Workflow
将AI工作负载从云端转移到本地
icon
Search documents
千兆瓦级数据中心难持续?Arm CEO:将AI工作负载迁到本地是方向
Sou Hu Cai Jing· 2025-10-16 07:55
Core Viewpoint - Arm Holdings CEO Rene Haas emphasized the importance of moving some AI functionalities from the cloud to local applications to reduce energy consumption, suggesting that the operation of large data centers is not sustainable in the long term [1][3]. Group 1: AI Functionality and Energy Consumption - Haas proposed two approaches to address energy consumption: adopting low-power solutions in the cloud and transferring AI workloads to local applications [3]. - He noted that while AI training may still require cloud resources, the inference process can be executed locally on devices such as smartphones and computers, indicating a shift towards a hybrid model in computing [3]. Group 2: Collaboration with Meta - Arm announced an expanded collaboration with Meta, focusing on enhancing AI efficiency across all layers of computing, including AI software and data center infrastructure [3][4]. - The partnership also involves Arm's technology in Meta's new Ray-Ban Wayfarer smart glasses, where AI processing occurs locally rather than in the cloud [4]. Group 3: Market Reaction - Following the announcement of the collaboration with Meta, Arm's stock price increased, closing with a gain of 1.49% on the day of the news [3].