黄仁勋终于得到了他最想要的东西。 本周美国政府正式批准英伟达向中国以及其他"经批准的客户"出售高端的H200 GPU芯片，但需要向美国政府缴纳25%的销售提成。这一提成比 例同样适用于AMD、英特尔等其他美国芯片巨头。不过，英伟达最新的Blackwell和未来的Rubin系列GPU仍然禁止出口。 这标志着黄仁勋长达数月的游说取得成功。过去半年时间，他不断造访佛罗里达与华盛顿，随着特朗普总统一道出访和出席国宴，向白宫宴会厅 建设工程捐款，就是为了这一刻。就在上周，他再一次来到白宫会见总统，终于如愿以偿得到了解锁禁运令。 受这一利好消息推动，英伟达股价盘后应声上涨。受美国政府连续多道芯片加码禁运令限制，过去两年时间，英伟达一步步失去迅猛增长的中国 市场，丢掉了在AI GPU市场原先高达95%的份额。在英伟达最核心的数据中心业务，中国市场的营收占比也从原先的四分之一急剧下滑。 心急如焚的黄仁勋在两个月前公开抱怨， "我们已经失去了全球最大的市场之一，在中国市场完全出局，市场份额已经归零。 "即便是向美国政府 缴纳四分之一的提成，对英伟达的业绩营收也意义重大，因为中国AI GPU今年规模估计高达200亿-300亿美元 ...

硅谷观察/郑峻 黄仁勋终于得到了他最想要的东西。 本周美国政府正式批准英伟达向中国以及其他"经批准的客户"出售高端的H200 GPU芯片，但需要向美 国政府缴纳25%的销售提成。这一提成比例同样适用于AMD、英特尔等其他美国芯片巨头。不过，英 伟达最新的Blackwell和未来的Rubin系列GPU仍然禁止出口。 这标志着黄仁勋长达数月的游说取得成功。过去半年时间，他不断造访佛罗里达与华盛顿，随着特朗普 总统一道出访和出席国宴，向白宫宴会厅建设工程捐款，就是为了这一刻。就在上周，他再一次来到白 宫会见总统，终于如愿以偿得到了解锁禁运令。 受这一利好消息推动，英伟达股价盘后应声上涨。受美国政府连续多道芯片加码禁运令限制，过去两年 时间，英伟达一步步失去迅猛增长的中国市场，丢掉了在AI GPU市场原先高达95%的份额。在英伟达 最核心的数据中心业务，中国市场的营收占比也从原先的四分之一急剧下滑。 心急如焚的黄仁勋在两个月前公开抱怨，"我们已经失去了全球最大的市场之一，在中国市场完全出 局，市场份额已经归零。"即便是向美国政府缴纳四分之一的提成，对英伟达的业绩营收也意义重大， 因为中国AI GPU今年规模估计高达2 ...

（文章来源：人民日报） 因此，这本书不是写给程序员的工具手册，也不是写给管理者的操作指南，而是希望为更广泛的读者提 供一个进入"计算社会"的入口。每一位读者都可以通过自己的视角阅读这张地图，也可以用自己的经验 验证其中的逻辑结构。 我们正生活在一个由算力编织的社会，理解它，不是为了技术本身，而是为了更好地理解我们自己的位 置与方向。 算力，已成热词。但对许多人而言，它是新闻标题里的技术术语，既熟悉又难以把握。如果说蒸汽机承 载了工业时代的想象，电网塑造了现代城市的形态，今天的算力，是否也具备了改变社会节奏的能力？ 政策文件频繁提及算力，媒体讨论着"东数西算""超算中心""AI芯片"等名词，企业决策越来越依赖模型 分析与数据模拟……结构性变化背后，算力在持续推动。算力生态不同于自然意义上的生态系统，并非 自发生长，也不具备自然平衡机制。它的运行依赖能源支撑、制度参与、商业投资、人才培养、数据供 给等多种要素持续投入。使用"生态"这个词，是为了强调这种结构的联动性、演化性与系统关联。 本书试图摆脱行业报告式写作。我们不以某一类设备为出发点、某一项技术为主线，而是从问题导向出 发，围绕算力如何融入社会运行、如何塑造 ...

Core Viewpoint - The U.S. has suddenly approved the sale of NVIDIA's high-end AI chip H200 to China, contrasting sharply with previous restrictions on the less capable H20, likely due to advancements in China's AI chip production and deployment [1][9]. Group 1: U.S. Policy Change - The approval of H200 indicates a significant shift in U.S. policy, possibly in response to China's advancements in AI chip technology [1][9]. - The previous restrictions on H20, which had only a fraction of H200's performance, highlight the rapid developments in China's AI capabilities [9]. Group 2: Competitive Landscape - Chinese chip companies claim to have developed AI chips that can rival NVIDIA's A100, suggesting that the gap between domestic and foreign technology is narrowing [3][7]. - Other U.S. chip manufacturers, such as AMD and Intel, are also developing AI chips, but they face challenges due to NVIDIA's established ecosystem [3]. Group 3: NVIDIA's Market Position - NVIDIA's dominance in the AI chip market has allowed it to increase prices, achieving a profit margin of 55.8%, with some chips priced higher than a Tesla Model Y [5]. - Despite complaints from AI companies about high prices, NVIDIA remains unyielding, while companies like Google and Facebook are taking steps to develop alternatives [5]. Group 4: Alternative Approaches - Google has successfully developed its TPU chips, which can compete with NVIDIA's offerings, and plans to further integrate these into its AI models [5][7]. - The development of advanced algorithms may reduce reliance on high-performance chips, opening new avenues for domestic AI models to utilize local chip capabilities effectively [7].

"国产GPU第一股"摩尔线程登陆科创板后的第五日，其股价突破900元，成为仅次于贵州茅台与寒武纪 的A股单价第三高公司，收盘总市值突破4400亿元，较其上市首日增长56%，与寒武纪最新市值也仅相 差1400亿元。 图片来源：视觉中国 根据摩尔线程在IPO中的首轮问询回复，在国内AI芯片市场中，2024 年，英伟达（全功能GPU）、华为 海思（ASIC）以及AMD（GPGPU）在国内AI芯片市场中的市场份额分别为54.4%，21.4%，15.3%，摩 尔线程在国内GPU市场中的份额占比低于1%。 12月5日为摩尔线程上市首日，其股价已暴涨逾4倍，市值冲至2823亿元，打新者中一签盈利最高接近28 万元。而与尚未盈利的硬科技新贵市值如火箭般蹿升形成尖锐对比的是，同期A股市场整体情绪持续低 迷以及价值投资者面对不断缩水的账户陷入失语状态。摩尔线程的这场资本盛宴，不仅是一次个股的狂 欢，更成为观测当下市场逻辑剧变的典型样本，也对A股的价值投资者提出新的课题。 摩尔线程估值之"奇"，在于即便按2025年前三季度7.85亿元营收估算，其当前动态市销率也接近300 倍。这一估值水平将国内同行远远甩在身后：业务更为成熟、已实 ...

根据新思科技于12月10日美股盘后公布，2025会计年度第四季（截至2025年10月31日止），营收 年增37.8%至22.6亿美元，略高于LSEG共识预期的22.5亿美元。 不计股票奖酬费用等一次性项目，经调整每股盈余2.90美元，虽低于2024年同期的3.40美元，但 仍优于华尔街预估值2.78美元。 如果您希望可以时常见面，欢迎标星收藏哦~ 全球IC设计自动化软件（EDA）第一大厂新思科技（Synopsys）揭晓最新财报，在AI与高效能运 算（HPC）需求强劲带动下，客制化ASIC芯片订单涌现，上季业绩超出市场预期，带动10日盘后 股价走扬逾2%。 旗 下 负 责 半 导 体 设 计 软 件 、 验 证 平 台 和 IC 制 造 软 件 销 售 业 务 的 「 设 计 自 动 化 」 （ Design Automation）部门为营运金鸡母，第四季营收成长65%，从2024年同期的11.2亿美元攀升至18.5 亿美元，占总营收高达八成，显示AI芯片设计热潮持续推升EDA工具需求。 展望2026会计年度第一季，新思科技预估营收可达23.6亿美元至24.2亿美元，中位数高于华尔街 预期的23.8亿美元，经调 ...

来源：中科创星 每日｜荐读 论坛： 新一批敲钟人，已在路上 荐读： 重新发现香港：科创时代的新蓝图 榜单： 全国首只AIC产业母基金来了 热文： 投资人"忙疯了" 12月11日，中科创星先导创业投资基金完成终关募集，基金总规模达到40.8亿元。本次终关签约的新增LP涵盖太保资本长航母基金、海淀区中关村 科学城科技成长基金、蚂蚁集团、广州产投、珠海科技产业集团、南京市创投集团、复旦科创母基金、国联民生等多元化机构。 整体来看，中科创星先导创业投资基金的LP组合，进一步推动了LP结构向"国家级母基金+险资+产业资本+高校资本+区域资本"的多元协同格局升 级。基金自首关以来，在不到半年的时间内已投资46笔项目， 包括AI芯片、TPU、具身智能、AI4Science、AI硬件、量子、可控核聚变、小型裂 变堆，以及基因编辑、免疫细胞治疗、病毒感染预防、生物基色素、农业微生物等领域。 基金"投早投小"特色鲜明，前两轮项目达42笔，占比超过90%，精准覆盖技术萌芽阶段。同时，源头创新转化成效显著，有32笔项目直接来源于科 研院所与高校的成果转化，有效激活了智力资源。 ...

Core Insights - TSMC's advanced packaging capacity is fully booked, with NVIDIA accounting for over half of the orders, indicating strong demand for semiconductor manufacturing [1] - NVIDIA has reserved 800,000 to 850,000 wafers for 2026, significantly outpacing competitors like Broadcom and AMD [1][3] Group 1: NVIDIA's Capacity Reservation - NVIDIA's large-scale capacity reservation is primarily to meet the growing production demands for the Blackwell Ultra chip and to prepare for the next-generation Rubin architecture [3] - Current orders do not include potential demand from the Chinese market for the H200 AI chip, suggesting that NVIDIA's capacity needs may increase further [3] Group 2: TSMC's Response to Demand - TSMC is actively expanding its advanced packaging facilities, planning to build eight wafer fabs at the AP7 plant and introducing two new packaging factories in Arizona, expected to start mass production in 2028 [3] - Due to limited capacity, TSMC has decided to outsource some processes of its CoWoS advanced packaging to companies like ASE and SPIL in Taiwan [3] Group 3: Industry Alternatives and Technology - The outsourcing decision has prompted some companies to consider alternative solutions, with Intel's EMIB technology gaining attention as a viable option [3] - EMIB offers advantages in area and cost, allowing for highly customized packaging layouts, but for GPU suppliers like NVIDIA and AMD, TSMC's CoWoS remains the preferred solution due to its bandwidth, transmission speed, and low latency requirements [3]

Group 1 - The H200 chip from Nvidia is expected to attract Chinese cloud vendors for procurement after its sales restrictions are lifted, despite its performance being lower than the current Blackwell series chips [2] - The H200 chip shows a significant performance improvement over the H20 chip, with a peak theoretical computing power of 1979 TFLOPS compared to H20's 148 TFLOPS, and it features 141 GB of HBM3e memory versus H20's 96 GB of HBM3 [2] - The procurement of the H200 chip will depend on formal policies, and cloud vendors are likely to test domestic AI chips to mitigate risks [3] Group 2 - The entry of the H200 chip into the Chinese market may intensify competition in the high-end chip sector, with local companies focusing on performance, compliance, and cost-effectiveness to strengthen their market position [4] - If Nvidia's H200 and AMD's MI325 can be exported to China, they could capture about 30% of the high-end AI chip market by 2026, while domestic ASIC chips are expected to hold 20% [5] - The overall supply of high-end AI chips in China is projected to grow by over 60% by 2026 [5]

Core Viewpoint - The article discusses NVIDIA's development of a location verification technology for its AI chips amid ongoing U.S. export restrictions on semiconductor technology to China, highlighting the geopolitical tensions and the company's response to concerns about chip smuggling [1][4]. Group 1: NVIDIA's Technology Development - NVIDIA is reportedly testing a location verification technology that can determine the country in which its chips are operating, aimed at preventing AI chips from being smuggled to countries affected by export bans [1]. - The technology will initially be available as an optional software for customers and will utilize NVIDIA's GPU capabilities for operation [1]. - The first implementation of this technology will be on the new Blackwell chips, with discussions ongoing for earlier generations [1]. Group 2: U.S. Export Restrictions and Geopolitical Context - The U.S. government has intensified export restrictions on semiconductor technology to China, citing national security concerns and the need to prevent AI chips from being smuggled into banned countries [4]. - NVIDIA's products are under scrutiny due to potential security vulnerabilities and monopoly risks, leading to investigations by Chinese authorities [4][5]. Group 3: Industry Response and Market Dynamics - In response to U.S. restrictions, Chinese companies are accelerating the development of their own AI chips to capture market share previously dominated by NVIDIA [5]. - Major Chinese tech firms, including Huawei, Alibaba, Tencent, and ByteDance, are increasing investments in chip research and development to enhance their supply chain autonomy [5]. - The Chinese government has expressed strong opposition to U.S. actions, arguing that such measures hinder global semiconductor industry development and ultimately harm the U.S. itself [5].