NVIDIA Unveils Physical AI Models and Next-Generation Robot Partners at CES 2026

NVIDIA founder Jensen Huang announced a breakthrough "ChatGPT moment for robotics" at CES 2026 by releasing a massive suite of open models and hardware. Global industry leaders including Boston Dynamics and Caterpillar are already using the new physical AI stack to debut next-generation autonomous machines. This shift aims to move robots from single-task automation toward reasoning generalists that can plan and act in unstructured environments. The software foundation rests on new models available through Hugging Face to lower entry barriers for developers. These include NVIDIA Cosmos for synthetic data generation and the GR00T N1.6 model for humanoid robot control. By providing these open-source tools, NVIDIA allows teams to bypass expensive pretraining and focus directly on task-specific robot learning. Hardware capabilities received a significant upgrade with the debut of the NVIDIA Jetson T4000 module. Built on the Blackwell architecture, it delivers 1,200 TFLOPS of compute power with four times the energy efficiency of previous generations. The module is designed specifically for autonomous machines that require high performance within a strict 70-watt power envelope.

To manage complex development cycles, NVIDIA introduced the Isaac Lab-Arena and the OSMO orchestration framework. Isaac Lab-Arena standardizes robot benchmarking by connecting simulation to industry-leading standards like Libero and Robocasa. Meanwhile, OSMO unifies the edge-to-cloud workflow, allowing developers to manage model training and testing from a single command center.

The ecosystem expansion includes a deep integration with the open-source community through the LeRobot framework. This collaboration enables models to run seamlessly on the NVIDIA Jetson Thor robotics computer for platforms like the Reachy 2 humanoid. Consumer giants like LG Electronics are also using these tools to power home robots capable of complex household tasks. Industrial and medical applications are scaling rapidly alongside these consumer advancements. Caterpillar is bringing advanced autonomy to construction and mining sites, while LEM Surgical utilizes NVIDIA Isaac tools to train autonomous surgical arms. These deployments demonstrate that physical AI is moving beyond research labs into critical real-world infrastructure.