8月21日，中国第一家美股IDC上市公司世纪互联(VNET.US)发布2025年第二季度未经审计的财务业 绩。二季度总营收同比增长22.1%，达到24.3亿元人民币；经调整现金毛利润同比增长34.9%至10.6亿 元；经调整EBITDA同比增长27.7%，达到7.32亿元；经调整现金毛利率从去年同期的39.5%增长至 43.6%；经调整EBITDA利润率由去年同期的28.8%提升至30.1%。 截至6月30日，公司基地型业务运营容量环比增长17.5%，新增约101兆瓦达674兆瓦；上架容量同步环 比提升17.0%，新增74兆瓦达511兆瓦，上架率保持在75.9%。 截至6月底，基地型数据中心在建容量约326兆瓦，近期储备容量374兆瓦，未来长期储备容量约418兆 瓦。城市型IDC业务稳健运营，持续贡献价值。第二季度，城市型业务规模保持稳定，运营容量达 52,131个机柜，上架率微增至63.9%，单机柜每月经常性收入提升至8,915元。 2025年上半年，世纪互联共交付约188兆瓦。在建数据中心8个，其中6个位于环京地区，2个位于长三角 地区。未来12个月计划交付约326兆瓦。强劲的交付规划不仅印证了客户对高品 ...

Core Insights - The company reported a revenue of 3.3 billion yuan for 2024, a year-on-year decrease of 7.89%, and a net loss attributable to shareholders of 63.54 million yuan, with a net profit excluding non-recurring items of -97.01 million yuan [1] - The losses are attributed to increased investments in AIDC transformation, the nascent stage of the digital cloud service business, reduced IT spending from government and industry clients, impairment provisions for certain assets, and decreased fair value changes [1] - The company aims to optimize its business structure and accelerate intelligent upgrades while focusing on the integration of cutting-edge technologies to build an AI industry ecosystem for the future [1] Group 1: AIDC Transformation and AI Product Expansion - In 2024, the company is intensifying its strategic shift from IDC to AIDC, focusing on artificial intelligence and intelligent computing cloud, and accelerating the deployment of computing resources [2] - The company has established an integrated service platform for large model development, AI education, and AIGC creation, leveraging its "West Lake Light" computing foundation [2] - New AI products targeting digital government and various industries, such as AI financial traders and AI digital personnel, are set to be launched to expand the application of artificial intelligence [2] Group 2: Intelligent Driving Strategy and Service System Enhancement - The company plans to deepen its "AI DRIVEN" strategy, enhancing the "computing-power-model-data" service system around the "Wangxin Cloud Engine Insphere AI" foundation [3] - It is developing an efficient and inclusive computing supply system through a nationwide network of intelligent computing centers and scheduling platforms [3] - The platform will support various mainstream models, providing services such as model training and multi-modal generation across multiple scenarios, while also participating in the construction of data open platforms to unlock data asset value [3]