点击上方蓝字关注 获得更多深度洞察
Green China
绿色中国要闻
CN
算力要增，碳排放要降
最近谷歌母公司Alphabet公布的2023年ESG报告吸引了外界对于人工智能潮流对气候变化影响的关注。据报告，因在数据中心内训练和运行人工智能大模型耗电巨大，Alphabet去年的温室气体排放量比起2019年的基线上升了48%，比2022年同比增长13%。
在中国，数据中心的耗电量占全社会耗电量的约3%。人工智能和可持续发展是否是一项“单选题”？作为在两个领域内都立下远大目标的大国，中国国家发展和改革委近日发布《数据中心绿色低碳发展专项行动计划》，以在加速抢占人工智能高地的同时完成“双碳”目标。《行动计划》要求，到2025 年底，全国数据中心平均电能利用效率要降至1.5以下，新建及改扩建大型和超大型数据中心电能利用效率降至1.25以内。
除了布置一系列节能和可再生能源利用的工作，《行动计划》重申要强化“东数西算”规划布局刚性约束。根据“东数西算”战略要求，国内的非实时算力需求应逐渐向土地资源和可再生能源丰富的西部地区转移，而东部的算力应集中在京津冀、长三角、大湾区等枢纽节点。《行动计划》要求，到2025年底，国家枢纽节点地区各类新增算力占全国新增算力的60%以上。
《行动计划》要求，加快推进低效数据中心节能降碳改造，并鼓励各地方政府结合实际对新上数据中心项目能效提出更高要求。同时，项目资金和税收优惠等措施也应尽快充分落地，支持各类数据中心节能降碳项目。
联系我们
如需了解更多可持续发展相关政策监测、宏观概览、专题分析或趋势展望，欢迎邮件联系 beijing@sandpipercomms.com咨询我司相关服务。
EN
China is trying to prevent an AI-driven carbon emission surge
Alphabet’s recent ESG report has highlighted the impact that the rise of AI could have on the climate. The company announced that it had recorded a 48 per cent increase in greenhouse gas emissions since 2019, among which a 13 per cent increase in the past year alone. This is mainly attributed to the electricity consumption of data centres for AI model training and operation.
China, a nation with ambitious objectives in both AI and sustainability, has recently unveiled an action plan aimed at accelerating the low-carbon transformation of its data centres. These facilities currently consume more than 3 per cent of the country’s electricity. The objective is to achieve an average power usage effectiveness (PUE) below 1.5 by the end of 2025, which means more than two thirds of the total amount of energy used should be spent directly on computing tasks. The PUE should reach at least 1.25 for new, large-scale and ultra-large-scale data centre projects.
In addition to a series of instructions on energy conservation and renewable energy use in data centres, the Action Plan emphasises the necessity of implementing a national strategy known as “Eastern Data - Western Computing”. This strategy involves transferring non time-sensitive storage and computing tasks to more spacious and resource-rich western regions, where wind, solar, and hydraulic power resources are exceptionally abundant. Necessary computing capacities on the east coast should be concentrated in a few “nodes” in the Jing-Jin-Ji region, Yangtze River Delta, and Greater Bay Area.
To reduce emissions and encourage sustainability in the burgeoning AI and data centre industry, local governments must gradually eliminate low-efficiency data centres and enforce even stricter energy efficiency standards for new data centre projects than the compulsory minimum. In the meantime, it is crucial that authorities take immediate action to ensure the full implementation of funding resources and tax incentives for low-carbon data centre projects, in order to expedite the desired outcomes.
Contact us
For more sustainability policy monitor, insights, thematic analyses or industry trend outlooks, please contact beijing@sandpipercomms.com by email to learn about the services we provide.
相关推荐 Related Articles
01 |
02 State Council issued a reminder of industrial decarbonization |
03 |
Sandpiper 基誉公关
Sandpiper is a strategic communications and public affairs consultancy that supports clients in understanding China’s dynamic policy environment through monitoring, reports, and timely alerts. For any business inquiries about stakeholder and issue mapping, expanding government outreach, and cultivating media relations, or any other services, please email beijing@sandpipercomms.com.
基誉公关是一家战略传播和公共事务咨询公司。我们通过政策监测、解读和预警，帮助客户实时追踪、理解中国的政策动态和市场环境。如需了解利益相关方和议题规划、扩展政府人脉、媒体关系，或者有任何其他业务咨询，请发送邮件至beijing@sandpipercomms.com。
This material has been prepared by Sandpiper, a strategic communications and public affairs consultancy, in the belief that it is fair and not misleading. The information upon which this material is based was obtained from sources believed to be reliable, but has not been independently verified; we do not guarantee its accuracy. This communication is being provided for informational purposes only.
