At the end of his GTC 2026 keynote address, NVIDIA CEO Jensen Huang did not take his bow alone as usual, but instead walked onto the stage with a special guest—a Disney-designed...Olaf, the robot from FrozenThis adorable robot can not only walk independently, but also engage in playful interactions with Jensen Huang, even complaining in public that "my legs get sore from standing for too long." This scene was not only the grand finale of the speech, but also NVIDIA's global declaration: the AI-driven "physical" revolution has officially arrived.
At the event, Jensen Huang stated, "The era of Physical AI has arrived, and every industrial company will become a robotics company. NVIDIA's full-platform architecture, encompassing computing, open models, and software frameworks, is the cornerstone of the entire robotics industry, connecting the global ecosystem to jointly create intelligent machines that will drive the next generation of factories, logistics, transportation, and infrastructure."
The four major industrial robot manufacturers have fully adopted this approach: from virtual debugging to real-time inference.
The most significant announcement at GTC 2026 was undoubtedly the adoption of NVIDIA technology by the world's four largest industrial robot manufacturers—FANUC, ABB Robotics, YASKAWA, and KUKA. These four companies, which together account for over 200 million industrial robots installed globally, are integrating the NVIDIA Omniverse library and Isaac simulation framework into their virtual commissioning solutions.
This means that future production lines can be fully designed, tested, and optimized in a digital twin environment, significantly reducing the time and cost required for actual deployment. At the same time, these companies are integrating NVIDIA Jetson modules into their controllers to provide production lines with real-time edge AI inference capabilities, enabling industrial robots not only to "move," but also to "think."
The Humanoid Robot Army Takes Shape: From Olaf to Figure, the GR00T Model Continues to Evolve
Humanoid robots remain a visual focus at GTC 2026. Besides Disney's Olaf robot (developed by Disney using the NVIDIA Warp framework and integrated into the Newton physics engine), companies including 1X, AGIBOT, Agility, Boston Dynamics, Figure, and Hexagon Robotics are currently using NVIDIA Cosmos world models, Isaac Sim, and Isaac Lab to accelerate the development and validation of their next-generation humanoid robots.
NVIDIA simultaneously announced several key technology updates:
• Isaac Lab 3.0 has entered early access phase, enabling faster large-scale robot learning on NVIDIA DGX-level infrastructure, and adding support for multiple physics simulations and complex dexterity manipulation.
• The GR00T N1.7 model is available for open commercial licensing, bringing general robotic skills, including advanced dexterity control, to production-ready robot deployments.
• GR00T N2 Early Preview: This next-generation robot model based on DreamZero research adopts a brand-new world action model architecture. It has more than twice the probability of successfully performing new tasks in unfamiliar environments compared to mainstream visual language action models. It is expected to be officially launched by the end of the year.
These systems are all powered by the NVIDIA Jetson Thor robotics platform, enabling developers to move from simulation training to real-world deployment more quickly, intelligently, and reliably.
New Blue Ocean for Medical Robots: CMR Surgical and Medtronic Adopted
The applications of AI in the physical world extend beyond factories, demonstrating immense potential in the medical field. CMR Surgical is using Cosmos-H simulation technology to train and validate its Versius surgical system before clinical deployment. Johnson & Johnson MedTech is employing a post-training workflow based on Isaac Sim and Cosmos to build a training and validation system for its Monarch urology platform. Medtronic is exploring the use of the NVIDIA IGX Thor platform to provide the accuracy and functional safety required for mission-critical surgical robotic systems.
A blueprint for a physical AI data factory: solving the "data hunger" problem in robot training.
NVIDIA announced the NVIDIA Physical AI Data Factory Blueprint, an open reference architecture that unifies and automates the process of training data generation, augmentation, and evaluation, significantly reducing the cost, time, and complexity of training physical AI systems at scale.
At the heart of this blueprint is to leverage NVIDIA Cosmos open-world base models and coding agents to transform limited training data into massive and diverse datasets, particularly capturing rare edge cases and long-tail scenarios in the real world that are extremely costly, time-consuming, and impractical.
Microsoft Azure and Nebius have announced the integration of this open blueprint into their cloud infrastructure and services. Physical AI developers, including FieldAI, Hexagon Robotics, Skild AI, Teradyne Robotics, and Uber, have already begun using this blueprint to accelerate the development of robots, visual AI agents, and self-driving cars.
The entire ecosystem is mobilized: from Foxconn to GXO, real-world applications are flourishing.
The collaboration announced by NVIDIA goes beyond the technical level and has already translated into real-world commercial applications:
• Skild AI collaborates with ABB Robotics and Universal Robots to deploy its general-purpose robotic intelligence across various industries and tasks; it also partners with Foxconn to develop high-precision assembly applications for NVIDIA Blackwell production lines.
• Lightwheel partnered with Samsung to enable its assembly robots to master complex cable handling skills in a simulated environment.
• KION Group, in collaboration with NVIDIA and Accenture, is developing an NVIDIA Jetson-based fleet of automated forklifts for GXO, the world’s largest contract logistics provider.
• PTC announced the launch of a seamless design-to-simulation workflow from the cloud-native CAD platform Onshape to NVIDIA Isaac Sim.
