量子计算、机器人……黄仁勋勾勒AI宏伟蓝图

Core Viewpoint - NVIDIA's GTC conference highlighted its expansive vision for AI and related technologies, showcasing partnerships and innovations that position the company at the forefront of the AI and telecommunications sectors [2][4]. Group 1: NVIDIA's Technological Advancements - NVIDIA's CEO Jensen Huang presented a roadmap extending to 2028, revealing that the company has accumulated $500 billion in orders for its Blackwell and Rubin chips by the end of the 2026 fiscal year [4][19]. - The introduction of the NVIDIA ARC (Aerial RAN Computer) aims to capture opportunities in the AI-RAN market, in collaboration with Nokia, which includes a $1 billion equity investment [6]. - NVIDIA's new BlueField-4 data processing chip is designed to accelerate AI infrastructure, supporting throughput of 800 Gb/s and high-performance inference processing [17]. Group 2: Partnerships and Collaborations - NVIDIA announced partnerships with T-Mobile and Cisco to develop the first AI-native wireless stack for 6G in the U.S. [9]. - A collaboration with the U.S. Department of Energy will see the construction of seven supercomputers, including the Solstice system, which will be the largest AI supercomputer for scientific discovery [11]. Group 3: AI Applications and Innovations - The AI traffic is experiencing explosive growth, with ChatGPT having 800 million active users, nearly half of whom access it via mobile devices [8]. - NVIDIA's Omniverse digital twin technology is being utilized to build modern factories and train robots, with a partnership with the startup Figure to accelerate humanoid robot development [26]. - The introduction of the IGX Thor edge AI platform offers 8 times the AI computing power compared to its predecessor, enabling seamless operation of large language models at the edge [29]. Group 4: Automotive and Autonomous Driving - The launch of the NVIDIA DRIVE AGX Hyperion 10 platform is aimed at L4-level autonomous driving, featuring a comprehensive sensor architecture [31]. - A partnership with Uber will allow vehicles using this platform to connect to the Uber network [33].