TAIPEI (Taiwan News) — Japanese real estate developer Mitsui Fudosan will establish a physical AI development center in Kumamoto Prefecture to develop next-generation chip technology, Nikkei Asia reported Monday.
The center will be built in the Kumamoto Science Park semiconductor industry cluster in Koshi, Kumamoto Prefecture, with completion targeted for 2030. It aims to create a shared development base where companies, universities, and research institutions can work on technologies for physical AI applications.
The facility will include a shared-use clean room for developing semiconductor assembly technologies used in physical AI systems. Mitsui Fudosan expects participation from semiconductor equipment makers, chemical companies, universities, and small and mid-sized companies.
Taiwanese companies and research institutions will also be invited to participate to develop prototypes using AI chips that TSMC plans to mass-produce in Kumamoto in 2028. The location could also benefit from nearby manufacturing operations by companies such as Yaskawa Electric and Toyota Motor.
The project will focus on integrating AI semiconductors into physical systems such as robots and automobiles while improving performance and reducing costs, per Nikkei Asia. Advanced AI chips produced by companies including TSMC and Nvidia require complex packaging technologies, making efficient assembly a key challenge.
Mitsui Fudosan said the hub could strengthen Japan’s technological capabilities and reduce supply chain risks by creating a domestic environment for advanced semiconductor development. TSMC’s expansion in Kumamoto has attracted companies seeking access to AI chips while reducing exposure to geopolitical risks involving Taiwan.
Funding is a concern as the shared clean room is expected to cost nearly NT$2 billion (US$61 million) to develop. The Japanese government plans to invest more than NT$611.5 billion by fiscal 2040 in growth areas including physical AI and semiconductors, and the project could become a major initiative under that strategy.
