Nvidia said four industrial robot makers are incorporating its simulation tools into their development processes, widening the company's effort to build AI for physical tasks as safety checks remain essential to deployment.
ABB Robotics, FANUC, KUKA and YASKAWA were named in Nvidia's March 16 announcement. They are adding the software to systems engineers use to develop and evaluate robot applications.
Collecting varied real-world examples for robot training is expensive, Nvidia said when it introduced Isaac GR00T N1 on March 18, 2025. The humanoid model arrived alongside simulation and synthetic data tools for training and testing. Nvidia said in March that a newer model, GR00T N1.7, was available in early access.
Nvidia CEO Jensen Huang described the broader ambition in a Hill & Valley Forum interview published May 3, 2025. Huang said most AI systems lack the physical reasoning needed to account for friction and inertia, or to infer what happens to an object after it leaves view. He hoped factories built over the following decade would become highly robotic, partly in response to what he described as a global labor shortage. That was Huang's forecast, rather than a reported result of the partnerships.
Nvidia's work also extends beyond humanoid models. The company said June 7 that Doosan Robotics was incorporating Nvidia robotics frameworks, models, a physics engine and an onboard computer into Doosan's robot software platform.
Putting robots to work around people brings a continuing safety obligation. In guidance released Sept. 21, Nvidia called for checks during deployment covering a robot's hardware, software, AI behavior and operating conditions. Its stated next milestone is the release of GR00T N2 by the end of 2026.
