NVIDIA and Japan-backed Noetra announced plans for a large Vera Rubin AI factory in Japan, with roughly 27,500 Rubin GPUs, 13,750 Vera CPUs and 140 MW of data-center capacity aimed at physical AI and robotics research.
NVIDIA and Noetra announced plans to build a large-scale AI computing facility in Japan using NVIDIA’s Vera Rubin platform.
NVIDIA said in a July 2026 press release that it is working with Noetra and Japanese industrial partners on what it describes as a national AI infrastructure initiative for Japan. The company said the project is intended to support “national physical AI,” including robotics and AI systems that interact with the physical world.
According to NVIDIA, the planned AI factory will use 13,750 NVIDIA Vera CPUs and 27,500 NVIDIA Rubin GPUs, with 140 MW of data-center capacity. NVIDIA framed the effort as part of Japan’s broader strategy to develop AI for robotics, manufacturing and industrial applications.
Noetra, a Japanese company backed by public and private investors, separately said in a press release distributed by NEC that it had begun full-scale research and development for a Japan-developed multimodal foundation model. Noetra said it plans to build AI computing infrastructure in collaboration with NVIDIA, equipped with about 27,500 NVIDIA Rubin GPUs.
Noetra said construction is planned to begin in April 2027, with operations expected to start in June 2028. The company said the infrastructure will be used for developing a multimodal foundation model, referring to AI systems that can process multiple types of data such as text, images, video and sensor inputs.
Data Center Dynamics, citing NVIDIA and Noetra’s announcements, reported that the planned facility will include 13,750 Vera CPUs and 27,500 Rubin GPUs and will support the Japanese government’s AI robotics strategy.
Reuters, in a report carried by Boursorama, said state-backed Noetra’s investors include Sony and that Noetra announced plans to buy 27,500 NVIDIA Rubin chips for physical-AI development work. Reuters also reported that NVIDIA is partnering with Japanese robotics-focused companies as part of the initiative.
The scale of the planned deployment is notable because NVIDIA’s Rubin platform is the company’s next-generation architecture following Blackwell and Vera Rubin is designed for high-performance AI training and inference. The announced numbers suggest Japan is preparing dedicated computing capacity for domestic AI model development rather than relying solely on overseas cloud resources.
The focus on physical AI also aligns with Japan’s industrial strengths. NVIDIA, Noetra and NEC’s materials point to robotics, manufacturing and multimodal model development as key use cases. Japan has a large automotive, electronics and robotics base, and the project appears designed to connect AI model development with industrial deployment.
Still, the project remains forward-looking. Noetra’s own timeline places construction in 2027 and operations in 2028, so the announced capacity is not yet online. The final scope will depend on execution, financing, supply availability and data-center construction.
NVIDIA called the initiative the “world’s first national AI infrastructure” in its press release. That description is NVIDIA’s characterization; the concrete, source-backed details are that NVIDIA and Noetra plan a 140 MW AI factory in Japan using 13,750 Vera CPUs and about 27,500 Rubin GPUs, with construction planned for April 2027 and operations expected in June 2028.
NVIDIA and Noetra announced plans to build a large scale AI computing facility in Japan using NVIDIA’s Vera Rubin platform.
A national push for physical AI NVIDIA said in a July 2026 press release that it is working with Noetra and Japanese industrial partners on what it describes as a national AI infrastructure initiative for Japan.
The company said the project is intended to support “national physical AI,” including robotics and AI systems that interact with the physical world.
Continue reading
