SoftBank is in talks with US Chip giant Nvidia and Taiwanese contract manufacturer Foxconn to build Japan’s AI servers. The company is actively exploring partnerships to expand its domestic AI infrastructure business, marking a significant transition of the company from a technology investor to a hardware manufacturer.SoftBank plans to begin the server design and component assembly before the end of the decade. As per the reports, the unit will focus on high-performance servers to run advanced graphics processing units at high speed.SoftBank's building of Japan’s AI server will strengthen Japan’s vision to develop independent AI computing capacity, eliminating its dependence on foreign hardware supply chains when the demand for AI is surging globally.The project is expected to be part of SoftBank’s medium-term management plan, and this comes after it invested more than USD 30 billion in OpenAI for 11 percent stake.Also Read: Microsoft's Massive $10B Move Signals Japan AI SurgeSoftBank Corp, the company’s telecom unit, has already designed and tested a rack-integrated AI server system built around Nvidia H200 GPUs, showing that the project has already started.It has worked in collaboration with Foxconn and ZutaCore, a manufacturer of liquid- cooling systems for data center chips, achieving a partial Power Usage Effectiveness rating of 1.03. This is exceptional power effectiveness for a system running heavy AI workloads.SoftBank Corp. has plans to supply AI computing capacity in Japan through a partnership with Microsoft and Sakura Internet, a cloud and internet infrastructure provider, with domestic hosted resources. For Japan, stressing its focus on developing AI infrastructure, SoftBank’s entry into the manufacturing sector will be critical in achieving this national ambition.