Japan to buy 27,500 Nvidia Rubin chips for homegrown robot AI

Japan announced plans to purchase 27,500 next-generation Rubin chips from Nvidia through a newly established government-backed company called Noetra, which has been allocated ¥387.3 billion ($2.4 billion（約3800億円）) in public funding through March of next year. The initiative brings together dozens of Japanese companies—including Sony Group, SoftBank, Toyota-backed Preferred Networks, NEC, and Fujitsu—to build a homegrown foundational AI model for robots and construct a roughly 140 megawatt data center to power the effort.

Noetra President Hironobu Tamba, who led the development of SoftBank's Sarashina large language model, framed the venture as an opportunity to consolidate Japan's fragmented AI efforts. The country is home to several competing models: SoftBank's Sarashina LLM, Preferred Networks' PLaMo, and NEC's cotomi. By centralizing these resources under a single national entity, Japan aims to reduce its reliance on foreign technology and strengthen national security. Tamba stated that Japan, as home to some of the world's largest industrial robot makers, may be able to build an alternative to U.S. and Chinese AI systems.

The project timeline is ambitious. Noetra's data center is scheduled to go online in June 2028. The company plans to release an initial AI model by March of next year, with regular updates to follow. Within a few years, Noetra intends to release a model tailored specifically for robotics applications. The broader government goal is to capture more than 30% of the estimated ¥60 trillion global robotics market by 2040. Minister of Economy, Trade and Industry Ryosei Akazawa emphasized that building a robust AI ecosystem is essential for "resource-poor and disaster-prone Japan" and will serve as a lifeline for the nation.

Nvidia co-founder Jensen Huang, present in Tokyo to promote the company's robotics and machine-learning capabilities, connected the initiative to Japan's demographic challenge: a generation of industrial craftspeople—welders, machinists, and others—are retiring without passing their skills to the next generation. Huang framed AI as a mechanism to preserve and transmit these specialized techniques. He also signaled that Japan would need substantially more infrastructure beyond this initial venture, stating, "We're going to be building a lot more infrastructure here. This is just the beginning." Stock prices for Sony, Toyota, Fujitsu, and NEC rose in Tokyo trading on Thursday, bucking a broader market decline.

Japan's Noetra initiative represents a coordinated national effort to consolidate domestic AI capabilities around a shared robotics-focused platform. Rather than allowing companies like SoftBank, Preferred Networks, and NEC to develop competing large language models in isolation, the government is channeling ¥387.3 billion ($2.4 billion（約3800億円）) into a unified venture that will leverage 27,500 Nvidia Rubin chips—a substantial order, though the body notes it is small compared with Microsoft's plans for hundreds of thousands of Vera Rubin systems. This consolidation serves two strategic purposes: lowering Japan's reliance on foreign technology infrastructure and positioning the nation to capture more than 30% of the estimated ¥60 trillion global robotics market by 2040.

The timing reflects both economic and demographic pressures. Japan faces a shortage of skilled craftspeople—welders, machinists, and other industrial workers—retiring without knowledge transfer mechanisms in place. The government and Nvidia co-founder Jensen Huang both frame AI as a tool to preserve and transmit these specialized skills to the next generation. Noetra's roadmap supports this vision: an initial model release by March of next year, followed by regular updates and a robotics-specific variant within a few years. The June 2028 data center launch provides the physical infrastructure foundation. By coordinating these efforts under a single entity led by Hironobu Tamba (who previously led SoftBank's language model development), Noetra aims to create an alternative to U.S. and Chinese AI systems rather than fragmenting investment across competing national champions.

No comments yet. Be the first to share your thoughts!