- Japan is backing Noetra’s physical AI project to develop domestic foundation models for robotics.
- Noetra plans a 140MW AI facility as Japan targets 10 million AI-enabled robots by 2040.
Japan is providing more than ¥380 billion ($2.33 billion) in first-year funding to Noetra, a government-backed company developing foundation models for physical AI and robotics.
The Ministry of Economy, Trade and Industry has allocated ¥387.3 billion to the programme for the current fiscal year. Published reports have put total government support at up to ¥1 trillion through fiscal 2030, although the full funding ceiling has not been confirmed in the programme’s public documents.
Noetra is backed by 44 companies and organisations, with SoftBank, Sony, NEC, and Honda among its core participants. The consortium also includes manufacturers, technology companies, research institutions, and AI developers.
The company plans to give Japanese organisations access to domestically developed models while allowing them to manage their own data. Noetra has not disclosed its licensing terms or how companies will access the models.
Noetra chief executive Hironobu Tamba described the project as an effort to strengthen Japan’s domestic AI capabilities. Tamba is also an executive at SoftBank.
“We’ve been able to bring together people who have been working on AI in Japan,” Tamba said. “Honestly, I think this may be our last chance.”
“Bringing physical AI into the real world requires enormous computing, data and foundational technologies — challenges no single company can solve alone,” Tamba said.
Noetra plans staged model development
Noetra is procuring 27,500 Nvidia Rubin GPUs for its main computing facility. Construction is scheduled to begin in April 2027, with operations expected to start in June 2028.
The systems will be installed at the FRONTia data centre, which is planned to provide 140 megawatts of capacity. The facility will also use 13,750 Nvidia Vera CPUs and Spectrum-X Ethernet networking.
Noetra plans to begin developing its models before FRONTia becomes operational, using computing infrastructure run by Japan-based providers. The company has not identified all of the providers or disclosed how much capacity will be available during this phase.
Rubin is Nvidia’s planned generation of data-centre processors for AI training and inference. Noetra intends to use the infrastructure to develop models for robotics, AI agents, and digital twins.
The company plans to begin developing a reasoning foundation model during fiscal 2026. The model will support Japanese-language understanding, logical reasoning, and instruction following.
An omni-modal model capable of processing text, images, video, and audio is planned for fiscal 2028. By fiscal 2030, Noetra aims to develop what it calls “Real-world Native AI,” designed to process spatial information and physical properties.
The roadmap begins with language and reasoning before moving into models designed for robotics and physical environments. Noetra has not disclosed the planned model sizes, architectures, or training datasets.
Nvidia said the programme would combine Japan’s manufacturing experience and industrial data with AI computing infrastructure. Noetra has not said whether participating companies will contribute proprietary data from factories, machines, or other operating environments.
The consortium includes companies from automotive manufacturing, electronics, telecommunications, and industrial technology. Noetra is also working with the National Institute of Advanced Industrial Science and Technology, Preferred Networks, and engineers seconded from participating organisations.
Noetra said its models would be made available externally in stages as development and testing progress. The company has not specified the licensing model or whether access will be provided through cloud platforms, APIs, or dedicated infrastructure.
The programme is intended to expand Japanese participation in model development, industrial data management, and deployment. FRONTia will nevertheless rely on Nvidia’s Rubin GPUs, Vera CPUs, Spectrum-X networking, and data-centre platform architecture.
Japan will therefore develop and operate the associated foundation models domestically while sourcing FRONTia’s main computing systems from Nvidia.
Government targets 10 million AI-enabled robots
Prime Minister Sanae Takaichi’s administration is supporting Noetra as part of a wider programme covering AI, semiconductors, robotics, and advanced manufacturing.
The government aims to deploy about 10 million AI-enabled robots across 18 fields by 2040. Target sectors include manufacturing, shipbuilding, logistics, healthcare, nursing, food production, and food service.
Japan’s AI robotics strategy also targets more than 30% of the global AI robotics market by 2040. METI has not disclosed how that market share will be measured.
METI has linked the strategy to structural labour shortages and plans to support AI-robot adoption across the 18 sectors. The programme is focused on industries that depend heavily on physical work and face constraints in recruiting workers.
The project also draws on Japan’s industrial robotics and factory-automation sector. Noetra’s participating companies include manufacturers and technology groups with experience in electronics, automotive production, industrial systems, and physical machinery.
Noetra has not detailed how its foundation models will be integrated into commercial robots or adapted for individual sectors. Japan’s broader robotics strategy separately calls for stronger robot manufacturers, systems integrators, computing infrastructure, and adoption support.
The government is also backing Rapidus, a domestic semiconductor foundry venture that plans to manufacture advanced chips in Japan. Rapidus and Noetra form separate parts of Japan’s efforts to expand domestic capabilities in semiconductors, AI models, and industrial applications.
Nvidia chief executive Jensen Huang said Japan was building AI factories intended to support its industries and economy. FRONTia will use Nvidia hardware and networking, while Noetra will direct development of the associated foundation models in Japan.
Noetra is organised as a government-supported industrial consortium rather than a standalone AI laboratory. Its programme combines government project funding, investment from 44 companies and organisations, shared computing infrastructure, and a published development roadmap through fiscal 2030.
Reuters reported that Japan is seeking greater technological self-sufficiency while US and Chinese companies remain prominent in advanced AI models and infrastructure. Noetra’s programme centres on shared computing resources and models intended for robotics, industrial systems, AI agents, and digital twins.
The company plans to begin developing its reasoning model during fiscal 2026. Its omni-modal model is targeted for fiscal 2028, followed by its physical-world model in fiscal 2030, while FRONTia is scheduled to begin operating in June 2028.
Want to learn more about AI and big data from industry leaders? Check out AI & Big Data Expo taking place in Amsterdam, California, and London. The comprehensive event is part of TechEx and is co-located with other leading technology events, click here for more information.
Tech Wire Asia is powered by TechForge Media. Explore other upcoming enterprise technology events and webinars here.
The post Japan backs Noetra with $2.3 billion for physical AI push appeared first on TechWire Asia.