Foxconn to Construct Taiwan’s Fastest AI Supercomputer Powered by Nvidia Blackwell

Nvidia and Foxconn are collaborating to build Taiwan's largest supercomputer, leveraging Nvidia's Blackwell chips.

The project, named the Hon Hai Kaohsiung Super Computing Center, was announced during Hon Hai Tech Day. It will harness Nvidia's advanced Blackwell graphics processing unit (GPU) architecture and feature the GB200 NVL72 platform, comprising 64 racks and 4,608 Tensor Core GPUs.

With an astonishing performance exceeding 90 exaflops of AI processing power, this supercomputer is poised to become the fastest in Taiwan.

Once operational, Foxconn intends to utilize the supercomputer to drive innovations in cancer research, enhance large language model development, and support smart city initiatives, positioning Taiwan as a frontrunner in AI-driven technologies.

Foxconn’s "three-platform strategy" emphasizes smart manufacturing, smart cities, and electric vehicles. The new supercomputer will be crucial in advancing digital twins, robotic automation, and smart urban infrastructure, delivering AI-assisted services to cities like Kaohsiung.

Construction is underway in Kaohsiung, Taiwan, with the first phase expected to be operational by mid-2025 and full deployment aimed for 2026.

The project will integrate Nvidia technologies such as Nvidia Omniverse and the Isaac robotics platform to revolutionize manufacturing through AI and digital twins.

"Powered by Nvidia’s Blackwell platform, Foxconn’s new AI supercomputer is among the most powerful worldwide, marking a significant advancement in AI computing efficiency," stated Foxconn vice president James Wu.

The GB200 NVL72 data center platform is optimized for AI and accelerated computing. Each rack is equipped with 36 Nvidia Grace CPUs and 72 Nvidia Blackwell GPUs, interconnected via Nvidia’s NVLink technology, providing an impressive 130TB/s bandwidth.

Nvidia NVLink Switch allows the 72-GPU system to operate as a unified GPU, making it ideal for training large AI models and performing complex inference tasks in real-time on trillion-parameter models.

Based in Taiwan, Foxconn, officially known as Hon Hai Precision Industry Co., is the world's largest electronics manufacturer, producing a diverse array of products, from smartphones to servers, for leading technology brands. Foxconn is also developing digital twins of its factories using Nvidia Omniverse and was one of the pioneers in employing Nvidia NIM microservices for domain-specific large language models (LLMs) to enhance its internal systems and processes for smart manufacturing, smart electric vehicles, and smart cities.