据上海市人民政府网站10月14日消息，上海市经济和信息化委员会日前印发《上海市智能终端产业高质 量发展行动方案（2026—2027年）》。 其中提到，围绕手机、计算机、眼镜等各类终端产品需求，加快SoC、CPU等核心芯片布局，覆盖 X86、ARM、RISC—V等三大技术路线。依托服务器算力芯片能力，积极布局端侧GPU芯片发展。支持 3D异构集成关键技术突破，探索新型存储技术路线，支持近存、存内计算技术发展，支撑芯片平台算 力和功耗水平提升。 方正证券指出，端侧AI作为对于算力、存力、连接有着高要求的技术创新，对硬件配置提出高要求， 因此AI终端将优先出现在硬件设计等打磨成熟的现有智能终端体中，如手机、PC、汽车、可穿戴、机 器人等领域。软、硬件的技术突破与生态协同让端侧AI成为可能。 其认为端侧AI行业技术核心两大趋势：一是制程工艺持续升级，从成熟制程向先进制程演进；二是架 构创新加速，存算一体架构使算力提升的同时有效降低功耗。由于端侧AI下游细分品类众多，参考 VerifiedMarketReports数据，预计到2033年全球AISoC市场将达900亿美元，2024-2033年复合增长率 15%。 对于A ...

Core Viewpoint - The company has developed intelligent computing modules that can be applied to AI servers, enabling customers to launch customized SoC array server products for centralized management and allocation of CPU, GPU, and NPU resources, achieving computing power sharing [2] Group 1: Product Development - The intelligent computing modules are designed to assist in the development of customized SoC array server products [2] - The company has already achieved mass production of these computing modules [2] Group 2: Market Engagement - The company supplies computing modules to leading domestic internet platforms [2]

Core Viewpoint - The company, Quectel (移远通信), has announced that its intelligent computing modules can be applied to AI servers, enabling customers to launch customized SoC array server products for centralized management and allocation of CPU, GPU, and NPU resources, achieving computing power sharing [1] Group 1 - The company has already supplied computing modules to leading domestic internet platforms and has achieved mass production [1]