NVIDIA Selects Yunar to Build First Human-like Robot AI Platform, Boosting Investment in 'Physics AI' Strategy

BlockBeats News, June 1st, NVIDIA announced at the Taipei GTC conference that it has selected Chinese robotics company Ubtech as its first humanoid robot AI platform partner for university and research institute sales. This platform will integrate the Ubtech H2 humanoid robot with the NVIDIA Jetson Thor computing platform, the Isaac GR00T humanoid robot AI model, and simulation software, targeting research institutions such as Stanford University and ETH Zurich.

NVIDIA CEO Jensen Huang stated that the platform aims to lower the barrier for humanoid robot development by integrating robot hardware, AI models, data generation, and simulation tools to provide researchers with a plug-and-play development environment. He believes that after generative AI, "physical AI" will be the next stage of artificial intelligence development.

It is understood that the Ubtech H2 robot is approximately 1.8 meters tall, equipped with the NVIDIA Jetson Thor chip and Blackwell GPU, allowing AI models to run directly on the robot. The platform also integrates the dexterous hand system provided by Singaporean robot hand manufacturer Sharpa.

NVIDIA stated that an upgraded version, the H2 Plus robot, is expected to be launched in October this year and will be available for public purchase. The first batch of institutions adopting this platform include the Stanford Robotics Center, ETH Zurich, UC San Diego Advanced Robotics Lab, and the American AI research institute Ai2.

It is worth noting that this collaboration comes as Ubtech is preparing to enter the capital market. According to public information, Ubtech is planning an IPO on the Shanghai Stock Exchange's Sci-Tech innovation board, aiming to raise 4.2 billion RMB. The company disclosed that over 40% of its revenue has come from markets outside China.

Prior to this, Jensen Huang had expressed that robotics is expected to be one of NVIDIA's largest growth opportunities in the next five years, and he expects the "physical AI" market to potentially reach trillions of dollars in the future.