A股GPU双雄市值破3000亿 壁仞科技港股上市能否复制狂热

Core Viewpoint - The recent listings of domestic GPU companies, including Moore Threads and Muxi Co., have generated significant market excitement, while Birran Technology is set to become the first domestic GPU stock on the Hong Kong Stock Exchange, despite substantial losses in the AI chip industry [1][2]. Group 1: Company Overview - Birran Technology, founded in 2019, is a provider of general intelligent computing solutions and has recently passed the Hong Kong Stock Exchange hearing [2][3]. - The company has accumulated over 4 billion yuan in revenue from 2022 to the first half of 2025, but has also incurred losses of 60 billion yuan during the same period [1][2]. - The company’s core technology team has backgrounds from Nvidia, HiSilicon, and AMD, enhancing its competitive edge in the market [5][6]. Group 2: Financial Performance - Birran Technology's revenue from 2022 to 2025 is projected to be 49,900 yuan, 62.03 million yuan, 337 million yuan, and 590 million yuan, respectively, with corresponding losses of 14.74 billion yuan, 17.44 billion yuan, 15.38 billion yuan, and 16.01 billion yuan [3][4]. - The total losses over three and a half years amount to 63.57 billion yuan, attributed mainly to non-cash changes in redemption liabilities [3][4]. - The company’s gross margin is expected to decline from 100% in 2022 to 31.9% in the first half of 2025, indicating increasing pressure from competition and potential pricing wars [8]. Group 3: Market Dynamics - The AI chip industry is experiencing a surge in demand for computing power, with the domestic intelligent computing chip market projected to reach 50.4 billion USD by 2025, where Birran Technology anticipates capturing approximately 0.2% market share [2][3]. - The market is highly concentrated, with Nvidia and HiSilicon expected to hold 94.4% of the market share in 2024 [2]. - Despite significant losses, the capital market remains optimistic, as evidenced by the soaring valuations of Moore Threads and Muxi Co. after their listings [4].