Jensen Huang’s Japan deals put Nvidia at the center of the AI factory boom

NVIDIA’s latest push in Japan shows how the company wants to expand beyond the giant cloud providers that drove the first wave of the AI boom. During CEO Jensen Huang’s July visit, Nvidia announced a national AI infrastructure project and deeper work with Japanese manufacturers, robotics companies, and technology groups.

The biggest project centers on Noetra, a Japanese AI company backed by 44 investors, including Sony, SoftBank, NEC, and Honda. Noetra plans to build an Nvidia Vera Rubin AI factory that will support homegrown AI models for robots, factories, and other physical systems. For Nvidia, the project opens another path for long-term infrastructure demand.

Noetra anchors the biggest project

Noetra is developing a Japanese multimodal foundation model under the government-backed FRONTia program. The company has received investment from 44 companies and organizations, including Sony, SoftBank, NEC, and Honda.

NVIDIA is working with Noetra on an AI factory planned with 13,750 Vera CPUs and 27,500 Rubin GPUs, delivering 140 megawatts of data center capacity. Construction is scheduled to start in April 2027, with operations expected to begin in June 2028.

FRONTia runs from fiscal 2026 through fiscal 2030 and focuses on multimodal models handling language, images, video, audio, and physical space. METI and NEDO selected Noetra and AIST to lead the domestic foundation model project.

Japan wants AI on the factory floor

Japan’s strategy focuses heavily on physical AI, a term for systems that can understand and act in the physical world. That includes robots, autonomous machines, digital twins, and industrial systems that use AI to respond to changing conditions rather than following only fixed instructions.

The country has a strong base in manufacturing and robotics, so the goal is to combine that industrial knowledge with newer AI models. Japan’s revised AI Robotics Strategy targets about 10 million AI-enabled robots by 2040 and calls for deployment across 18 areas, including manufacturing and health care.

Noetra is meant to provide part of the software foundation for that push. Its planned models will be developed for Japanese users and businesses, while the Nvidia-powered infrastructure will provide the computing capacity needed to train and run increasingly complex systems.

Huang brought manufacturers into the Nvidia stack

The Noetra project was only one part of Nvidia’s July activity in Japan. Huang also met leaders from Fujitsu, Kawasaki Heavy Industries, FANUC, and Yaskawa, companies with major roles in industrial technology, robotics, and factory automation.

NVIDIA says each is building on its platform in some form. Fujitsu is exploring a collaborative control platform for physical AI, while FANUC, Yaskawa, and Kawasaki are integrating Nvidia technologies into robotics and industrial systems. Other Japanese companies, including Hitachi, Kubota, NEC, and Sony, are also working with Nvidia’s physical AI tools.

That matters because Nvidia is trying to sell more than GPUs. Its physical AI stack includes Cosmos, Isaac, Metropolis, Jetson, and networking products. A factory that adopts several layers of that stack can create demand across chips, systems, networking, and software.

The hardware scale is significant

The Noetra system alone would be a large deployment. NVIDIA says the Vera Rubin AI factory will contain 27,500 Rubin GPUs and 13,750 Vera CPUs. The 140-megawatt facility is designed to support large AI training workloads and models used for robotics, digital twins, and other physical AI applications.

Noetra has also secured substantial government support. Reuters reported that the project is receiving more than 380 billion yen in government funding in its first year. IDC describes the broader FRONTia commitment as 1 trillion yen through 2030, although not all of that amount represents direct purchases from Nvidia.

That distinction is important. NVIDIA has not disclosed the value of its Noetra business, and the project’s government funding should not be treated as NVIDIA revenue. The clearest confirmed information is the planned hardware scale, construction schedule, and Nvidia’s role as a core computing provider.

Growth is moving beyond hyperscalers

NVIDIA’s financial results show why industrial and sovereign AI projects matter. In its reported quarter ended April 26, the company posted record revenue of $81.6 billion. Data Center revenue reached $75.2 billion, up 92% from a year earlier.

Within the Data Center, Nvidia now separates hyperscale customers from a category called ACIE, which covers AI clouds, industrial, and enterprise customers. In the first quarter of fiscal 2027, hyperscale revenue was about $38 billion, while ACIE revenue was about $37 billion and grew 31% from the previous quarter.

Japan fits directly into that second category. National AI infrastructure, industrial computing, and factory automation can help Nvidia reduce its dependence on spending from a small group of giant cloud companies. That does not replace hyperscalers, but it gives the company another source of potential demand.

Japan is meaningful, but not the whole story

The size of Nvidia’s business puts the Japan announcements in perspective. A project with tens of thousands of advanced GPUs is significant, but Nvidia generated more than $81 billion in total revenue in its most recent reported quarter alone. One national program will not determine the company’s growth on its own.

The bigger opportunity is replication. NVIDIA says its AI infrastructure is already deployed across nearly 40 countries, and sovereign revenue grew more than 80% year over year in its latest quarter. Similar national or industrial projects could add up if more governments and companies build their own AI factories.

Japan also gives Nvidia a showcase for physical AI. If major manufacturers successfully use its platform for robotics, inspection, logistics, or factory control, those examples could help Nvidia sell the same technology stack to industrial customers elsewhere.

What Nvidia still has to prove

Noetra’s AI factory is not expected to begin operating until June 2028, so the infrastructure must still be built, supplied, tested, and brought online before it can support national models at scale.

NVIDIA reported $81.6 billion in first-quarter fiscal 2027 revenue, up 85% year over year. Its second-quarter fiscal 2027 results are scheduled for Aug. 26, after that quarter ended July 26.

Japan supports NVIDIA’s case that AI spending is extending into factories, robotics, and national infrastructure. Those projects could broaden demand, but they cannot guarantee Nvidia’s future revenue growth on their own.

TL;DR:

• NVIDIA is working with Noetra on a 140-megawatt AI factory planned around 13,750 Vera CPUs and 27,500 Rubin GPUs for Japan’s physical AI push.
• Noetra is backed by 44 companies and organizations, including core investors Sony, SoftBank, NEC, and Honda, and is developing homegrown multimodal AI models.
• Major Japanese industrial companies, including Fujitsu, FANUC, Kawasaki Heavy Industries, and Yaskawa, are also expanding work with Nvidia’s physical AI platform.
• NVIDIA’s ACIE business, which includes AI cloud, industrial, and enterprise customers, generated about $37 billion in its latest reported quarter and grew 31% sequentially.
• Japan alone will not determine Nvidia’s growth, but the projects show how the company is expanding demand for AI infrastructure beyond hyperscalers into governments and factories.

Don’t forget to follow us for more exclusive content.

If you liked this, you might also like:

• NVIDIA unveils next AI breakthrough at Megaconference
• NVIDIA invests $2 billion in Marvell technology amid AI race
• Nvidia’s CEO Jensen Huang highlights Europe as a major opportunity

This article was made with AI assistance and human editing.