据路透社消息，百度旗下的AI芯片业务昆仑芯近期已完成估值逼近30亿美元的新一轮融资，目 前正筹备在香港进行首次公开募股(IPO)。投资材料显示，昆仑芯希望最晚在2027年初完成首 次公开发行(IPO)。两位消息人士称，该公司表示其目标是2026年第一季度向香港交易所提交 上市申请。 综合 | 路透社 公司官网 HKET 腾讯新闻 编辑 | Arti 本文仅为信息交流之用，不构成任何交易建议 据接近昆仑芯的投资人表示，早些时候昆仑芯曾准备筹备科创板上市，在与多家券商进行沟通 后，于今年下半年决定转向港股。 官网显示，昆仑芯(北京)科技有限公司前身为百度智能芯片及架构部。公司的技术积累可追溯 至2011年，彼时百度为满足深度学习运算需求，已启动基于FPGA的AI加速器研发，并同步大 规模部署GPU。2018年7月，百度在AI开发者大会上正式推出云端全功能AI芯片"百度昆仑"， 宣告全面进军AI芯片领域。 昆仑芯于2021年4月完成独立融资，领投方为CPE源峰，IDG、君联、元禾璞华等知名机构参与 投资，首轮估值约130亿元，由百度芯片首席架构师欧阳剑出任CEO。 据介绍，欧阳剑是北京航空航天大学学士、中国科学技术 ...

Core Insights - The integration of AI and cloud computing is reshaping the technology and business landscape, marking a new starting point for the industry [2][4] - In 2024, China witnessed the emergence of 50 trillion-level parameter models, leading to a 74.1% year-on-year growth in intelligent computing power, significantly outpacing general computing growth [1] - Baidu's intelligent cloud has become a leader in the AI field, leveraging its comprehensive AI technology stack to drive innovation and transformation in cloud services [4][6] Industry Trends - The cloud computing sector is undergoing unprecedented changes due to the incorporation of AI technologies, which are redefining service architectures and value propositions [1][4] - The competitive landscape in cloud computing is shifting from cost-based competition to a focus on innovative solutions that address specific business challenges [4][6] Company Developments - Baidu has made significant advancements in AI, launching the first deep learning framework "PaddlePaddle" in 2016 and the AI chip "Kunlun" in 2018, establishing a robust foundation for its AI capabilities [4][5] - Baidu's intelligent cloud has achieved a 24.6% market share in China's AI public cloud market, maintaining its position as the leader for six consecutive years [6][7] - The "Cloud Intelligence Integration" strategy of Baidu's intelligent cloud is set to evolve further, aiming to enhance intelligence capabilities and expand its market penetration [6][7] Product Innovations - The Baidu "Baihe" AI computing platform has been upgraded to version 5.0, focusing on enhancing inference capabilities and optimizing resource utilization [6][7] - The "Qianfan" platform has transitioned to version 4.0, providing a one-stop enterprise service platform centered around AI agents, aimed at accelerating application deployment [7] Financial Performance - Baidu's intelligent cloud revenue grew by 27% year-on-year, contributing to the company's AI new business surpassing 10 billion yuan for the first time, with a 34% year-on-year increase [7]

Core Viewpoint - The integration of AI and cloud computing is creating a new paradigm known as "AI cloud," which is driving significant growth in the cloud computing industry and reshaping its competitive landscape [1][2]. Group 1: AI and Cloud Computing Dynamics - In 2024, China is expected to add 50 large models with over 100 billion parameters, leading to a 74.1% year-on-year growth in intelligent computing power, significantly outpacing general computing power growth [1]. - AI is fundamentally altering the architecture and value logic of cloud services, pushing the demand for comprehensive end-to-end solutions rather than just basic AI computing infrastructure [1][9]. - The relationship between AI and cloud computing is redefining the boundaries of technology and business, creating a new starting line for competition in the cloud computing sector [1]. Group 2: Market Trends and Competitive Landscape - In 2020, the Chinese cloud computing market grew by 33.6% to reach 178.1 billion RMB, making it one of the fastest-growing regions globally, despite facing intense price wars and homogenization challenges [5]. - Major cloud providers like Amazon AWS, Microsoft Azure, and Google Cloud engaged in aggressive price cuts, with reductions of up to 60% in some services, intensifying competition [5][8]. - The shift from standardized cloud resources to customized solutions is seen as a potential remedy for the homogenization issue, although it has led to fragmentation challenges [8]. Group 3: Baidu's Strategic Positioning - Baidu has established a complete AI technology stack from hardware to applications, which is a competitive advantage not shared by its rivals [10][11]. - The "cloud-intelligence integration" strategy initiated by Baidu aims to transform the AI development paradigm and lead the evolution of the Chinese cloud computing industry [11]. - Baidu's intelligent cloud revenue grew by 27% year-on-year, with AI new business surpassing 10 billion RMB for the first time, indicating strong market performance [21]. Group 4: Future Outlook and Innovations - Baidu's intelligent cloud is set to upgrade its strategy to focus on intelligent-first AI cloud infrastructure, aiming to enhance the capabilities and penetration of AI technologies [18]. - The Baidu Baige AI heterogeneous computing platform has achieved significant performance improvements, allowing for efficient training and inference of large models [19]. - The introduction of the Agent-based AI development framework in Baidu's Qianfan platform aims to lower development barriers and accelerate application deployment across various industries [20].

Group 1 - Huang Renxun, CEO of Nvidia, emphasizes the importance of global supply chain cooperation during his visit to China for the third Chain Expo, highlighting the interdependence of the global technology ecosystem [1] - The approval for the sale of H20 GPU in China marks a significant development for Nvidia, as the company prepares to launch the new RTX PRO GPU in the market [2] - The rise of AI in China is seen as a critical factor in the ongoing US-China tech competition, with Nvidia playing a vital role in providing the underlying computing power [2] Group 2 - Huang Renxun praises China's open-source AI as a catalyst for global progress, allowing various countries and industries to participate in the AI revolution [3] - The competition between Nvidia and domestic chip manufacturers like Huawei and Baidu is acknowledged, with Huang stating that there is room for collaboration despite competition [2] - The overall AI industry benefits from the competitive landscape, as the presence of multiple players drives down GPU costs and fosters innovation among various stakeholders [2][4]