During Foxconn's annual tech day in Taipei, the company announced that its $1.4 billion supercomputing center, built in partnership with NVIDIA, will be ready in the first half of 2026. The facility will become Taiwan's largest advanced GPU cluster, powered by NVIDIA Blackwell GB300 hardware and operated under Foxconn's new AI and cloud unit, Visonbay.ai. NVIDIA VP Alexis Bjorlin said that with GPU performance rising so quickly, many companies will find it more economical to rent compute capacity instead of building their own facilities. Foxconn plans to invest $2-3 billion per year in AI as Chairman Young Liu says the company can already build around 1,000 AI racks per week, a number set to rise next year.

In a separate announcement, OpenAI confirmed a new partnership with Foxconn to co-develop hardware for data centers. Foxconn will design and manufacture components such as cabling, power systems, and other infrastructure gear. While the agreement doesn't include purchase commitments, OpenAI will get early access to test the hardware. Foxconn's deepening collaboration with OpenAI mirrors the model it already uses with Google, AWS, and Microsoft. The partnership gives Foxconn better feedback on rack design and helps speed up new system development. OpenAI CEO Sam Altman said that demand for advanced AI infrastructure continues to exceed supply, and that new AI models will require new types of server racks, cooling systems, and power hardware.

More...