Weeks after his keynote speech in Taiwan and months after his visit to South Korea, Nvidia CEO Jensen Huang spent July 15th and 16th in Tokyo courting Japan’s industrial and chip-supply elite. He leaves with deals across Japan’s technology ecosystem, including a national AI factory, partnerships with the nation’s leading robotics companies, and deals with chip material suppliers to power Nvidia’s next-generation AI chips. His message was clear. Nvidia is targeting Japanese factory floors, and many of Japan’s major manufacturers are participating. Huang said the next chapter of AI will belong to factory floors, robots and machines, and he wants Japan to build it.
Thirty years ago, Sega’s $5 million investment kept Nvidia afloat from the brink of bankruptcy. Today, Nvidia and the Japanese industrial giant need each other again. This time we’re building the era of physical AI, starting with the following three projects:
Noetra — A battle between the Japanese sovereign and the AI. This country does not want to run its factories and robots using AI from the US or China. Therefore, the government brought together approximately 44 domestic companies, mainly Softbank, Sony, NEC, and Honda, to build its own AI for robots, vehicles, and factory floors. Tokyo is spending up to 1 trillion yen ($6.2 billion) over five years on domestically produced “physical AI,” the basic model built to run machines. Japan wants to own the software brain. However, the hardware to build it is still provided by Nvidia. The US chip giant is building a ‘Vera Rubin AI Factory’. It will be a massive data center packed with next-generation chips that will feature 13,750 Vera CPUs and 27,500 Rubin GPUs, delivering 140 megawatts and is scheduled to go live in 2028. Noetra is overseeing this effort and plans to build the data center. Noetra’s plan will be carried out in three stages. The first is an inference model that focuses on Japanese language skills and will begin in 2026. Omnimodal version to process text, images, video and audio by 2028. The “real-world native AI” built to run robots by 2030 was released to external Noetra developers in stages.
Robot Alliance — Japanese industrial giants line up behind Cosmos. Nvidia is targeting the Japanese factory floor, with many of Japan’s top robotics and manufacturing companies signing up. Fanuc, Yaskawa Electric, Kawasaki Heavy Industries, Fujitsu, Hitachi, NEC, Sony, Softbank, Kubota, and robotics group AIRoA announced plans to build on Nvidia’s Cosmos model. This is an open model initiative that Nvidia launched in May with several global AI labs. In Tokyo, Nvidia announced Cosmos 3 Edge, a version of the model that runs on the Jetson Thor chip in the machine itself, giving them a reason to commit. Some are already testing shared control systems. Other companies are currently building this tool, including Honda R&D and Omron. “The next frontier for AI is in the physical world, and this is a once-in-a-generation opportunity for Japan,” Huang said in a company statement. “Japan invented modern manufacturing, and now Japan has an opportunity to reinvent manufacturing for the era of intelligent industry.”
Toyota — Cars and Physics AI. Toyota uses Nvidia chips in much of its stack. The company committed its next-generation vehicle to Nvidia’s Drive platform at CES in January 2025. New efforts extend Nvidia to manufacturing sites that use simulation to design production lines, software that moves vehicles, and systems that read road traffic. Toyota’s cars will run advanced driver assistance that provides functional steering and brakes but still requires a driver, a more conservative approach than Waymo and Tesla, which are developing systems that rely less on human drivers.
Hwang’s visit puts physical AI at the center of Japan’s industrial strategy, with the Japanese government pouring money into supporting it. Facing a shrinking workforce, Japan wants to deploy 10 million AI-powered robots in 18 fields by 2040, backed by $65 billion in public and private physical AI investment.
The longer the game, the bigger it is. Japan’s AI robotics strategy, announced in March, aims to capture more than 30% of the global AI robotics market by 2040. This market is valued by Tokyo at about 20 trillion yen, or about $133 billion. METI is funding the national foundational model to run the machine, and Noetra’s Nvidia-powered factory will train models of that scale, spanning trillions of parameters. The bet is whether Japan’s factory floor data and manufacturing sites can be turned into physical AI.
Under industrial litigation there are sovereign litigation. As the U.S. and China develop large-scale AI, the Japanese government wants its own data and its own computing, reducing its dependence on infrastructure it doesn’t control. On July 16, Huang attended the government’s physics AI presentation together with Trade Minister Yoshinari Akazawa, and Prime Minister Sanae Takaichi also participated via video. The Takaichi administration has placed AI and semiconductors at the center of its growth plan, which aims to invest 370 trillion yen ($2.3 trillion) in public and private sectors by 2040. The Noetra factory, which NVIDIA touts as “the world’s first national AI infrastructure,” is its clearest bet yet. Japan’s push for independence, at least for now, relies on American chips, and ndence runs on American silicon.
Over the course of two days, Hwang had lunch with nearly every important figure in Japan’s technology industry, including the CEOs of Toyota, Fanuc, Yaskawa Electric, Fujitsu, and Kawasaki Heavy Industries, and sat across from dozens of supply chain executives over skewers and whiskey at a Kanda izakaya.
This is the same strategy he implemented a few weeks ago. A return keynote in Taiwan, fried chicken, and a 50,000 GPU deal in Seoul last fall. Now it was Tokyo’s turn with its robots, supply chain, and chips underneath.
If you buy through links in our articles, we may earn a small commission. This does not affect editorial independence.
