3 6 Ke·2025-08-26 08:25Core Viewpoint - NVIDIA has launched the new robotic computing platform Jetson Thor, which is referred to as the "robot brain," featuring server-level computing power integrated into robots, significantly enhancing their capabilities [1][6]. Group 1: Technological Advancements - Jetson Thor is based on the latest Blackwell GPU architecture, achieving AI computing power exceeding 2000 TFLOPS, a 7.5 times increase from the previous generation, and equipped with 128GB of memory [1][4]. - The platform allows robots to process data from multiple sources such as cameras and LiDAR, enabling high-speed, multimodal perception and real-time decision-making [4][6]. - Jetson Thor can run multiple AI models simultaneously, providing developers with a flexible environment for creating advanced robotic systems [3][6]. Group 2: Market Impact and Applications - The Jetson AGX Thor developer kit is priced at $3,499 in the U.S. (approximately 25,000 RMB), and several robotics companies, particularly in China, have already adopted it [3][8]. - The platform is expected to enhance the autonomy and decision-making speed of robots across various sectors, including logistics, healthcare, and industrial production [8][12]. - Companies like Yushutech and Boston Dynamics are integrating Jetson Thor into their products, demonstrating its effectiveness in real-world applications [8][10]. Group 3: Industry Perspective - The release of Jetson Thor is seen as a pivotal moment for the robotics industry, potentially marking a new development phase as AI computing power and software ecosystems improve [10][12]. - The current limitation in robotics is attributed to the lack of advanced "brains" capable of governing physical actions, which Jetson Thor aims to address [9][10]. - NVIDIA's investment in Jetson Thor is not just about a single product but about establishing an ecosystem for the intelligent robotics era, which could significantly impact its valuation in the future [12].