1月4日，成都星拓微电子科技股份有限公司正式启动上市辅导备案。这家来自成都的互联芯片领域"潜力股"，正式踏上了冲击资本市场的新征程。 | | | 公开发行辅导公示 | | 辅导对象 | 辅导机构 | 备客时间 辅导状态 | 派出机构 | 报告类型 | 报告标题 | | --- | --- | --- | --- | --- | --- | --- | --- | --- | --- | | - | 北京 | 天津 | | 天津七一二移动通信股份有 順公司 | 中国银河证券股份有限公司 | 2026-01-04 博导演室 | 天津证监局 | 第三章案报告 | 关于大津七一二移 ... | | - | 河北 | - | шта | | | | | | | | | 内蒙古 | - | 辽宁 | 无锡市晶源微电子股份有限 公司 | 中信证券股份有限公司 | 2024-01-11 辅导备宽 | 江苏证监局 | | 辅导进展情况报告 关于无锡市晶源微 ... | | - | 吉林 | 黑龙江 | | 成都星拓微电子科技股份有 | | | | | | | - | 奥 | | 江苏 | 增√自 | 中国国际金融股份有限公司 ...

Core Insights - TianShu ZhiXin Semiconductor Co., Ltd. has officially launched its IPO, planning to issue 25.43 million H-shares to raise approximately HKD 3.7 billion, with the offering period from December 30, 2025, to January 5, 2026 [1] - The company is recognized as the first domestic chip design firm to achieve mass production of general-purpose GPU chips for inference and training, utilizing 7nm technology [1] Business Deployment Value - TianShu ZhiXin has successfully delivered over 52,000 general-purpose GPU products to more than 290 clients across various sectors, achieving over 900 deployments [2] - The company ranks third in China's training GPU market share and second in the inference GPU market among domestic companies for 2024 [2] Differentiation Strategies - The company adopts an integrated hardware-software design approach, treating compilers, drivers, and libraries as inseparable from the product, which enhances the interaction and iteration between hardware and software [3] - TianShu ZhiXin's solutions support major Linux distributions and are compatible with both x86 and ARM architectures, optimizing for key AI frameworks like PyTorch and TensorFlow [3] Full-Stack Solution Capability - The company positions itself not just as a hardware provider but as a full-stack solution provider, offering tailored AI computing solutions from single machines to clusters [4] - TianShu ZhiXin provides specialized training and inference products, integrating its GPUs with third-party infrastructure for flexible model deployment [4][6] Innovation Engine - The R&D team consists of over 480 professionals, with around 70% holding master's degrees or higher, ensuring continuous technological iteration [7] - The company has launched multiple generations of products, including TianGu and ZhiKai series, with a commitment to a "produce one generation, design one generation, and research one generation" philosophy [7] Market Timing and Growth Potential - The IPO coincides with a critical period for domestic chip replacement, with the general GPU market in China expected to grow at a compound annual growth rate of 72.8% from 2022 to 2024 [8] - The domestic GPU market share is projected to increase from 8.3% in 2022 to over 50% by 2029, indicating significant growth opportunities for the company [8] Future Product Development - The company plans to develop next-generation products, including ZhiKai Gen 2 and Gen 3, and TianGu Gen 4 and Gen 5, with expected production timelines from 2025 to 2027 [9] Strategic Positioning - The IPO provides TianShu ZhiXin with a unique capital platform, facilitating international investment and future expansion [10] - The company emphasizes its engineering capabilities and sustainable business model, positioning itself as a pragmatic player in the domestic chip industry [11]

截至2024年，杰理科技TWS耳机芯片累计销量已突破46.59亿颗，全球市场占有率约40%，充分印 证其广泛的市场认可与行业引领地位。 2025年12月20日，由半导体投资联盟和集成电路投资创新联盟主办、ICT知识产权发展联 盟协办、爱集微承办的"2026半导体投资年会暨IC风云榜颁奖典礼"在上海隆重举行。杰理 科技荣获IC风云榜"年度领军企业奖"。 杰理科技专注于系统级芯片（SoC）的设计与研发，产品广泛覆盖蓝牙音视频、智能穿戴、智能物 联终端等多个领域，致力于为全球客户提供高规格、高灵活性与高集成度的芯片解决方案。 作为国家级制造业单项冠军企业、国家重点集成电路设计企业，杰理科技芯片累计销量已突破百亿 颗，其蓝牙音频芯片连续多年位居全球市场占有率首位，展现出卓越的品牌影响力与行业话语权。 在科研创新与人才建设方面，杰理科技依托"广东省射频智能企业重点实验室"等四大省级科研平 台，并设立国家级博士后科研工作站及广东省博士工作站，持续推动集成电路领域高端人才的引进 与培养，为产业注入源源不断的智力支持。知识产权是保障创新成果、激励技术突破的重要支柱。 截至目前，杰理科技已累计获得国内外知识产权700余项，构建 ...

Core Viewpoint - The article discusses the challenges and opportunities for domestic AI chips, particularly Baidu's Kunlun chip, in supporting large-scale training for next-generation models, amidst the ongoing dominance of Nvidia in the market [1][5]. Group 1: Challenges in Large-Scale Training - The evaluation of chip capabilities has shifted from mere computational power to the ability to stably support training for models ranging from hundreds of millions to trillions of parameters [1][5]. - The first major challenge is cluster stability, where any interruption in a large-scale training system can lead to significant downtime, especially in systems with thousands of GPUs [7][10]. - The second challenge involves achieving linear scalability in large clusters, which requires advanced communication optimization and system-level coordination [10][11]. - The third challenge is the model ecosystem and precision system, where Nvidia's extensive model ecosystem provides a competitive edge in training accuracy [15][19]. Group 2: Solutions and Strategies - To address cluster stability, the company emphasizes the need for detailed monitoring and verification to preemptively identify potential issues [8][9]. - For scalability, the company has developed a communication strategy that bypasses CPU limitations, allowing for optimized task management across different workloads [14][20]. - The company is focusing on a highly generalized operator system to ensure reliability in large-scale training, adapting to various model sizes and shapes [19][27]. Group 3: Current Developments and Future Directions - The company has successfully implemented large-scale training with its Kunlun chip, achieving significant results with models like Qianfan-VL and Baidu Steam Engine, which have demonstrated state-of-the-art performance in various tasks [28][30]. - The future direction includes expanding the capabilities of domestic chips to support even larger clusters and more complex models, aiming for a comprehensive coverage of major model systems [27][31]. - The article highlights the importance of binding advanced self-developed models to the Kunlun chip to enhance its acceptance and performance in the market [29].

12月17日，"国产GPU第二股"沐曦股份正式上市。此前"国产GPU第一股"摩尔线程上市后即迎来大涨， 催生了市场乐观情绪，沐曦股份也因而受到投资者热捧，开盘后同样迎来大涨，开盘价700元较发行价 暴涨569%。盘中涨幅继续扩大，截至12月17日午盘报824.50元，涨687.79%，总市值3298.82亿元。以此 计算，中一签将赚35.99万元。 多位高管来自AMD、华为海思 葛卫东、经纬创投、红杉资本都是股东 据南都此前报道，沐曦股份诞生于2020年9月。彼时，受中美科技战影响，国产芯片替代概念快速升 温，涌现出沐曦、摩尔线程、壁仞科技等一批研发GPU或AI芯片的企业。 沐曦股份核心创始团队主要由陈维良、彭莉和杨建三人组成。招股书显示，三人均长期任职于AMD上 海公司，其中陈维良于2007年1月至2020年8月任超威半导体（上海）有限公司高级总监；彭莉于2007年 7月至2020年9月，任超威半导体（上海）有限公司企业院士；杨建则于2007年1月至2019年11月，任超 威半导体（上海）有限公司企业院士，离开AMD后，他又先后任上海华为技术有限公司架构师以及海 思技术有限公司架构师，2020年9月加入沐曦 ...

Group 1 - The core viewpoint of the articles is the announcement of the subscription situation and winning rates for the IPOs of two semiconductor companies, Muxi Co., Ltd. and Angrui Microelectronics, on the Sci-Tech Innovation Board [1][2] Group 2 - Muxi Co., Ltd. has an issuance price of 104.66 yuan per share, with a total of 40.1 million shares offered, aiming to raise 3.904 billion yuan [1] - The effective number of subscription accounts for Muxi Co., Ltd. reached 5.1752 million, with a total of 28.862 billion shares subscribed, resulting in an initial winning rate of 0.02223023% [1] - After the adjustment mechanism was activated, the final number of shares issued online for Muxi Co., Ltd. was 9.6655 million, accounting for 29.74% of the total issuance [1] Group 3 - Angrui Microelectronics announced an issuance price of 83.06 yuan per share, with a total of 24.8829 million shares offered, aiming to raise 2.067 billion yuan [2] - The effective number of subscription accounts for Angrui Microelectronics reached 5.1292 million, with a total of 17.314 billion shares subscribed, resulting in an initial winning rate of 0.02299243% [2] - Following the activation of the adjustment mechanism, the final number of shares issued online for Angrui Microelectronics was 5.9720 million, accounting for 30% of the total issuance [2]

来源：大观国研 12月5日，在中国外交部例行记者会上，发言人林剑对一则技术出口问题作出了明确回应。英媒记者提 问中提到，AMD首席执行官苏姿丰称其公司获得了部分MI308芯片对华出口许可，并愿意将对华销售 收入的15%上缴美国政府。表面看是一项"有诚意"的商业交易，但林剑仅用一句"请美方以实际行动维 护全球供应链稳定"，就点破了其中的政治本质。 这番回应看似平淡，实则传递了明确信号：中国不会接受打了折扣的技术产品，更不会为美方的规则设 限背书。林剑没有就"是否购买"展开论述，而是直接回到了本源问题——谁在设限、谁在控制、谁在制 造不平等的技术规则。 从回应中可以读出，中方已不再容忍美方一边用行政手段限制中国发展高端技术，一边又想继续向中国 市场倾销经过性能阉割的"次级芯片"。这是一场在对等尚未建立的前提下，拒绝继续妥协的外交表达， 也是对当前出口机制的系统性反制。 "阉割版芯片"背后的双重陷阱 MI308并非AMD旗下的高端产品，而是在MI300X基础上为中国市场"量身定制"的降级版，其定位与此 前英伟达的H20芯片如出一辙：牺牲算力、压缩带宽、降低互联能力，只为拿到美国政府的出口许可标 签。 这种"特供化" ...

Core Points - Moer Technology, known as the "first domestic GPU stock," officially listed on the Sci-Tech Innovation Board on December 5, 2025, with an opening price of 650 CNY per share, marking a 468.78% increase from its IPO price of 114.28 CNY [1][3] - The company aims to raise 8 billion CNY through this IPO, primarily for the development of next-generation AI training and inference chips, as well as other GPU projects [7][10] - Moer Technology has raised over 10 billion CNY in total funding since its inception, with a notable investor base including Sequoia Capital and Tencent [8][10] Company Overview - Moer Technology was founded in June 2020 by Zhang Jianzhong, a former vice president of NVIDIA, and has quickly gained attention in the GPU market [8][10] - The company has launched four generations of GPU architecture chips from 2021 to 2024, with plans to release an intelligent SoC chip named "Changjiang" in 2024 [7][10] - As of the latest financial reports, Moer Technology has not yet achieved profitability, with projected revenues of 0.46 billion CNY, 1.24 billion CNY, and 4.38 billion CNY for 2022 to 2024, respectively [10][12] Market Position - Moer Technology's full-function GPU is positioned to compete directly with NVIDIA, offering unique advantages in AI training, rendering, and high-performance computing [15] - The company has a low subscription rate of approximately 0.0363% for its IPO, indicating high demand but limited access for individual investors [5][10] - The competitive landscape is intensifying, with other domestic companies like Muxi Technology also pursuing IPOs, which could impact Moer Technology's market share [14][15]

2025年12月5日，三大股指集体收涨。图为一股民手机上呈现的摩尔线程收盘价为600.50元/股。 IC供图 12月5日，摩尔线程头顶"国产GPU第一股"光环正式登陆科创板，证券代码为"688795"，上市首日高开 468.78%，报650元/股，开盘市值超3000亿元。最高飙升至688元/股，涨幅超500%，打新中一签最高可 获利28.7万元。有投资者惊呼"摩伟达"来了。 此前，摩尔线程披露上市将发行7000万股，上市发行价为114.28元/股，这也是2025年以来发行价最高的 新股。截至昨日收盘，摩尔线程报收600.5元/股，当日累计涨幅为425.46%。在上市首日便一跃成为A股 第三，目前仅次于贵州茅台与寒武纪。 创年内科创板多项纪录 万名打新者不足4人中签 回顾摩尔线程的上市之路，其创下了年内科创板多项纪录。摩尔线程最早于2024年11月启动上市辅导， 后在今年6月8日完成辅导验收，历时7个月，6月30日，其IPO申请正式获得受理，9月26日，上交所上 市委审议通过了摩尔线程发行上市申请，其于当日向证监会提交注册申请，从申报到过会仅用时88天， 成为今年科创板从受理到过会用时最短的企业，10月30日， ...

12月5日，摩尔线程头顶"国产GPU第一股"光环正式登陆科创板，证券代码为"688795"，上市首日高开468.78%，报650元/股，开盘市值超3000亿元。有投资 者惊呼"摩伟达"来了。 此前，摩尔线程披露上市将发行7000万股，上市发行价为114.28元/股，这也是2025年以来发行价最高的新股，按开盘价计算，中一签赚超26.7万元。 不过，在高开之后，摩尔线程股价开始下滑，截至今日午间收盘，报590.59元/股，较之发行价上涨416.79%，市值为2775.9亿元，居A股"老四"之位，仅次于 贵州茅台、寒武纪和源杰科技。 | | 最新 . | 主力净流入 | 涨幅 ▶ | | --- | --- | --- | --- | | 贵州茅台 600519 | 1424.00 | 1.33亿 | 0% | | 寒武纪-U 科创 688256 融 | 1392.00 -6817.7万 | | 1.68% | | 源杰科技 科创 688498 融 | 605.61 | 1.37亿 | 4.64% | | N摩尔-U 科创 688795 新 | 590.59 | | 21.59亿 416.79% | 11月24日 ...