After concluding a two-day trip to Tokyo, NVIDIA has further clarified its partnership landscape in Japan. The focus of this initiative goes beyond selling chips to include national AI infrastructure, deployment of robotics models, and deeper integration of AI into automotive and manufacturing systems. For Japan, this is also part of its broader plan to upgrade manufacturing and achieve AI autonomy.
Japan launches a national AI factory
Japan is advancing its domestic AI initiative called Noetra, with key participants including SoftBank, Sony, NEC, and Honda, aiming to develop native models for robotics, automotive, and factory applications. According to reports, Japan plans to invest up to 1 trillion yen over five years, betting on "physical AI" capable of directly controlling machine operations.
NVIDIA will build an AI factory based on the Vera Rubin architecture—a large-scale data center—under this framework. The project is scheduled to launch in 2028, featuring 13,750 Vera CPUs and 27,500 Rubin GPUs, with a power capacity of approximately 140 megawatts. Noetra will oversee the overall implementation and plans to construct the data center.
Noetra’s roadmap is divided into three phases: first, launching a reasoning model optimized for Japanese language capabilities in fiscal year 2026; then, expanding around 2028 to a multimodal version capable of processing text, images, video, and audio; and finally, advancing to “reality-native AI” capable of directly operating robotic systems by 2030, with gradual openness to external developers.
Robot companies are gathering around Cosmos.
NVIDIA’s second focus during this trip is to bring Japanese manufacturing and robotics companies into its Cosmos model ecosystem. The report states that Fanuc, Yaskawa Electric, Kawasaki Heavy Industries, Fujitsu, Hitachi, NEC, Sony, SoftBank, Kubota, and the robotics organization AIRoA all plan to develop based on the Cosmos model.
NVIDIA also launched Cosmos 3 Edge in Tokyo. This version runs on the Jetson Thor chip and can be deployed directly within machines, rather than relying solely on the cloud. For factory robots and industrial equipment, this means models can operate closer to the edge.
Some companies have already begun testing shared control systems, with Honda’s R&D division and Omron among those already using related tools. NVIDIA aims to extend the application of generative AI from data centers further into factory floors and robotic systems.
The collaboration with Toyota extends to factory and vehicle software.
Toyota is another key partner in Huang’s visit. Toyota has already integrated NVIDIA chips into its automotive technology stack and committed in January 2025 to adopt the NVIDIA Drive platform for its next-generation vehicles. The latest collaboration extends further into manufacturing and transportation systems.
Reports indicate that the collaboration between the two parties covers production line simulation, vehicle operation software, and road traffic recognition systems. Toyota remains cautious in its approach to autonomous driving, with its current focus on advanced driver-assistance systems that require driver involvement, rather than solutions that minimize human intervention.
Japan has incorporated physical AI into its industrial strategy.
The broader signal from Huang’s visit is that Japan is placing physical AI at the center of its industrial policy. Facing a shrinking workforce, Japan aims to deploy 10 million AI-equipped robots across 18 industries by 2040, supported by approximately $65 billion in public and private investment.
Japan’s AI robotics strategy, released in March this year, aims to capture more than 30% of the global AI robotics market by 2040. The Japanese government estimates this market size at approximately 20 trillion yen. The Ministry of Economy, Trade and Industry is also supporting the development of domestic foundational models, seeking to keep data, computing power, and industrial applications within the national system.
However, while Japan is advancing its AI autonomy, its underlying computing power still heavily relies on NVIDIA. In other words, Japan aims to build a more independent AI industrial system, but at this stage, it still depends on U.S. chips.