From NVIDIA: 2024-10-08 01:00:28

NVIDIA and Foxconn collaborate on building Taiwan’s largest supercomputer using NVIDIA’s Blackwell architecture and GB200 NVL72 platform, with 4,608 Tensor Core GPUs. Expected to deliver over 90 exaflops of AI performance, the supercomputer aims to drive breakthroughs in cancer research, language models, and smart city innovations.

Foxconn’s supercomputer will support its “three-platform strategy” focusing on smart manufacturing, smart cities, and electric vehicles. Construction has commenced in Kaohsiung, Taiwan, with plans for partial operation by mid-2025 and full deployment by 2026.

The supercomputer integrates NVIDIA technologies like NVIDIA Omniverse and Isaac robotics platforms for AI and digital twins, transforming manufacturing processes. Foxconn’s Vice President James Wu highlights the power and efficiency of the new AI supercomputer, utilizing state-of-the-art data center platforms optimized for AI and accelerated computing.

The GB200 NVL72 features 36 NVIDIA Grace CPUs and 72 Blackwell GPUs connected via NVLink technology, delivering 130TB/s bandwidth. The NVIDIA NVLink Switch enables the 72-GPU system to function as a single GPU, ideal for training large AI models and real-time complex inference tasks on trillion-parameter models.

Foxconn, a key player in global technology supply chains, leads in smart manufacturing and industrial AI with digital twin technology. Utilizing NVIDIA’s NIM microservices, Foxconn embeds large language models into AI factories for smart manufacturing, electric vehicles, and smart cities.

The Hon Hai Kaohsiung Super Computing Center is part of a global network of advanced supercomputing facilities powered by NVIDIA, enabling cutting-edge computational power to accelerate research and innovation in scientific disciplines. Explore more at Hon Hai Tech Day.



Read more at NVIDIA: Foxconn to Build Taiwan’s Fastest AI Supercomputer With NVIDIA Blackwell