started · updated
Japan's Noetra launches multimodal AI model R&D with Sony, SoftBank, NEC, Honda
Noetra Corp., backed by 44 Japanese companies, has begun full‑scale research and development of a homegrown multimodal foundation model that will underpin AI‑enabled robots and physical AI. Core investors – Sony Group, SoftBank, NEC and Honda – together with partners such as the National Institute of Advanced Industrial Science and Technology (AIST), Preferred Networks and NVIDIA – will provide engineering talent and computing infrastructure. The project will use roughly 27,500 NVIDIA Rubin GPUs, with construction of the AI compute facility slated for April 2027 and operational start in June 2028.
The development roadmap targets a reasoning foundation model with advanced Japanese language understanding, logical reasoning and instruction following by fiscal year 2026, followed by an omni‑modal model capable of processing text, images, video and audio by fiscal 2028. By fiscal 2030, Noetra aims to deliver a “real‑world native AI” that can recognise physical space and act autonomously. The initiative receives up to ¥1 trillion in support from Japan’s Ministry of Economy, Trade and Industry and NEDO, reflecting a national push to lead in physical AI technologies.