI, TSMC can identify nanometer-scale defects during manufacturing, improving inspection accuracy while reducing the need for repeated data labeling and retraining as production conditions evolve.

C.C. Wei, chairman and CEO of TSMC, highlighted the long-standing partnership with NVIDIA, stating, “By using NVIDIA accelerated computing and AI across fab operations optimization, lithography, process control, and inspection, TSMC is strengthening our technology leadership and manufacturing excellence to support our customers’ future products and success.”

Digital twins for future fabs

TSMC is exploring NVIDIA Omniverse technologies to develop FabTwin, a virtual environment for simulating fab layouts and manufacturing workflows before physical deployment. This approach allows TSMC to digitally test different factory configurations, identify bottlenecks early, and enhance planning efficiency without the need for costly physical infrastructure commitments.

As semiconductor fabs continue to grow in size and automation, the integration of digital twins and AI-based optimization is expected to play a crucial role in reducing operational complexity and enhancing manufacturing flexibility.