nicholashan - Fotolia
Foxconn and Nvidia to build Taiwan’s fastest supercomputer
The supercomputer, powered by Nvidia’s Blackwell architecture, will drive breakthroughs in cancer research, large language models and smart city initiatives
Nvidia and contract manufacturing giant Foxconn have teamed up to build Taiwan’s largest and fastest supercomputer, marking a significant milestone in the island’s artificial intelligence (AI) capabilities.
Announced at Hon Hai Tech Day, the supercomputer will leverage Nvidia’s Blackwell architecture and the GB200 NVL72 platform comprising 64 racks and 4,608 Tensor Core graphics processing units (GPUs), boasting an expected AI performance exceeding 90 exaflops.
Once operational, Foxconn plans to leverage the supercomputer to drive breakthroughs in key areas such as cancer research, large language model (LLM) development and smart city innovations, positioning Taiwan as a global leader in AI-driven industries.
The supercomputer will also be instrumental in supporting Foxconn’s “three-platform strategy” focused on smart manufacturing, smart cities and electric vehicles. It will also enhance ongoing developments in digital twins, robotic automation and smart urban infrastructure, bringing AI-assisted services to urban areas such as Kaohsiung.
Construction of the supercomputer has commenced, with the first phase expected to be operational by mid-2025 and full deployment targeted for 2026. The project will integrate with Nvidia technologies, including Omniverse and Isaac robotics platforms, to further advance AI and digital-twin technologies and transform manufacturing processes.
“Powered by Nvidia’s Blackwell platform, Foxconn’s new AI supercomputer is one of the most powerful in the world, representing a significant leap forward in AI computing and efficiency,” said Foxconn vice-president James Wu.
The GB200 NVL72 is Nvidia’s datacentre platform optimised for AI workloads. Each rack incorporates 36 Nvidia Grace server processors and 72 Nvidia Blackwell GPUs interconnected via Nvidia’s NVLink technology to deliver 130TB per second of bandwidth. The GPUs can operate as a single, unified GPU, ideal for training large AI models and executing complex inference tasks on trillion-parameter models in real time.
Read more about AI in APAC
- Some 500 customer service officers at Singapore’s DBS Bank will soon be able to tap a GenAI-powered virtual assistant to improve workflows and better serve customers.
- Snowflake’s regional leader Sanjay Deshmukh outlines how the company is helping customers to tackle the security, skills and cost challenges of AI implementations.
- The Australian government is experimenting with AI use cases in a safe environment while it figures out ways to harness the technology to benefit citizens and businesses.
- Pure Storage’s global CTO discusses the data and sustainability challenges in AI adoption, which can be addressed by centralising datasets and focusing on data quality and management.
Foxconn, also known as Hon Hai Precision Industry, was an early adopter of Nvidia inference microservices, developing domain-specific LLMs embedded within its AI factories for smart manufacturing, smart electric vehicles, and smart cities.
The new supercomputer, when ready, will join a growing global network of advanced supercomputing facilities powered by Nvidia across Europe and Asia, representing a substantial advancement in computational power, and driving research and innovation across various scientific disciplines.
