“全球大模型第一股”登陆港股，谁是背后资本？

Core Viewpoint - The successful IPO of Zhiyuan (2513.HK) marks the first listing of a large model company in the global market, providing a valuation benchmark for the long-underserved large model industry, despite the company's increasing losses due to substantial investments in computing power [1][2][3]. Group 1: IPO Details - Zhiyuan's stock opened at 120 HKD, rose by 13.17% to close at 131.5 HKD, with a total market capitalization of 57.89 billion HKD on its first trading day [1]. - The IPO attracted significant interest, with a subscription amount of nearly 197.8 billion HKD and an oversubscription rate of 1164 times, involving over 206,000 participants [2]. - The company plans to raise approximately 4.348 billion HKD through the issuance of 37.42 million shares at an offer price of 116.2 HKD per share [4]. Group 2: Company Background and Business Model - Zhiyuan, founded in June 2019, is recognized as a leading domestic large language model provider, focusing on enterprise clients through a MaaS (Model as a Service) platform [3][8]. - The revenue structure is primarily derived from localized deployments, which contributed 95.5% of revenue in 2022, decreasing to 84.8% by mid-2025, while cloud API services are rapidly growing [8][9]. - As of mid-2025, Zhiyuan's models supported over 8,000 institutional clients and approximately 80 million devices, with a domestic market focus [9]. Group 3: Financial Performance and Challenges - Zhiyuan's revenue grew from 57.4 million RMB in 2022 to 312 million RMB in 2024, with a compound annual growth rate of 130%, and further increased to 191 million RMB in the first half of 2025, representing a 325% year-on-year growth [10]. - Despite revenue growth, the company reported a net loss that expanded from 144 million RMB in 2022 to 2.958 billion RMB in 2024, with a loss of 2.358 billion RMB in the first half of 2025 [10]. - The company faces intense competition from major tech giants like Alibaba and ByteDance, as well as other independent large model firms [10][11].