NVIDIA Launches Next-Gen Robotics Chip Jetson Thor with 7.5x AI Performance Boost!
-
Recently, NVIDIA officially launched its next-generation robotics-specific chip, Jetson Thor. This chip is designed to significantly enhance AI computing power in the robotics field, better supporting various forms of intelligent robots, especially humanoid robots.
Compared to the previous Jetson Orin, Jetson Thor delivers a staggering 7.5x improvement in computing performance, reaching up to 2070 FP4 TFLOPS with 130W power consumption and 3.5x better energy efficiency. Additionally, Thor's memory capacity has been significantly increased to 128GB, with a memory bandwidth of 273GB/s.
The Jetson Thor chip features the latest Blackwell architecture GPU, specifically designed for generative AI model inference, enabling support for next-generation "physical AI" agents. These agents can operate in real-time on the edge, minimizing reliance on cloud computing. To meet the low-latency and high-performance demands of real-time applications, Jetson Thor is equipped with a suite of powerful software tools, supporting all mainstream generative AI frameworks and inference models, allowing developers to quickly conduct model experiments and inference locally.
NVIDIA also stated that, with optimizations in FP4 precision and speculative decoding, Jetson Thor's performance is expected to further improve. Moreover, the chip supports the full NVIDIA AI software stack, accelerating nearly all physical AI workflows across robotics, video analytics, sensor processing, and other fields. The newly released Jetson Thor products include developer kits and production-grade modules, with the developer kit starting at $3,499 (approximately ¥25,000) and the module priced at $2,999 (approximately ¥21,400).
At the recently concluded World Robot Conference, several leading Chinese robotics companies, such as Unitree Robotics and Galaxy General, announced they would be among the first to adopt the Jetson Thor chip. Galaxy General's Galbot robot showcased its applications across various industrial scenarios at the conference. This new chip has not only garnered industry attention but also encouraged active participation from hardware and sensor companies in building Jetson Thor-based systems.
Additionally, NVIDIA's autonomous vehicle-specific chip, Nvidia Drive AGX Thor, will soon be available for pre-order. As AI technology advances, NVIDIA is not only committed to providing robust computing foundations but also driving new research breakthroughs, solidifying its core position in the AI field. Moving forward, NVIDIA will continue to explore vast market opportunities in robotics, autonomous driving, and beyond, supporting industry growth.
Key Highlights:
AI Performance Boost: The Jetson Thor chip delivers a 7.5x improvement in AI computing power over its predecessor, meeting the demands of various intelligent robots.
Memory & Efficiency: The Thor chip features 128GB memory capacity and 273GB/s memory bandwidth, with exceptional energy efficiency.
Industry Collaboration: Multiple domestic robotics companies have announced plans to adopt the new chip, driving the integration and application of AI and robotics technologies.