Nvidia CEO Jensen Huang spent two days in Tokyo, July 15 and 16, and walked away with deals touching nearly every corner of Japan’s tech economy. According to TechCrunch AI, he left with a national AI factory, partnerships with the country’s top robotics firms, and agreements with the chip-material suppliers feeding Nvidia’s next-generation chips. His pitch was simple: AI’s next chapter belongs to factory floors, robots, and machines, and he wants Japan to build it.
This matters because it moves the AI story off the chatbot and onto the assembly line. TechCrunch AI reports that Japan is betting big on “physical AI,” the foundation models built to run machines rather than answer questions. And Tokyo is spending real money to back it.
The three projects that anchor the visit
- Noetra, Japan’s sovereign-AI play. The government pulled together roughly 44 domestic firms, with SoftBank, Sony, NEC, and Honda at the core, to build homegrown AI for robots, vehicles, and factories. Tokyo is committing up to 1 trillion yen ($6.2 billion) over five years. Japan wants to own the software brain. The hardware still comes from Nvidia.
- The robotics coalition lines up behind Cosmos. Fanuc, Yaskawa, Kawasaki Heavy, Fujitsu, Hitachi, NEC, Sony, SoftBank, Kubota, and robotics group AIRoA plan to build on Nvidia’s Cosmos models. In Tokyo, Nvidia unveiled Cosmos 3 Edge, a version that runs on its Jetson Thor chips inside the machines themselves.
- Toyota goes all in. Toyota already runs Nvidia chips across much of its stack and committed its next-generation vehicles to Nvidia’s Drive platform back in January 2025. The newer work pushes Nvidia into Toyota’s manufacturing simulations, its in-car software, and its traffic-reading systems.
The factory underneath it all
The centerpiece is what Nvidia calls a “Vera Rubin AI factory,” a data center packed with its next-generation chips and expected to launch in 2028. TechCrunch AI details the specs: 13,750 Vera CPUs and 27,500 Rubin GPUs, delivering 140 megawatts. Noetra will oversee the build.
Noetra’s software roadmap runs in three stages. A reasoning model heavy on Japanese-language skills starts in fiscal 2026. An omni-modal version handling text, images, video, and audio follows by 2028. Then “Real-world Native AI” built to run robots arrives by 2030.
Why it matters
What stands out here is the scale of the national commitment. Facing a shrinking workforce, Japan wants 10 million AI-equipped robots across 18 sectors by 2040, backed by $65 billion in public and private physical-AI investment. The country’s AI Robotics Strategy, released in March, aims to grab more than 30% of a global AI robotics market Tokyo values at roughly $133 billion.
There’s a sovereignty story running underneath the industrial one. As the U.S. and China pull ahead in large-scale AI, Japan wants its own data, its own compute, and less dependence on infrastructure it doesn’t control. Huang appeared July 16 alongside trade minister Ryosei Akazawa, with Prime Minister Sanae Takaichi joining by video. That administration has made AI and semiconductors the core of a growth plan chasing $2.3 trillion in investment by 2040.
The irony is hard to miss. Nvidia bills Noetra’s factory as “the world’s first national AI infrastructure,” yet Japan’s push for independence, at least for now, rests on American chips.
What to watch next
This is the same playbook Huang ran weeks earlier, according to TechCrunch AI: a homecoming keynote in Taiwan, then a 50,000-GPU deal in Seoul last fall. Tokyo was next, with the robots, the supply chain, and the chips underneath. “Japan invented modern manufacturing,” Huang said. “Now, it has the opportunity to reinvent it for the age of intelligent industries.”
For anyone tracking where AI spending flows next, the signal is clear. The build-out is shifting from cloud data centers toward robots and production lines, and governments are now writing the checks. Watch whether Noetra’s software timeline holds, whether these robotics commitments turn into shipping products, and whether Japan’s sovereign-AI bet can ever run on something other than Nvidia silicon. Full details are available at the original TechCrunch AI report.