国盛证券研报指出，蓝思科技收购服务器业务公司，加码AI算力核心布局。通过此次收购，公司将快 速获得国内外特定客户服务器机柜业务（涵盖机架、滑轨、托盘、Busbar等组件）的成熟技术及客户认 证，以及先进液冷散热系统集成能力。这将与公司自身强大的精密结构件与"材料-模组-整机"垂直整合 能力形成深度互补协同，使得公司能够提供从核心部件到散热系统的更完整AI硬件解决方案，大幅增 强核心竞争力，为公司开拓AI算力基础设施新赛道、加速向全球AI硬件创新平台转型注入强劲动能。 公司作为智能终端全产业链一站式精密制造解决方案提供商，一方面具备以结构件及功能模组核心技术 为支撑的产业化能力体系，另一方面具备全产业链垂直整合能力，未来将深度受益AI技术发展，各条 业务线预计全面增长。维持公司"买入"评级。 ...

蓝思科技(300433)拟收购液冷和服务器机柜公司裴美高国际（PMG International Co.,LTD.）100%股权， 开拓AI算力基础设施新赛道。 12月10日晚间，蓝思科技公告称，公司于2025年12月10日与吕松寿（LEU, SONG-SHOW）签订股权收 购意向协议，拟以现金及其他合法方式购买其拥有的裴美高国际100%股权。 具体交易金额、交易方案等将由相关各方根据尽职调查、审计及评估结果等进一步协商谈判，并由交易 双方在正式收购协议中确定。 通过本项收购，蓝思科技将快速获得国内外特定客户服务器机柜业务（包括但不限于机架、滑轨、托 盘、Busbar、其他组件）的成熟技术与客户认证，以及先进液冷散热系统集成能力。 蓝思科技表示，此次收购将与公司较强的精密结构件制造及"材料—模组—整机"垂直整合能力形成强力 互补和高度协同，大幅增强在AI算力硬件方案的核心竞争力，为蓝思科技开拓AI算力基础设施新赛 道、加速向全球AI硬件创新平台转型增添强劲动力。 作为智能设备制造的明星企业，蓝思科技今年频频接受机构投资者调研。在9月的一场投资者调研上， 蓝思科技就表示，当前算力是AI发展的基础，公司正全力拥抱 ...

12月10日晚间，蓝思科技(300433)发布公告，公司与吕松寿LEU, SONG-SHOW签订了《股权收购意向 协议》，拟以现金及其他合法方式购买其持有的PMG International Co.,LTD.（裴美高国际有限公司） 100%股权。具体交易金额、交易方案等将由相关各方根据尽职调查、审计及评估结果等进一步协商谈 判，具体交易金额及收购方式将由交易双方在正式收购协议中确定。 蓝思科技此前在投资者调研活动中表示，已开始逐步布局算力基础设施赛道。公司与北美AI核心硬件 客户深度配合，从2024年开始启动AI服务器相关产品开发，计划从机箱结构件（含导轨、托盘等）加 工开始，逐步向液冷模组、固态硬盘等组装领域拓展。目前，国内外头部大型服务器客户均在同步开拓 中，其中机箱结构件已批量出货，固态硬盘组装预计近期通过客户验证，明年实现规模量产。 数据显示，蓝思科技近年来业绩持续增长，2022—2024年营业收入及归母净利润均实现双位数增长， 2025年前三季度实现营业收入536.63亿元，同比增长16.08%，实现归母净利润28.43亿元，同比增长 19.91%。 （文章来源：证券时报网） 公司提示，本次签订的股 ...

通过本项收购，蓝思科技将快速获得国内外特定客户服务器机柜业务（包括但不限于机架、滑轨、托 盘、Busbar、其他组件）的成熟技术与客户认证，以及先进液冷散热系统集成能力。 蓝思科技表示，此次收购将与公司较强的精密结构件制造及"材料—模组—整机"垂直整合能力形成强力 互补和高度协同，大幅增强在AI算力硬件方案的核心竞争力，为蓝思科技开拓AI算力基础设施新赛 道、加速向全球AI硬件创新平台转型增添强劲动力。 作为智能设备制造的明星企业，蓝思科技今年频频接受机构投资者调研。在9月的一场投资者调研上， 蓝思科技就表示，当前算力是AI发展的基础，公司正全力拥抱AI领域。 12月10日晚间，蓝思科技公告称，公司于2025年12月10日与吕松寿（LEU, SONG-SHOW）签订股权收 购意向协议，拟以现金及其他合法方式购买其拥有的裴美高国际100%股权。 具体交易金额、交易方案等将由相关各方根据尽职调查、审计及评估结果等进一步协商谈判，并由交易 双方在正式收购协议中确定。 在11月最新投资者调研活动上，蓝思科技高管也表示，在AI算力基础设施领域，蓝思科技从AI服务器 机箱结构件起步，逐步拓展至液冷模组、固态硬盘等关键模组 ...

Core Viewpoint - The consumer electronics sector continues to show strong growth, particularly in sub-segments such as AIPC, AI smartphones, AI glasses, and humanoid robots, with significant stock performance noted in related ETFs and companies [1] Group 1: Market Performance - As of the report, the consumer electronics ETF, 富国 (561100), saw an intraday increase of 3.85%, with stocks like 环旭电子 hitting the daily limit, and 立讯精密 rising over 9% [1] - Other notable performers included 歌尔股份 and 闻泰科技, both of which saw increases exceeding 6% [1] Group 2: Sales Data - According to a report by Counterpoint, early sales of the iPhone 17 series in China and the US are strong, with sales 14% higher than the iPhone 16 series [1] - The basic model of the iPhone 17 nearly doubled its sales in China [1] Group 3: Long-term Outlook - Analysts suggest that AI hardware innovation and contributions from the second growth curve will continue to open up growth opportunities in the medium to long term [1] - The supply chain for Apple products has emerged as a high-quality sector within electronics, characterized by both safety margins and resilience [1] Group 4: Investment Opportunities - The 富国 consumer electronics ETF closely tracks the 中证消费电子主题指数 (931494), which selects 50 listed companies involved in component production and brand design within the consumer electronics space [1] - Investors without stock accounts can also access opportunities through the 富国 consumer electronics ETF connection (Class A 015876; Class C 015877) [1]

Investment Rating - The report assigns a "Buy" rating for Tonglian Precision (688210) with a first coverage [1]. Core Views - The rapid evolution of generative AI technology is driving a new wave of hardware innovation in consumer electronics and smart devices. This includes advancements in foldable smartphones, AI personal computers, smart glasses, and wearable devices, which enhance user experience and stimulate consumer demand [4]. - The company is expanding its precision manufacturing capabilities to meet the demand for lightweight and high-performance components, utilizing new materials and technologies such as 3D printing. This positions the company favorably in the market for new smart terminal products [5]. - The company plans to issue convertible bonds to raise up to 595 million yuan, primarily for a new intelligent manufacturing center focused on lightweight materials, which will enhance its role from a traditional component supplier to a comprehensive technology service provider [6]. Financial Projections - Revenue projections for the company are 1.12 billion yuan in 2025, 1.52 billion yuan in 2026, and 2.10 billion yuan in 2027, with net profits of 101.11 million yuan, 193.90 million yuan, and 329.62 million yuan respectively [7][10]. - The company is expected to achieve a revenue growth rate of 44.93% in 2024, followed by 37.62% in 2025, 35.80% in 2026, and 38.30% in 2027 [10][11].

Core Viewpoint - Nvidia is expected to unveil significant advancements in AI hardware, including the Blackwell Ultra chip and details about the Rubin platform, at the upcoming GTC 2025 conference, which may help revive market sentiment towards AI stocks [1][2]. Group 1: Blackwell Ultra Chip - The Blackwell Ultra (GB300) chip is anticipated to be a highlight of the GTC conference, featuring improvements in HBM memory capacity and power consumption compared to its predecessor B200 [3]. - The changes in the Blackwell Ultra system are expected to benefit suppliers in power, battery, cooling, connectors, ODM, and HBM sectors [3]. Group 2: Rubin Platform - The Rubin platform is projected to be a new engine for AI computing by 2026, with Nvidia likely to share some details at the GTC conference [4]. - The Rubin GPU is expected to have a massive HBM capacity of 288GB, a thermal design power (TDP) of 1.4kW, and a 50% performance increase in FP4 computing compared to B200, with shipments starting in Q3 2025 [4][5]. - The Rubin platform may feature a dual logic chip structure, HBM4 memory with a total capacity of 384GB, and an expected TDP of around 1.8kW [5]. Group 3: CPO Technology - Nvidia's CPO (Co-Packaged Optics) technology is anticipated to be another major highlight at the GTC conference, aimed at enhancing bandwidth, reducing latency, and lowering power consumption [6][7]. - Initial applications of CPO are expected in switches, with widespread GPU-level adoption projected for the Rubin Ultra era in 2027 [8]. Group 4: Physical AI and Humanoid Robots - There is an increasing market focus on physical AI and humanoid robots, with Nvidia expected to showcase advancements in these areas at the GTC conference [9]. - Nvidia has already introduced platforms like Cosmos and GR00T, and further announcements regarding multimodal AI, robotics, and digital twins are anticipated [9][10].