Japan Launches Domestic AI Project 'Noetra': NVIDIA to Supply 27,500 Rubin GPUs, 44 Companies Including Sony and SoftBank Participate

BlockBeats News, July 17 — Japan's domestic multimodal foundational model development project "Noetra" officially launched on July 16. The core companies involved are Sony Group, SoftBank, NEC, and Honda, with a total of 44 companies and organizations investing in the project. These participants span industries such as manufacturing, finance, logistics, and telecommunications. Engineers from institutions like the National Institute of Advanced Industrial Science and Technology (AIST) and Preferred Networks will also join the development efforts. This project is part of the Ministry of Economy, Trade and Industry's initiative for the "Development of Multimodal Foundational Models for AI Robots and Physical AI" (referred to by NVIDIA as the FRONTia Project). Its goal is to create Japan's domestic foundational models for physical AI scenarios such as manufacturing sites and robotics, rather than merely developing another Japanese-language conversational AI. On the hardware side, Noetra will collaborate with NVIDIA to build a computing platform equipped with approximately 27,500 of the latest Rubin GPUs and 13,750 Vera CPUs. The platform will utilize NVIDIA Vera Rubin NVL72 racks and the DSX platform architecture, with a designed power capacity of 140 megawatts. Construction is scheduled to begin in April 2027, with operations expected to start in June 2028. Once completed, it will become Japan's largest AI computing infrastructure. The development roadmap will proceed in three phases: - Starting in fiscal year 2026, the focus will be on developing inference foundational models centered on AI agents and natural language processing. - By fiscal year 2028, the goal is to achieve seamless integration of text, images, video, and audio into a fully multimodal foundational model. - By fiscal year 2030, the aim is to realize real-world native AI capable of understanding spatial and physical properties, with applications in manufacturing, logistics, healthcare, and telecommunications.