
数据中心能耗预计在三年内翻倍,如何利用端侧人工智能降低能源需求?
数据中心耗电量巨大,预计其能耗将在2024年至2027年间翻倍。相较于在云数据中心处理人工智能,在终端设备上处理人工智能可使单次任务能耗降低99%至99.9%。能源信用交易系统可以激励企业采用更节能的AI。
Kim Lokwon is driven to advance low-power, high-performance AI chips for edge devices. With deep expertise in system semiconductors, he has held key roles at Broadcom, IBM, Cisco Systems and Apple Inc., where he contributed to application processor design and large-scale production, including AI processors. He holds a PhD from the University of California Los Angeles and has authored over 100 patents and 14 journal papers in the Institute of Electrical and Electronics Engineers and Association for Computing Machinery publications.
数据中心耗电量巨大,预计其能耗将在2024年至2027年间翻倍。相较于在云数据中心处理人工智能,在终端设备上处理人工智能可使单次任务能耗降低99%至99.9%。能源信用交易系统可以激励企业采用更节能的AI。
AI is transforming industries while driving up energy demand – on-device AI and an energy credit trading system can combine to ensure a sustainable future.
