NVIDIA used its GTC Berlin keynote on October 21, 2026 to give robot builders a complete starter kit: a ready-made reference humanoid, a new runtime, a generally available robot-learning framework, and open datasets — all aimed at cutting the time it takes to get a general-purpose robot working from weeks to days. Why it matters: humanoid development has been bottlenecked by every lab building its own body, sensors, and software stack from scratch; a shared reference platform is how NVIDIA wants to speed the whole field up.
Berlin, October 21, 2026
What NVIDIA announced for Isaac
The expansion of the NVIDIA Isaac open robotics platform adds four main pieces. Isaac Lab 3.0, NVIDIA’s open source robot learning framework, is now generally available, with multi-GPU workflows and physics engines including Newton and NVIDIA PhysX. A new open source runtime, Isaac OS, is built on NVIDIA Holoscan and connects AI, sensors, controls, and robot hardware for deterministic, real-time performance. The SimReady Central Library puts more than 2,000 simulation-ready robotics and factory assets on Hugging Face, alongside the Human 1M Dataset — human activities captured with synchronized three-view video and 3D hand tracking, with the full dataset open to Cosmos Coalition partners. NVIDIA also released three open source reference workflows for DGX Station, including post-training NVIDIA Cosmos world models for synthetic data generation using ChatGPT Astra.
Agile ONE SN: a reference humanoid you can build on
The headline hardware is Agile Robots’ Agile ONE SN, a full-stack reference humanoid preconfigured with NVIDIA Isaac GR00T and powered by the NVIDIA Jetson Thor edge AI platform. The Agile ONE S body stands 1.40 meters tall, weighs 40 kilograms, and offers 31 degrees of freedom with a 3-kilogram payload per arm. Its dual WUJI Hand 2 five-finger hands each provide 20 degrees of freedom, 1kHz force control, and tactile sensing. Jetson Thor delivers 2,070 FP4 teraflops of compute and 128GB of unified memory onboard, and a sensing suite combines head and wrist cameras, lidar, and inertial measurement units. Agile Robots also plans to integrate on-robot safety protocols with NVIDIA Halos, NVIDIA’s full-stack safety system. NVIDIA says the integrated package cuts setup time from weeks to days. Agile ONE SN is scheduled to be available in early 2027, aimed at frontier humanoid research, with interest from institutions including Stanford University and ETH Zurich.
Europe’s robotics industry is building on it
NVIDIA named a long list of companies building on the platform: ABB Robotics, Flexion, Hexagon Robotics, Humanoid, KUKA, NEURA Robotics, Sereact, and Universal Robots among them. Announcing the expansion, NVIDIA founder and CEO Jensen Huang said general-purpose robotics is advancing at extraordinary speed and that “the ChatGPT moment for robotics is fast approaching.”
Why it matters
For anyone tracking physical AI, this is NVIDIA applying its AI-factory playbook to robots: own the reference design, the runtime, the training framework, and the data, and let an ecosystem build on top. The honest caveat is timing — the reference humanoid does not arrive until early 2027, and a reference platform only matters if labs actually adopt it. But with Europe’s biggest robotics names already building on Isaac, NVIDIA is positioning itself as the default starting point for the humanoid era. Everything AI, tested — we will be watching whether Agile ONE SN shows up in labs on schedule.
FAQ
What is Agile ONE SN?
It is a reference humanoid robot from Agile Robots, preconfigured with NVIDIA Isaac GR00T software and Jetson Thor compute, designed so researchers can develop and validate humanoid skills on one integrated system instead of assembling their own.
When will Agile ONE SN be available?
NVIDIA says it is scheduled for availability in early 2027. No price was announced.
What is new in Isaac Lab 3.0?
Isaac Lab 3.0 is now generally available as an open source, agent-ready framework for training and refining robot policies from desktop to data center, using multi-GPU workflows and physics engines including Newton and NVIDIA PhysX.
What are the SimReady Central Library and Human 1M Dataset?
Both are open resources on Hugging Face: more than 2,000 simulation-ready robotics and factory assets, and a dataset of human activities with synchronized three-view video and 3D hand tracking for teaching robots by example.
Sources: NVIDIA (GlobeNewswire), Unite.AI