Nvidia and Japan Partner to Build World's First National AI Infrastructure for 'Physical AI'

Nvidia and the Japanese government have launched a massive initiative to build the world's first national computing infrastructure dedicated entirely to "physical AI." The project centers on a 140-megawatt data center designed to train artificial intelligence models that understand real-world physics and can operate autonomous robots.[2][3]

The facility will be established and operated by Noetra Corp, a private consortium backed by Japanese industrial and technology heavyweights including SoftBank, NEC, Sony Group, and Honda. It will house 27,500 of Nvidia's next-generation Rubin GPUs and 13,750 Vera CPUs, making it one of the most powerful AI factories ever constructed.[4][5]

The initiative is heavily subsidized by the Japanese state. Japan's Ministry of Economy, Trade and Industry (METI) has committed an initial ¥387.3 billion ($2.4 billion) to the project, with total funding potentially reaching ¥1 trillion ($6.1 billion) over the next five years, subject to annual reviews.[2][4]

The investment is driven by demographic necessity. Japan is facing an acute labor shortage as its population rapidly ages, threatening its position as a global manufacturing powerhouse. By developing advanced physical AI, the country hopes to deploy millions of smart robots across factories, logistics centers, and elder-care facilities.[1][3]

The technology area known as "physical AI" refers to smart, futuristic robots that can think on their own, not just follow programmed directions, noting that they are designed to work safely alongside humans. Unlike large language models that generate text, physical AI models must process complex sensor data to navigate and manipulate the real world in real time.[1][6]

To build the software side of this ecosystem, Nvidia CEO Jensen Huang visited Tokyo to announce the "Nvidia Cosmos Coalition." The alliance brings together Japan's premier robotics manufacturers—including Fanuc, Yaskawa Electric, Kawasaki Heavy Industries, and Fujitsu—to develop open world models using Nvidia's simulation and training platforms.[6][7]

Huang framed the partnership as a natural fit for the country that pioneered modern industrial automation. "Japan invented modern manufacturing. Now, it has the opportunity to reinvent it for the age of intelligent industries," Huang said during the announcement, emphasizing Japan's reputation for precision engineering.[1][7]

The strategy represents a distinct approach to "sovereign AI." While European nations have debated the risks of relying on American hardware vendors, Japan has opted to purchase Nvidia's entire hardware and networking stack outright. In exchange, Japanese companies and researchers will retain ownership of the resulting AI models and deployment platforms.[4][5]

The stakes for Japan's industrial future are immense. The government's recently published AI Robotics Strategy sets a target for the country to capture more than 30% of the global AI robotics market by 2040, an industry projected to be worth $133 billion.[3]

By centralizing the immense compute power required to train these models, the Noetra consortium aims to lower the barrier to entry for domestic startups and researchers. The resulting open multimodal foundation models will be shared broadly with Japanese developers, accelerating the country's transition to a highly automated economy.[2][7]