Priced at $3,499, Nvidia's Jetson Thor Enables real-time Intelligent Interaction Between Robots and the Physical world.

Jensen Huang once publicly stated,Robotics is Nvidia's biggest growth opportunity outside of artificial intelligence.At CES 2025, which opened in early January 2025, Huang even shouted, "The next frontier of AI is physics." Just like when he bet on OpenAI, his choice now is the robotics track led by Physical AI.

On Monday (August 25th) local time, NVIDIA announced that the Jetson AGX Thor development kit is officially on the market with a starting price of US$3,499. The mass production module Thor T5000 is also now available to corporate customers.

Nvidia calls the Jetson AGX Thor the "robot brain," aiming to power millions of robots across industries such as manufacturing, logistics, transportation, healthcare, agriculture, and retail. Initial users include Agility Robotics, Amazon Robotics, Boston Dynamics, Caterpillar, Figure, Hexagon, Medtronic, and Meta. 1X, John Deere, OpenAI, and Physical Intelligence are also evaluating the platform.

“We built Jetson Thor for the millions of developers who are developing embodied systems that are interacting with the physical world and reshaping our lives,” Jensen Huang said at the launch event. “With its unparalleled performance and energy efficiency, and the ability to run multiple generative AI models at the edge, Jetson Thor will be the ultimate supercomputer driving the era of physical AI and general-purpose robotics.”

7.5 Times the Computing Power + 3.5 Times the Energy Efficiency: A Key Breakthrough in Physical AI

Jetson AGX Thor is equipped with NVIDIA's latest Blackwell GPU architecture and up to 128GB of memory, capable of providing 2070 FP4 TFLOPS of AI computing power at 130 watts of power consumption. Compared with the previous generation Jetson Orin,Thor has increased computing power by 7.5 times and energy efficiency by 3.5 times.It has significantly broken through the performance bottleneck of robots running generative AI models.

This means robots can run not only the mainstream Large Language Model (LLM), but also the Visual Language Model (VLM) and the Isaac GR00T N1.5, a humanoid robot foundation model, enabling real-time understanding and reasoning about the physical world. NVIDIA emphasizes that this performance advantage enables Jetson Thor to support multiple AI workflows, enabling real-time, intelligent interaction between robots and humans and the real environment.

Building a Physical AI Supercomputer

Jetson Thor is more than just a hardware upgrade; it also carries the complete NVIDIA Jetson software stack, fully supports mainstream AI frameworks and generative AI models, and is fully compatible with NVIDIA's software ecosystem from cloud to edge.

include:

* NVIDIA Isaac: robotics simulation and development platform;

* Isaac GR00T: humanoid robot base model;

NVIDIA Metropolis: Visual AI platform.

* NVIDIA Holoscan: Real-time sensor data processing.

Since its launch in 2014, the Jetson platform has attracted over 2 million developers and over 150 ecosystem partners. Jetson Orin has been used by over 7,000 customers for edge AI deployments. With the launch of Jetson Thor, NVIDIA aims to further promote the implementation of complex systems such as visual AI agents, humanoid robots, and surgical robots.

At the same time, NVIDIA also released a series of new world AI models, libraries and other infrastructure for robot developers, such as Cosmos-Reason1-7B, a 7-billion-parameter inference-based visual language model designed for Physical AI applications and robots; Cosmos Transfer-2 can accelerate the generation of synthetic data based on 3D simulation scenes or spatial control inputs; and a distilled version of Cosmos Transfer with further optimized running speed.

Get high-quality papers and in-depth interpretation articles in the field of AI4S from 2023 to 2024 with one click⬇️