Hon Hai Precision Industry, the world’s largest contract electronics manufacturer, disclosed plans to allocate up to NT$42 billion ($1.37 billion) for AI computing equipment through the end of 2026. The Taiwan-based company, better known as Foxconn, specified the purchase will include 10,000 Nvidia GB300 modules for its supercomputing center in Kaohsiung.
The facility, operated through subsidiary Big Innovation Company, represents Foxconn’s latest effort to capitalize on surging demand for AI infrastructure beyond its traditional iPhone assembly business. The center will feature 27 megawatts of power capacity and emphasize inference computing rather than model training, according to Big Innovation CEO Yao Yan-zong.
Foxconn positions the project as addressing three objectives: retaining Taiwanese corporate data domestically, providing low-latency computing for AI applications, and hosting international large language models rather than developing proprietary ones. The company cited partnerships with Cathay Financial Holdings for financial AI testing.
The infrastructure demands significant reinforcement, requiring floor loads of two metric tons per square meter versus 500 kilograms for conventional office space. A second construction phase remains under consideration, though financial details weren’t disclosed. The announcement comes as Foxconn projects AI server revenue will exceed NT$1 trillion in 2025.
