TLDR
- At COMPUTEX/GTC Taipei on June 1, 2026, Nvidia revealed the Isaac GR00T Reference Humanoid Robot, marking its inaugural open-source humanoid robotics platform.
- Chinese company Unitree Robotics manufactured the physical robot body—the H2 Plus model measures 6 feet in height and tips the scales at 150 pounds.
- Leading academic institutions including Stanford University, ETH Zurich, and UC San Diego are pioneering adopters of this robotics platform.
- Unitree is pursuing a public offering on Shanghai’s STAR Market with fundraising targets between $610 million and $620 million, gaining momentum from the Nvidia collaboration.
- More than 80% of global humanoid robot manufacturing occurs in China, though U.S.-based production is projected to surge significantly within the coming year.
In a strategic move that surprised many industry observers, Nvidia has selected a relatively unknown Chinese robotics company as the hardware foundation for its groundbreaking open humanoid robot platform.
During its GTC Taipei presentation on June 1, 2026, Nvidia introduced the Isaac GR00T Reference Humanoid Robot to the world. The robotics system features the Unitree H2 Plus as its core hardware component, manufactured by Unitree Robotics, headquartered in Hangzhou, China. This humanoid machine boasts impressive specifications: a 6-foot stature, 150-pound weight, and 31 degrees of freedom for movement. The robotic hands, supplied by Singapore’s Sharpa, deliver 25 additional degrees of freedom enabling sophisticated object manipulation. Processing power comes from Nvidia’s Jetson Thor chip integrated with a Blackwell GPU.
During the unveiling, Nvidia CEO Jensen Huang drew a playful comparison, observing that the robot matched his own physical dimensions at six feet tall and 150 pounds. “Just like me,” he remarked.
Understanding the Platform’s Capabilities
The Isaac GR00T ecosystem encompasses the complete robotics development pipeline. This comprehensive system includes data collection, virtual simulation, model training, performance assessment, and real-world implementation. Developers can leverage Isaac Teleop for capturing demonstration datasets, utilize Isaac Sim and Isaac Lab for virtual environment training, and deploy Isaac ROS middleware to transfer trained algorithms onto physical robotic systems.
Additionally, Nvidia introduced open-source foundation models enabling robots to acquire new skills and adjust to unfamiliar settings without requiring manual programming for individual use cases. The complete reference workflow designed for Unitree’s G1 robot model will become publicly accessible through GitHub and Hugging Face repositories.
Nvidia’s strategic objective centers on establishing a universal software infrastructure for humanoid robotics, mirroring how its CUDA platform became the industry standard for artificial intelligence development in previous years.
Strategic Significance for Unitree Robotics
Nvidia’s selection of Unitree represents a calculated decision. The Chinese manufacturer has distinguished itself by producing humanoid robots at significantly lower price points compared to Western manufacturers and has cultivated strong relationships within the research community through accessible hardware offerings.
This partnership elevates Unitree to stand alongside prestigious institutions like Stanford University, ETH Zurich, Ai2, and UC San Diego’s Advanced Robotics and Controls Laboratory as initial adopters of the robotics platform.
The announcement’s timing carries particular weight for Unitree’s financial trajectory. The company has submitted registration documents for a public listing on the Shanghai Stock Exchange’s STAR Market, targeting capital raises between $610 million and $620 million. With Qiming Venture Partners among its investment backers, the Nvidia collaboration provides substantial validation immediately preceding the planned IPO.
Competition in the Humanoid Robotics Arena
This platform launch positions Nvidia in indirect rivalry with Tesla, which maintains its own humanoid robotics ambitions. Tesla has recently halted manufacturing of its Model S and X automobile lines, reallocating that production infrastructure at its Fremont, California facility toward Optimus robot assembly.
Tesla stock experienced a 4.6% decline on the date of Nvidia’s platform announcement. Coincidentally, OpenAI revealed it was actively recruiting personnel for a robotics division on the identical day.
Chinese manufacturers currently account for over 80% of worldwide humanoid robot production volume. Industry analysts anticipate substantial growth in U.S. domestic humanoid robot manufacturing capacity throughout the next 12-month period.
While Nvidia’s data center operations continue generating the majority of company revenue, the June 1 platform introduction clearly indicates where the technology giant is constructing its next major computing ecosystem.
