Sign up for our daily and weekly newsletters to receive the latest updates and exclusive content on cutting-edge AI topics. Learn More
Nvidia and Foxconn have collaborated to build Taiwan’s largest supercomputer utilizing Nvidia Blackwell chips.
Announced at Hon Hai Tech Day, the Hon Hai Kaohsiung Super Computing Center will be powered by Nvidia’s Blackwell GPU architecture and feature the GB200 NVL72 platform, incorporating 64 racks with a total of 4,608 Tensor Core GPUs.
Expected to deliver over 90 exaflops of AI performance, this supercomputer will set a new benchmark in Taiwan’s computing landscape.
Once operational, Foxconn intends to leverage the supercomputer for advancements in cancer research, large language model development, and innovations in smart city initiatives, establishing Taiwan as a frontrunner in AI-driven industries.
Foxconn’s strategic focus on smart manufacturing, smart cities, and electric vehicles will be bolstered by the new supercomputer, supporting initiatives in digital twins, robotic automation, and smart urban infrastructure in places like Kaohsiung.
Construction is underway in Kaohsiung, Taiwan, with the first phase scheduled for completion by mid-2025 and full deployment by 2026.
The project will integrate Nvidia technologies such as Nvidia Omniverse and Isaac robotics platforms to enhance manufacturing processes with AI and digital twin capabilities.
“Empowered by Nvidia’s Blackwell platform, Foxconn’s new AI supercomputer stands as one of the most potent globally, marking a significant milestone in AI computing and efficiency,” stated James Wu, Vice President of Foxconn.
The GB200 NVL72 is a cutting-edge data center platform tailored for AI and accelerated computing, boasting 36 Nvidia Grace CPUs and 72 Nvidia Blackwell GPUs per rack connected through Nvidia’s NVLink technology, delivering 130TB/s bandwidth.
Featuring the Nvidia NVLink Switch, the system enables the 72-GPU setup to function as a unified unit, ideal for training large AI models and executing complex inference tasks on trillion-parameter models in real-time.
Foxconn, also known as Hon Hai Precision Industry Co., a top electronics manufacturer globally, has been utilizing Nvidia Omniverse to create digital twins of its factories, alongside pioneering the use of Nvidia NIM microservices in developing domain-specific large language models (LLMs) embedded into internal systems for smart manufacturing, smart electric vehicles, and smart cities.
