Thomas Unise
NVIDIA has announced a comprehensive suite of new open models, frameworks, simulation tools, datasets, and AI infrastructure designed to advance the capabilities of robotics and autonomous vehicles. The company is positioning these technologies as catalysts for a transformative moment in physical AI development, enabling machines to better understand, reason about, and interact with the physical world.
The ChatGPT Moment for Robotics
According to NVIDIA CEO Jensen Huang, “The ChatGPT moment for robotics is here,” highlighting how breakthroughs in physical AI are unlocking entirely new applications. The company’s vision centers on creating “generalist-specialist” robots that combine broad knowledge with deep expertise, making them both versatile and reliable in complex situations.
Key New Models and Frameworks
NVIDIA has released several new open models available on Hugging Face:
- Cosmos Transfer 2.5 and Cosmos Predict 2.5: World models enabling synthetic data generation and robot policy evaluation
- Cosmos Reason 2: A reasoning vision language model (VLM) that helps machines perceive and understand the physical world
- Isaac GR00T N1.6: A vision-language-action model designed for humanoid robots
- Alpamayo models: Reasoning-based models for autonomous vehicles
To support development workflows, NVIDIA introduced:
- Isaac Lab-Arena: An open-source framework for robot policy evaluation and benchmarking
- NVIDIA OSMO: A cloud-native orchestration framework unifying robotic development
Hardware Advancements
On the hardware front, NVIDIA announced the Jetson T4000 module, bringing the Blackwell architecture to robotics at $1,999 per 1,000 units. This module delivers quadruple the performance of the previous generation with 2,070 FP4 TFLOPS and 64GB of memory, all within a 70-watt power envelope ideal for energy-constrained autonomous systems.
Additionally, NVIDIA’s IGX Thor extends robotics capabilities to the industrial edge, offering high-performance AI computing with enterprise software support and functional safety features.
Industry Adoption and Partnerships
NVIDIA’s physical AI technologies are already being adopted across industries:
- LEM Surgical is using NVIDIA Isaac for Healthcare and Cosmos Transfer to train autonomous surgical robots
- Boston Dynamics is launching a new Atlas humanoid running on Jetson Thor
- Franka Robotics is powering dual-arm manipulator robots with GR00T N
- Caterpillar is expanding its collaboration with NVIDIA for AI-powered construction and mining equipment
- Automotive companies including Jaguar Land Rover, Lucid, and Uber are leveraging Alpamayo for autonomous vehicle development
A significant partnership with Hugging Face connects 2 million NVIDIA robotics experts with 13 million Hugging Face AI builders, with NVIDIA’s Isaac and GR00T technologies now built into the Hugging Face LeRobot library.
Evolution of Autonomous Driving
NVIDIA’s automotive VP Ali Kani described autonomous driving as “the first real example of physical AI,” outlining its evolution from perception-only systems to the current “AV3” stage featuring agentic physical AI that reasons, plans, and drives like a capable assistant in the real world. The new Alpamayo models and tools are designed to address these advanced capabilities.
With these announcements, NVIDIA continues to position itself at the forefront of physical AI development, providing the models, tools, and infrastructure needed to create more capable and intelligent autonomous systems across industries.
GIPHY App Key not set. Please check settings