NVIDIA Launches Jetson Thor Robotics Computing Platform with 7.5x Performance Boost, Priced at 25,000 RMB
-
NVIDIA has officially launched its new Jetson Thor robotics computing platform, which adopts the latest Blackwell GPU architecture, delivering 2070 TFLOPS of AI computing power—a 7.5x improvement over the previous Jetson Orin—while also achieving 3.5x better energy efficiency.
It is reported that Jetson Thor comes with a massive 128GB memory, making it a top-tier configuration among edge computing devices. The platform is designed to provide global developers with powerful tools for building physical AI robotic systems, facilitating deeper interaction between robots and the real world.
At the product launch event, NVIDIA CEO Jensen Huang stated that Jetson Thor can run multiple AI models simultaneously and supports mainstream generative AI frameworks, including Qwen and DeepSeek. He emphasized that this platform marks the official arrival of the era of physical AI and general-purpose robotics.
In terms of hardware specifications, Jetson Thor's GPU features up to 2560 CUDA cores and 9 fifth-generation Tensor Cores, supporting multi-instance GPU technology for efficient multitasking in complex scenarios. The processor utilizes a 14-core Arm Neoverse V3AE architecture, specifically designed for real-time control and task management, ensuring stable robot operation.
On the software side, the platform natively integrates the NVIDIA Isaac simulation platform, providing developers with a unified development environment from cloud to edge. Through FP4 quantization technology, the performance of certain AI models can be further doubled, effectively supporting real-time human-robot dialogue functionality.
It is reported that several prominent Chinese companies, including Unitree Robotics and Boston Dynamics, have become the first adopters of Jetson Thor. These companies have provided feedback indicating that the platform's powerful computing capabilities and flexibility significantly enhance robots' navigation and interaction abilities in real-world environments.
Deepu Talla, NVIDIA's Vice President of Robotics and Edge AI, pointed out that the true advancement of physical AI and robotics requires the synergy of training, simulation, and deployment. Jetson Thor is designed as a comprehensive solution based on this philosophy.
Regarding pricing, the Jetson AGX THOR developer kit is priced at $3,499 in the U.S. market, approximately 25,000 RMB. NVIDIA stated that the specific pricing for the Chinese market will be announced separately.
Industry experts believe that the launch of Jetson Thor will further lower the barrier to robotics development, accelerating the scaling and practical application of the robotics industry. With the breakthrough in computing power, robotic technology is expected to achieve commercial applications in more scenarios.