Nvidia and Foxconn to build 10,000-GPU AI supercomputer in southern Taiwan
Nvidia and Foxconn have extended their long-standing collaboration with plans for an “AI factory” supercomputer in southern Taiwan. The system, scheduled to be delivered by Foxconn subsidiary Big Innovation Company as an Nvidia Cloud Partner, will combine 10,000 Nvidia Blackwell GPUs into a single installation.
With roughly several hundred million US dollars behind it, the project is smaller than Elon Musk’s 200,000 GPU Memphis Supercluster but still ranks among the world’s most substantial AI investments. Nvidia will supply the hardware, while Foxconn provides the datacenter infrastructure and operational support.
Taiwan’s National Science and Technology Council intends to allocate the cluster’s computing power to local researchers, start-ups, and established industries, accelerating domestic AI adoption. Taiwan Semiconductor Manufacturing Company (TSMC) also plans to run advanced R&D workloads on the machine, expecting performance several orders of magnitude above previous systems.
Company leaders frame the initiative as a foundation for a wider AI ecosystem. Nvidia chief executive Jensen Huang described AI as the engine of “a new industrial revolution,” while Foxconn chair Young Liu highlighted the goal of connecting public agencies and private enterprises through shared infrastructure. TSMC chair C.C. Wei regards the cluster as a means to speed semiconductor innovation.
Foxconn will use the supercomputer internally to refine smart-city services, improve driver-assistance functions for its electric-vehicle platform, and streamline manufacturing flows with digital-twin technology. Taiwan’s science minister, Wu Cheng-Wen, says the broader objective is to build “a smart AI island” that links citizens, businesses, and government through advanced computing resources.
The supercomputer also fits Nvidia’s wider expansion strategy. In recent weeks, the company has announced a Shanghai R&D center and signalled plans to assemble a half-trillion-dollar fleet of AI servers in the United States.
