《麦肯锡：2025放眼算力之外：AI数据中心供能与冷却基础设施研究报告（英文版）（11页）.pdf》由会员分享，可在线阅读，更多相关《麦肯锡：2025放眼算力之外：AI数据中心供能与冷却基础设施研究报告（英文版）（11页）.pdf（11页珍藏版）》请在三个皮匠报告上搜索。

1、October 2025McKinsey DirectBeyond compute:Infrastructure that powers and cools AI data centers Power and cooling equipment are the backbones of data center infrastructure.Innovations and on-time supply of this technology will become increasingly relevant as the demand for data centers grows.This art

2、icle is a collaborative effort by Ammanuel Zegeye and Pankaj Sachdeva,with Arjita Bhan,Kenza Bouhaj,Rishi Gupta,Vaibhav Chandak,and Wendy Zhu,representing views from McKinseys Industrials&Electronics Practice.As AI use becomes continuously more widespread,the demand for data centers is surging in ta

3、ndem.Data center demand is expected to grow at a CAGR of 22 percent and reach 220 gigawatts by 2030nearly six times larger than demand in 2020 and driven predominantly by AI adoption and hyperscaler spending.1 Moreover,McKinsey research shows that by 2030,data centers are projected to require$6.7 tr

4、illion in cumulative capital outlays worldwide to keep pace with the demand for compute.2 A large part of that capital spending will be allocated to the underlying systems that deliver electricity to IT equipment and the cooling systems that remove the heat generated by the IT equipment.As data cent

5、er designs evolve to keep up with increasing compute requirements,power,cooling,and IT components are no longer seen as separate entities;they must be codesigned and considered in the context of the holistic performance of the data center.More opportunities are opening for stakeholders across the da

6、ta center value chain to design equipment with the overall architecture in mind,either reducing silos or allowing manufacturers to capture vertical integration opportunities to expand their market footprint.Whats more,time to market is now one of the most important buying preferences for data center