Japan’s government, Nvidia, and a newly formed 44-company consortium called Noetra unveiled plans this week to build what they describe as the world’s first national AI infrastructure dedicated to physical AI, according to a joint announcement from Nvidia.
A 140-megawatt AI factory
The planned facility will run on 27,500 Nvidia Rubin GPUs and 13,750 Vera CPUs, built from Vera Rubin NVL72 racks on Nvidia’s DSX reference platform and linked by Spectrum-X Ethernet networking, delivering roughly 140 megawatts of AI computing capacity. Nvidia says the system is designed to train open, multimodal foundation models for robotics, digital twins, and industrial automation using Japan-specific data.
A government-backed consortium
Noetra was founded by SoftBank, Sony, NEC, and Honda, with investment from 44 companies and organizations spanning manufacturing, logistics, and other industries. The project forms the computing backbone of FRONTia, a state-backed initiative run with Japan’s Ministry of Economy, Trade and Industry (METI). METI has said it wants Japan to capture 30% of the global AI robotics market by 2040, an opportunity it estimates at $133 billion.
“Japan invented modern manufacturing. Now, it is building the AI factories” that will drive the next industrial revolution, Nvidia CEO Jensen Huang said. METI Minister Ryosei Akazawa said the project draws on “Japan’s strengths, such as its onsite expertise and manufacturing technology,” while Noetra CEO Hironobu Tamba said the resulting foundation models would be “broadly shared” to speed physical AI deployment nationwide.
Funding and rollout
Nvidia’s announcement did not disclose funding figures, but Noetra and the AIST research institute reportedly won a NEDO government tender on June 30 worth about ¥387.3 billion ($2.4 billion) in first-year funding, with up to ¥1 trillion ($6.1 billion) available through fiscal 2030, according to Nikkei Asia and Tom’s Hardware. Construction is expected to begin in April 2027, with operations starting in June 2028. Noetra’s roadmap targets a Japanese-language reasoning model by fiscal 2027, an omni-modal model handling text, images, video and audio by fiscal 2028, and a “real-world native” model with physical spatial understanding by fiscal 2030.
Part of a broader push
The Japan deal extends Nvidia’s sovereign AI strategy, under which the company has struck similar national-infrastructure agreements with governments including Singapore, Switzerland and India — a strategy analysts estimate has generated more than $30 billion in revenue over the past year, according to TechWireAsia.