Group 1 - The core viewpoint of the news highlights the strong performance of the AI sector, with the Zhongzheng AI Theme Index rising by 1.33% and individual stocks like Fudan Microelectronics and Zhongke Star Map showing significant gains [1] - OpenAI is set to release the GPT-5.2 update soon, aiming to close the performance gap with Google's Gemini 3, which has received high praise from industry leaders [1] - The AI Artificial Intelligence ETF has shown a consistent upward trend, marking a 1.15% increase and achieving three consecutive days of gains [1] Group 2 - Major cloud service providers are increasing investments in AI chip development, with Amazon AWS launching the next-generation Trainium3 chip and previewing Trainium4, which is expected to enhance training efficiency and reduce costs [2] - The Zhongzheng AI Theme Index comprises 50 listed companies involved in providing resources, technology, and application support for AI, reflecting the overall performance of AI-related securities [2] - As of November 28, 2025, the top ten weighted stocks in the Zhongzheng AI Theme Index account for 63.92% of the index, indicating a concentrated investment in key players within the AI sector [2]

头顶"GPU第一股"称号，摩尔线程以600.5元/股的收盘价直接跻身第三大高价股，仅次于贵州茅台、寒 武纪，这也让目前的A股百元股阵营扩容至153只。经同花顺iFinD统计，截至12月5日收盘，在目前已 有的153只百元股中，上纬新材、天普股份、鼎泰高科等88股为今年来加入的百元股成员，除摩尔线程 之外，云汉芯城、同宇新材在内的8股均为年内上市新股。另外，与年初相比，百元股数量翻倍，在不 复权形式下，年初百元股数量为72只。 摩尔跻身第三大高价股 经统计，截至12月5日收盘，A股市场百元股阵营扩容至153只，较年初百元股的数量增加了81只。 具体来看，A股市场上共有两只千元股，与寒武纪年内多次胶着，贵州茅台目前仍稳居A股"股王"，最 新股价报1430.01元/股，排在第二位的寒武纪，最新股价1355元/股。 12月5日上市的摩尔线程，以600.5元/股的收盘价位居A股市场第三位。交易行情显示，摩尔线程上市当 日大幅高开468.78%，开于650元/股，截至收盘，公司股价最终收涨425.46%，收于600.5元/股。此外， 还有源杰科技、中际旭创、国盾量子等12股股价在300元/股以上。 百济神州、春风动力、海 ...

AI 需求持续增长及国家大力支持推动国产算力发展，尽管面临先进制程 瓶颈，但以 Deepseek 为代表的算法已在全球表现出色，硬件和软件算 法在兼容国际标准方面亦有创新突破，如 FP8 运算和 CUDA 兼容性。 国产半导体设备国产化率自 2019 年起稳步提升，表明行业正逐步克服 技术壁垒。寒武纪、升腾、海光等公司技术不断迭代，已涉及 12nm 至 7nm 制程，单纯算力角度已领先，生态系统兼容性是当前挑战，但有望 通过时间和投入突破。 预计未来两三年内，国产算力将从流片量转向订单量，行业将经历从 0 到 1 到 10 再到 100 的扩张。大厂和创业公司积极下单扩产，当前是投 资该领域的时机。2025 年底至 2026 年，国产算力仍将是二级市场关 注焦点，尤其在中美博弈背景下。 中芯国际为首的新制程进展乐观，良率和产能预期良好，2026 年或是 业绩兑现和股价上升的关键年，存在投资机会。建议关注华为、海光、 寒武纪等龙头企业，以及木兮、摩尔、阿里平头哥、百度昆仑芯等第二 梯队公司。 Q&A 近年来国产算力的发展情况如何？ 近年来，国产算力领域取得了显著进展。尤其是摩尔线程和牧原等新生代企业 的上市，吸 ...

"寒王"寒武纪(688256.SH)紧急声明，传言不实。 根据公开信息，当地时间12月3日，英伟达CEO黄仁勋在一场活动中发出警告称："如果将中国市场完全 让给其本土企业，还将为中国向其他国家输出先进技术创造便利条件。" 当日，彭博社报道称：寒武纪计划在2026年将其AI芯片产量提高3倍以上，目标是从华为手中夺取中国 市场份额，并填补英伟达被迫退出后留下的空白。 此前经营一直亏损的寒武纪，在2025年迎来了爆发的好势头。前三季度，公司实现营业收入约46亿元， 同比增长接近24倍；归母净利润约16亿元，首次实现盈利。 寒武纪继续加码研发，2025年前三季度，公司研发投入超过7亿元。 寒武纪也有风险需要化解，比如客户集中度偏高。2024年，公司向前五名客户销售占比达94.63%。 "产能提升3倍以上"系谣言 针对市场传闻，寒武纪发布了紧急声明，予以澄清。 12月4日晚，寒武纪发布严正声明称，当日媒体及网络传播的关于公司产品、客户、供应及产能预测等 相关信息，均为误导市场的不实信息。 当日，有外媒报道称，2026年，寒武纪计划将人工智能(AI)芯片产量提升逾3倍，对标华为抢占市场份 额，并填补因英伟达被迫退出而产生 ...

头顶"GPU第一股"称号，摩尔线程以600.5元/股的收盘价直接跻身第三大高价股，仅次于贵州茅台、寒 武纪，这也让目前的A股百元股阵营扩容至153只。经同花顺iFinD统计，截至12月5日收盘，在目前已 有的153只百元股中，上纬新材、天普股份、鼎泰高科等88股为今年新加入的百元股成员，除摩尔线程 外，云汉芯城、同宇新材在内的8股均为年内上市新股。此外，与年初相比，百元股数量翻倍，在不复 权形式下，年初百元股数量为72只。 摩尔线程跻身第三大高价股 经统计，截至12月5日收盘，A股市场百元股阵营扩容至153只，较年初百元股的数量增加81只。 具体来看，A股市场上共有两只千元股，与寒武纪年内多次胶着，贵州茅台目前仍稳居A股"股王"，最 新股价报1430.01元/股；排在第二位的寒武纪，最新股价1355元/股。 12月5日上市的摩尔线程，以600.5元/股的收盘价位居A股市场第三位。交易行情显示，摩尔线程上市当 日大幅高开468.78%，开于650元/股，截至收盘，公司股价最终收涨425.46%，收于600.5元/股。此外， 还有源杰科技、中际旭创、国盾量子等12股股价在300元/股以上。 百济神州、春风动力、海 ...

这背后，是卢春霖在摩尔线程（SH688795，股价600.50元，市值2823亿元）的融资过程中押下的重 注。 "一下就达成共识了，（团队里）没有一票是否决的。因为之前投过相关领域，看得越来越清楚。而且 寒武纪的成功让我们认清楚了这个方向。"回忆团队投资摩尔线程的决策时，卢春霖说道。按照12月5日 收盘价计算，拓锋在摩尔线程的投资浮盈超过700%。 窗外车水马龙，而卢春霖的脑海里装的是一张价值千亿元的中国半导体版图。如果说寒武纪 （SH688256，股价1355元，市值5714亿元）的成功让卢春霖崭露头角，那么摩尔线程的上市，则让这 家隐匿在福建的投资机构浮出水面。 每经记者｜赵李南 每经编辑｜廖丹 福州闽江畔，卢春霖沏了一壶肉桂，这位福州市拓锋私募基金管理有限公司（以下简称拓锋）的董事长 显得既松弛又敏锐。 作为摩尔线程的重要股东，卢春霖不仅押中了中国GPU（图形处理器）的"独角兽"，更投资了沐曦集成 电路（上海）股份有限公司（以下简称沐曦）、北京清微智能科技有限公司（以下简称清微智能）等多 个半导体王牌。 这是一场关于技术信仰的豪赌，也是中国本土资本的生存进化论。 始于寒武纪：从"财务投资"到"产业资本" ...

头顶"GPU第一股"称号，摩尔线程以600.5元/股的收盘价直接跻身第三大高价股，仅次于贵州茅台、寒武纪，这也让目前的A股百元股阵营扩容至153只。经 同花顺iFinD统计，截至12月5日收盘，在目前已有的153只百元股中，上纬新材、天普股份、鼎泰高科等88股为今年来加入的百元股成员，除摩尔线程之 外，云汉芯城、同宇新材在内的8股均为年内上市新股。另外，与年初相比，百元股数量翻倍，在不复权形式下，年初百元股数量为72只。 处于100—200元/股区间的百元股数量为118只，包括天普股份、上纬新材、中芯国际、五粮液等诸多知名个股。 按照申万一级行业划分，电子行业百元股数量最多，共57只；其次是计算机行业，有17只百元股；机械设备行业有16只百元股；电力设备、通信行业各有13 只百元股；医药生物、汽车、食品饮料行业百元股分别有11只、8只、7只；剩余行业个股数量分布较少。 天使投资人、资深人工智能专家郭涛表示，年内百元股数量的增加是产业、政策等因素合力的结果。一方面，电子、AI芯片、创新药等领域受益于国产替 代与技术突破，业绩增长确定性强，推动股价攀升；另一方面，政策支持下优质科技企业加速上市，叠加退市制度趋严， ...

Core Viewpoint - Moer Thread has emerged as the "king of new stocks" in A-shares, with its stock price soaring 468.78% on the first day of trading, reaching a market capitalization of over 300 billion yuan, highlighting its significant market interest and potential as a leading domestic GPU player [2][4]. Company Background - Founded by Zhang Jianzhong, a former Nvidia executive, Moer Thread aims to create a "full-function GPU" that integrates graphics rendering, AI computing, and scientific calculations, distinguishing itself from other AI-focused startups [3][4]. - The company has developed the MUSA unified system architecture, which supports multiple tasks and aims to provide a comprehensive solution for various computing needs [3]. Financial Performance - Moer Thread has experienced rapid growth since its establishment in 2020, achieving significant financing and product launches, with projected revenues of 438 million yuan in 2024 and an IPO completed in December 2023 [4][8]. - Despite its growth, the company has incurred substantial losses, exceeding 5 billion yuan over three years, primarily due to high R&D expenditures [7][8]. Market Position and Competition - Moer Thread's technology, particularly the MUSA architecture, is seen as a key player in the domestic GPU market, providing compatibility with Nvidia's CUDA ecosystem and addressing the need for domestic alternatives [6][9]. - However, there are significant gaps in technology and market presence compared to Nvidia, including manufacturing process differences and market share concentration risks [9][11]. Long-term Opportunities - The company aims to build a "full-scenario computing ecosystem" that integrates AI, graphics, and scientific computing, positioning itself as a pivotal player in the digital economy and AI era [14][20]. - With increasing demand for versatile computing capabilities across various industries, Moer Thread's strategy to develop a comprehensive ecosystem could provide substantial long-term growth potential [16][19].

Core Viewpoint - The domestic chip industry in China is undergoing significant transformation, moving from a perception of being "stuck" or "ineffective" to showcasing advancements and capabilities by 2025 [1][3]. Group 1: Consumer Electronics and Automotive - The gaming performance of domestic processors is improving, with the Haiguang C86 processor achieving frame rates of nearly 200 and 300 for major games, indicating readiness for high-performance gaming [2][3]. - The Haiguang C86 processor, featured in the "Black Warrior·Hunting Blade Pro" gaming PC, has a base frequency of 2.8GHz and supports PCIe 5.0, demonstrating its capability to handle high-load tasks [5][6]. - The GAC Aopao GT "Climbing Edition" achieved 100% domestic chip design, utilizing the RK3588M chip for its intelligent cockpit, showcasing the reliability and performance of domestic chips in high-end vehicles [12][13]. Group 2: Storage and Memory Chips - Changxin Storage launched DDR5 products with speeds up to 8000Mbps, marking a significant advancement for domestic manufacturers in the high-end memory market [9][10]. - The introduction of DDR5 technology represents a leap from DDR4, with performance improvements and the ability to address high-frequency signal interference [10][11]. - Longjiang Storage's TiPlus7100s SSD achieved read speeds of 7400MB/s and write speeds of 6900MB/s, indicating competitive performance against international brands [11]. Group 3: AI and Computing Power - The exit of international giants from the AI chip market has created opportunities for domestic companies, with Cambrian Technologies reporting a 1332.52% year-on-year revenue increase in Q3 2025 [15][16]. - Domestic firms are adopting "super node" technology to enhance computing power by integrating multiple AI acceleration cards into a single cabinet, addressing performance bottlenecks in large model training [17][18]. - The demand for storage has shifted from being a cost component to a strategic asset, with significant implications for the supply chain and market dynamics [18][19]. Group 4: Semiconductor Manufacturing and Equipment - Domestic companies are making strides in semiconductor manufacturing equipment, with the shipment of a 350nm lithography machine that supports the production of power devices and RF chips [23][24]. - The development of high-speed oscilloscopes and EDA tools is filling critical gaps in the chip design and testing process, enhancing the capabilities of domestic semiconductor firms [25][26]. - The Chinese semiconductor industry is transitioning from "point breakthroughs" to "chain collaboration," indicating a shift towards more integrated and cooperative development efforts [20][27].

图片来自AI生成 【数智周报将整合本周最重要的企业级服务、云计算、大数据领域的前沿趋势、重磅政策及行研报 告。】观点华为任正非：大量建设大模型是正确的探索，未来算力一定过剩；AI重在"应用"，着眼于未 来3-5年 华为创始人兼总裁任正非与ICPC（国际大学生程序设计竞赛）主席、教练及获奖选手座谈会纪要公 布。在回答关于计算资源的问题时，任正非直言，"以后是算力过剩的，不是你想象的算力不足"。他认 为，建数千个、数百个大模型都是正确的探索。华为在解释超节点时提到了950、960、970等型号， 但"我们需要多少个'970'，970在哪儿能用？需要多少台，集群怎么连接"，任正非表示，这是一种线性 技术推演，是一定能实现的算力假设。而需求是不是一种线性结构尚未可知，"万一是非线性的呢"。因 此，他说："算力过剩的时代一定会到来的。" 谈及当下火热的AI，任正非的观点很明确：AI重在应用，不重在发明。"发明AI只是一家IT公司，应用 AI是会强大一个国家。"任正非表示，IT公司对人类的贡献只2%，AI在产业上的贡献则会占到98%。仅 以洗煤、炼钢为例，洗煤精度提高0.1%，乘以中国40亿吨的煤炭产量，价值惊人；高炉炼 ...