沐曦股份科创板上市，全栈“中国芯”联动AI企业实现“算力集群+行业拓展”全方位合作

Core Viewpoint - The collaboration between SenseTime and Muxi signifies a pivotal shift in China's GPU industry from "catching up" to "running alongside," showcasing confidence in domestic alternatives and the potential for growth in AI computing power [1]. Group 1: Financial Performance - SenseTime's revenue from generative AI grew by 103.1% in 2024, reaching 2.4 billion yuan, maintaining a doubling growth trend for two consecutive fiscal years [3]. - The company's strategic focus on high-performance computing is driven by a rigid demand for computing power [3]. Group 2: Strategic Collaboration - SenseTime and Muxi have established a comprehensive partnership that extends beyond product procurement to long-term strategic agreements [5]. - The collaboration aims to create a "computing cluster + industry expansion" model, integrating supply and demand in the GPU market [4]. Group 3: Technological Development - Both companies are engaged in a model of open co-construction and sharing of underlying technologies, focusing on deep adaptation and joint development of product solutions [6]. - They have developed multiple product solutions, including model adaptations and compatibility certifications for domestic chips, enhancing the training and inference capabilities of AI models [6]. Group 4: Ecosystem Development - The partnership has led to the establishment of the "SenseTime Big Device Computing Mall," facilitating effective resource integration and sharing [7]. - This collaboration enhances the diversity and autonomy of computing power supply while reducing construction costs, supporting the ongoing iteration of AI technology and its industrial applications [7]. Group 5: Market Impact - The partnership has significantly improved Muxi's brand influence and market share by gaining recognition in mainstream demand markets through deep collaboration with industry clients [9]. - The ongoing collaboration is expected to explore new paths and models for AI computing infrastructure, potentially leading to a more open and collaborative domestic AI computing ecosystem [9].