
Japan's Noetra, a consortium of Sony, SoftBank, NEC, and Honda, is building a domestically developed multimodal AI foundation model specifically designed for robotics and physical AI applications. The project, backed by 44 total investors, will construct a computing infrastructure with approximately 27,500 Nvidia Rubin GPUs starting in April 2027 and operational by June 2028. The effort reflects Japan's push to develop sovereign AI capabilities and strengthen competitiveness in the AI-driven industrial transformation while maintaining data control within the country.
Summaries like this, in your inbox every morning.
Sign up free →What happened
Noetra, backed by Sony, SoftBank, NEC, and Honda as core members, launched full-scale R&D for a Japan-developed multimodal foundation model designed for AI-enabled robots and physical AI. The company has secured investments from 44 organizations total. In partnership with Nvidia, Noetra plans to build AI computing infrastructure equipped with approximately 27,500 Nvidia Rubin GPUs, with construction beginning in April 2027 and operations expected in June 2028.
Why it matters
Japan is developing a domestically controlled sovereign AI foundation model to strengthen the nation's industrial competitiveness in physical AI and robotics. By building the infrastructure and models within Japan, the country aims to secure its competitive advantage while addressing data sovereignty and economic security concerns, particularly as AI becomes critical across manufacturing and other strategic sectors.
What to watch
Noetra's development roadmap spans three phases—a reasoning foundation model by fiscal 2026, an omni-modal model (text, image, video, audio) by fiscal 2028, and a "Real-world Native AI" understanding physical properties by fiscal 2030. Models will be released in stages as R&D progresses and real-world implementation opportunities arise.
Noetra, a newly launched Japanese AI initiative, is undertaking full-scale development of a multimodal foundation model specifically engineered for robots and physical AI systems. The project brings together four major core members—Sony, SoftBank, NEC, and Honda Motor—along with 40 additional investing organizations drawn from manufacturing, enterprise software, and AI research sectors. This broad coalition reflects Japan's ambition to build sovereign AI infrastructure and reduce dependence on overseas models.
The technical strategy centers on building a custom AI computing infrastructure in collaboration with Nvidia. Noetra plans to construct facilities equipped with approximately 27,500 Nvidia Rubin GPUs, the latest graphics processors optimized for agentic AI workloads and large-scale foundation models. Construction begins in April 2027, with full operations commencing in June 2028. Parallel to infrastructure buildout, Noetra has assembled a dedicated R&D organization staffed with engineers seconded from core members, the National Institute of Advanced Industrial Science and Technology (AIST), and Preferred Networks, allowing it to leverage deep expertise in AI model development.
The development timeline is structured in three ambitious phases. Starting in fiscal 2026 (ending March 31, 2027), Noetra will develop a reasoning foundation model serving as the core for AI agents, featuring advanced Japanese language understanding, logical reasoning, and instruction-following capabilities. By fiscal 2028, the roadmap calls for an omni-modal foundation model capable of seamlessly processing text, images, video, and audio data together. Looking further ahead to fiscal 2030, Noetra aims to realize "Real-world Native AI"—systems that understand physical properties such as spatial awareness and are architected for actual deployment in operational environments like factories and robotics applications.
Leadership statements underscore the strategic rationale. Noetra's president Hironobu Tamba framed the initiative as essential for Japan to "become a global leader in physical AI" while strengthening industrial competitiveness and addressing societal challenges. Sony's Hiroki Totoki emphasized that developing multimodal foundation models domestically will accelerate real-world deployment of physical AI and allow Sony to apply insights across entertainment, semiconductors, and other domains. SoftBank's Junichi Miyakawa highlighted data sovereignty: securing Japan's industrial and business data within domestic infrastructure is "essential to strengthening the country's industrial competitiveness." NEC's Takayuki Morita stressed the company's end-to-end AI capabilities and framed the effort as critical for "economic security." Honda's Toshihiro Mibe connected the initiative to the company's founding philosophy and decades of manufacturing expertise, positioning the project as a platform to accelerate real-world deployment in mobility and robotics.
Models developed by Noetra will be released externally in stages as development progresses and real-world implementation opportunities emerge, signaling that the foundation models are intended to become shared infrastructure for the broader Japanese AI ecosystem rather than proprietary assets locked within the consortium.
Noetra's launch represents Japan's coordinated response to global competition in AI foundation models, particularly as the technology becomes essential for robotics and physical AI—areas where Japan has historical manufacturing expertise. The consortium structure, with participation from a major electronics manufacturer (Sony), telecom and infrastructure provider (SoftBank), enterprise IT and systems integrator (NEC), and automotive-robotics leader (Honda), reflects a strategy to combine domain expertise across sectors while maintaining Japanese control over model development and data.
The investment from 44 organizations signals broad industrial backing for the sovereign AI agenda. By concentrating compute infrastructure domestically and building models from scratch rather than relying on foreign platforms, Noetra explicitly addresses economic security concerns—a priority Sony's and NEC's statements emphasize. The phased roadmap (reasoning model by 2027, omni-modal by 2028, real-world-native by 2030) suggests a pragmatic path from language and multimodal understanding toward deployment-ready physical AI, aligned with Japan's strengths in manufacturing systems and robotics.
AI-summarized, only the topics you pick — one digest a day via Email, Slack, or Discord.
Free · takes 30 seconds · unsubscribe anytime
No comments yet. Be the first to share your thoughts!
Log in to join the discussion





Get curated AI news from 200+ sources delivered daily to your inbox. Free to use.
Get Started FreeFree · takes 30 seconds · unsubscribe anytime
1 minute a day. The AI essentials.
200+ sources · Email / LINE / Slack