Foxconn announced at its annual tech day in Taipei that a $1.4 billion supercomputing center developed in partnership with NVIDIA is due in the first half of 2026. The company said the facility will become Taiwan’s largest advanced GPU cluster and will be powered by NVIDIA Blackwell GB300 hardware. Foxconn will operate the site under its new Artificial Intelligence and cloud unit, Visonbay Artificial Intelligence, and highlighted the project as part of a broader push into large-scale compute services.
NVIDIA vice president Alexis Bjorlin told attendees that rapid gains in GPU performance are shifting the economics of compute, and that many companies may find it more economical to rent capacity rather than build their own facilities. Foxconn said it plans to invest $2-3 billion per year in Artificial Intelligence, and chairman Young Liu said the company can already build around 1,000 Artificial Intelligence racks per week, a volume he expects to increase next year. Those production capabilities are presented as a core advantage for scaling new systems quickly.
Separately, OpenAI confirmed a new partnership with Foxconn to co-develop hardware for data centers. Under the agreement, Foxconn will design and manufacture components such as cabling, power systems and other infrastructure gear. The deal does not include purchase commitments, but gives OpenAI early access to test the hardware, mirroring the supplier feedback model Foxconn uses with Google, AWS and Microsoft. OpenAI CEO Sam Altman said demand for advanced Artificial Intelligence infrastructure continues to exceed supply and that upcoming models will require new types of server racks, cooling systems and power hardware, driving further hardware innovation and collaboration.
