NVIDIA and Japan: Full AI and Robotics Stack for All Industries

This week, NVIDIA and its partners in Japan are showcasing the latest advances in the AI ecosystem. NVIDIA founder and CEO Jensen Huang visited the Build-a-Claw event, where developers tested physical AI—using open models and the NVIDIA platform to create robots capable of grasping objects. Huang awarded two autographed copies of the NVIDIA DGX Spark supercomputer to giveaway winners and stated that "you can have your own personal AI" and build your own robot agents. At a press conference, he noted that combining precision manufacturing with AI gives Japan an opportunity to create robots and address labor shortages. At a Tokyo restaurant, Huang met with leaders of 16 leading supply chain companies, including Advantest, Tokyo Electron, and Panasonic. Over lunch with the CEOs of Fujitsu, Kawasaki Heavy Industries, FANUC, and Yaskawa, they discussed integrating physical AI into manufacturing processes. At the launch of the government initiative Physical AI Initiative, with Minister of Economy Ryosei Akazawa in attendance, Huang announced the development of open multimodal foundation models for digital twins, robots, and physical AI. SoftBank confirmed the use of the full NVIDIA stack—from GB200 infrastructure to large telecom models based on Nemotron—to transform the network into an intelligent one. In collaboration with SB Intuitions and Yaskawa, open models and physical AI will be further developed. In healthcare, Japanese companies—including Astellas, Ono Pharmaceuticals, Daiichi Sankyo, and Xeureka—are using NVIDIA BioNeMo to accelerate drug discovery. SyntheticGestalt announced molecular AI models ZAO and KOYA. Biomy is building a virtual cell foundation model. Takeda is deploying BoltzMol-1 and BoltzProt-1 models. Kawasaki Heavy Industries plans to use NVIDIA Holoscan IGX, Isaac GR00T, and Cosmos to develop surgical and hospital robots. Canon and Fujifilm have introduced new CT systems based on NVIDIA GPUs.