Nvidia wants to be the Android of generalist robotics

Nvidia released a new batch of robotics models, simulators, and peripherals The cost of CES 2026it will make it possible to show the desire of the company to be an unchanging generalist platform roboticsjust as Android became the mobile operating system.

Nvidia’s move to robotics represents a major shift in the industry as AI moves from the cloud to machines that can learn how to think in the environment, with the help of low-cost sensors, advanced simulation, and AI models that can integrate all tasks.

Nvidia revealed more on Monday about its full physical AI ecosystem, including new open-source models that allow robots to think, process, and adapt to many tasks and environments, beyond the narrow bots, all found on Hugging Face.

Those examples include: Cosmos Transfer 2.5 and Cosmos Predict 2.5, two international models for data processing and robot process analysis in comparison; Because of the Cosmos 2, a vision language model (VLM) that allows AI systems to see, understand, and act in the environment; and Isaac GR00T N1.6, his next-gen vision language action (VLA) for humanoid robots. GR00T relies on Cosmos Reason as its brain, and it unlocks the control of the entire body of humanoids to be able to move and hold objects at the same time.

Nvidia also introduced the Isaac Lab-Arena at CES, an open-source testing environment hosted on GitHub that also serves as part of the company’s AI platform, enabling safe testing of robotics capabilities.

The platform promises to solve a major industry problem: As robots learn more complex tasks, from handling objects to laying cable, proving these skills in virtual environments can be expensive, slow, and risky. Isaac Lab-Arena addresses this by combining resources, work experiences, training tools, and establishing benchmarks like Libero, RoboCasa, and RoboTwin, creating a level of cooperation where the industry did not exist.

Supporting the ecosystem is Nvidia OSMO, an open source platform that acts as a connector that connects all the processes from data creation through education on the desktop and the cloud.

And to top it all off, there’s a new Blackwell-powered Jetson T4000 graphics card, the newest member of the Thor family. Nvidia is positioning it as an affordable hardware upgrade that offers 1200 teraflops of AI compute and 64 gigabytes of memory while running smoothly at 40 to 70 watts.

Nvidia is also getting deep collaboration with Hugging Face allowing more people to try out robotics training without the need for expensive equipment or special knowledge. The partnership integrates Nvidia’s Isaac and GR00T technologies into Hugging Face’s LeRobot framework, connecting Nvidia’s 2 million robotics developers with Hugging Face’s 13 million AI developers. open source platform Results 2 humanoid now works directly with Nvidia’s Jetson Thor chip, allowing developers to experiment with different types of AI without being locked into its hardware.

The big picture here is that Nvidia is trying to make the development of robotics more accessible, and it wants to be the basis of hardware and software for selling powerful software, just as Android is not stable for mobile manufacturers.

There are early signs that Nvidia’s strategy is working. Robotics is the fastest growing category at Hugging Face, with Nvidia models leading the charge. Currently robotics companies, from Boston Dynamics and Caterpillar to Franka Robots and NEURA Robotics, are using Nvidia technology.

Follow along with all of TechCrunch Coverage of the annual CES conference here.