Core Viewpoint - The article discusses the contrasting perspectives on the AI sector and Nvidia's recent financial performance, emphasizing that despite concerns about a potential AI bubble, Nvidia's Q3 results indicate strong growth and demand for its products [2][11][62]. Financial Performance - Nvidia reported a record total revenue of $57.006 billion for Q3, a year-on-year increase of 62%, surpassing market expectations of $54.92 billion [5]. - The net profit reached $31.91 billion, reflecting a 65% year-on-year growth, equating to a daily profit of $3.5 million [6]. - For Q4, Nvidia expects revenue to reach $65 billion, significantly exceeding analyst expectations of $61.66 billion [7]. Business Segments - The data center business, accounting for 89.5% of total revenue, generated $43 billion, primarily driven by the performance of the Blackwell series chips [44]. - The gaming and AI PC segment reported $4.3 billion in revenue, with a 30% year-on-year growth, solidifying Nvidia's position in the consumer market [50]. - The professional visualization segment saw a revenue increase of 62% year-on-year, driven by demand in design tools and medical imaging [55]. Market Dynamics - Nvidia's GPUs are experiencing exponential demand, with every GPU being utilized for training and inference, indicating a robust growth cycle in AI [10]. - The article highlights the potential risks associated with the AI sector, including concerns about circular financing and inflated revenue figures, but emphasizes that the actual demand for Nvidia's products remains strong [18][24]. Future Outlook - Nvidia has committed to a revenue target of $500 billion, with clear visibility into data center revenues for 2025-2026, indicating strong future demand [64]. - Key growth drivers include the anticipated shift from training to inference demand, the emergence of embodied AI, and the ongoing development of sovereign AI projects globally [65]. - The success of the upcoming Rubin platform is crucial for sustaining Nvidia's growth trajectory, with expectations for advanced technology and increased production capacity [68][70].

| Non-GAAP | | --- | 英伟达被视为AI行业的风向标，作为AI算力的核心供给方，其业绩状况直接映射AI行业的真实需求与繁荣程度。 美东时间周三盘后，人工智能（AI）龙头股英伟达公布了超出预期的2026财年第三季度财报。 财报显示，英伟达第三财季营收为570.1亿美元，超出市场预期的549.2亿美元，同比增长62%；净利润为319.1亿美元，同比大涨65%，经调整后的每股收 益为1.30 美元，高于市场预期的1.25美元。 | GAAP | | | | | | | --- | --- | --- | --- | --- | --- | | ($ in millions, except earnings per share) | Q3 FY26 | Q2 FY26 | Q3 FY25 | Q/Q | Y/Y | | Revenue | $57,006 | $46,743 | $35,082 | 22% | 62% | | Gross margin | 73.4% | 72.4% | 74.6% | 1.0 pts | (1.2) pts | | Operating expenses | $5,8 ...

这份备受关注的业绩报告，或缓解市场对人工智能泡沫日益增长的担忧。"我并未看到人工智能泡沫。"黄仁勋说。数据公布后，收涨近3%的英伟达在盘 后交易中涨幅迅速扩大，截至发稿，盘后涨幅一度超过6%。 英伟达业绩超预期 盘后涨幅一度超过6% 美东时间11月19日周三美股盘后，英伟达公布了三季度财务数据，再次超出预期。财报显示，英伟达三季度营收570亿美元，同比增长62%，市场预期 551.9亿美元；净利润为319亿美元，同比增长65%；经调整后的每股收益为1.30美元，高于市场预期的1.25美元。 数据中心业务是英伟达最重要的业务板块。财报显示，英伟达三季度数据中心收入达到512亿美元，同比增长66%。其中，"计算业务"（即GPU）贡献了 430亿美元的营收；网络业务（即允许多个GPU能协同工作的组件）贡献了82亿美元。 英伟达首席财务官Colette Kress在声明中表示，公司当前最畅销的芯片系列是Blackwell Ultra，这是Blackwell芯片的第二代产品。黄仁勋在声明中表 示："云GPU也已售罄。" 在投资者表现出对人工智能泡沫日益增长的担忧之际，AI芯片龙头英伟达再次交出了一份超预期的财报。美东时 ...

第三财季毛利率为73.4%，与上年同期的74.6相比下降1.2%，但与上一财季的72.4%相比增长1%。按照 非公认会计准则计算，毛利率为73.6%，与上年同期的75相比下降1.4%，但与上一财季的72.7%相比增 长0.9%。 新华财经北京11月20日电（王媛媛）美东时间周三盘后，人工智能（AI）龙头股英伟达公布了超出预 期的三季度财报。 受此影响，英伟达股价在美股盘后交易中直线猛拉，一度暴涨超6%，截至发稿涨近5%。美股其他AI概 念股盘后亦全线走强，AMD大涨超3%，台积电、甲骨文、美光科技涨超2%。 有分析指出，随着英伟达公布的财报及业绩指引超出预期，过去几周在市场上不断蔓延的"AI泡沫"忧虑 有望缓解，市场对英伟达AI芯片的需求依然旺盛。 Q3财报及Q4营收展望均超市场预期 财报显示，英伟达第三财季营收为570.06亿美元，与上年同期的350.82亿美元相比增长62%，与上一财 季的467.43亿美元相比增长22%，超过LSEG所调查分析师平均预期的549.2亿美元。 | GAAP | | | | | | | --- | --- | --- | --- | --- | --- | | (S in m ...

美国当地时间周三，芯片巨头英伟达发布了截至2025年10月26日的2026财年第三财季（对应自然年2025年第三季度）财报。 财报显示，英伟达第三财季营收为570.06亿美元，与上年同期的350.82亿美元相比增长62%，与上一财季的467.43亿美元相比增长22%， 超过LSEG所调查分析师平均预期的549.2亿美元。 净利润为319.1亿美元，与上年同期的193.09亿美元相比增长65%，与上一财季的264.22亿美元相比增长21%；按照非公认会计准则（Non- GAAP）计算，净利润为317.67亿美元，与上年同期的200.1亿美元相比增长59%，与上一财季的257.88亿美元相比增长23%。 毛利率为73.4%，与上年同期的74.6相比下降了1.2%，但与上一财季的72.4%相比增长了1%。按照非公认会计准则计算，毛利率为 73.6%，与上年同期的75相比下降了1.4%，但与上一财季的72.7%相比增长了0.9%。 运营支出为58.39亿美元，与上年同期的42.87亿美元相比增长36%，与上一财季的54.13亿美元相比增长8%。按照非公认会计准则计算， 运营支出为42.15亿美元，与上年同期的30.4 ...

11月20日，美东时间周三盘后，人工智能（AI）龙头股英伟达公布了超出预期的三季度财报。在美股 市场因估值担忧而持续低迷之际，英伟达再次不负众望，交出了一份优异答卷。 英伟达周三收盘时上涨近3%。截至发稿，随着财报公布，英伟达股价盘后进一步上涨超6%。 财报显示，英伟达第三财季营收为570.1亿美元，超出市场预期的549.2亿美元；净利润为319.1亿美元， 同比大涨65%，经调整后的每股收益为1.30 美元，高于市场预期的1.25美元。 数据中心——英伟达最重要的业务——Q3营收为512亿美元，轻松超过分析师预测的490.9亿美元，同比 增长66%。其中，"计算"（即GPU）业务贡献了430亿美元营收；网络业务（即允许多个GPU能协同工 作的组件）贡献了82亿美元。 英伟达首席财务官Colette Kress在声明中表示，公司当前最畅销的芯片系列是Blackwell Ultra，这是 Blackwell芯片的第二代产品。 英伟达CEO黄仁勋还在声明中提到："云端GPU已经全部售罄。" 这一表态回应了投资者的担忧，即英伟达快速增长的营收高度依赖于少数几家大型云服务商（即超大规 模客户），而这些客户必须找到终端 ...

Core Insights - Nvidia will hold the 2026 GTC conference from March 16 to 19, 2026, in San Jose, USA [1] - CEO Jensen Huang shared Nvidia's latest strategies, technological breakthroughs, and industry vision in AI during the 2025 GTC conference [1] - The core viewpoint is that AI is transitioning from "generation" to "reasoning" and "action," with a deeper integration into the physical world [1] Technical Developments - The reasoning process of generative AI is expected to produce 100 times more tokens than before, with computational demands projected to be at least 100 times higher than anticipated a year ago [1] - The conference also introduced the Blackwell Ultra chip and outlined the future architecture roadmap for Rubin and Feynman [1]

Investment Rating - The report rates the industry as "Outperforming the Market" [5] Core Insights - The high power consumption of AI chips is accelerating the adoption of liquid cooling technology, with cold plate solutions becoming mainstream in data centers [1] - Policy incentives and the demand for computing power are driving the rapid growth of the liquid cooling industry, with significant market potential anticipated by 2025 [2] - The penetration rate of liquid cooling technology is still in its early stages, but is expected to see rapid growth as high-power chips become more widely used [3] Summary by Sections Section 1: Industry Dynamics - The latest Blackwell Ultra chip from NVIDIA has a power consumption of 1.4kW, leading to a total rack power consumption of 120-130kW, making traditional air cooling inadequate [1] - Liquid cooling technology is projected to be adopted in over 50% of new data center projects by the end of 2025, driven by policies such as China's "East Data West Computing" initiative [2] Section 2: Market Growth - The Chinese liquid cooling server market is expected to grow at a compound annual growth rate (CAGR) of 46.8% from 2024 to 2029, potentially reaching a market size of $16.2 billion by 2029 [2] - The global penetration rate of liquid cooling in data centers is currently around 10%, with expectations to rise to over 30% in China by 2025 [3] Section 3: Industry Opportunities - The increasing shipment of NVIDIA's Blackwell series chips is a key driver for the adoption of liquid cooling, with projections indicating that Blackwell will account for over 80% of high-end GPU shipments by 2025 [3] - The report highlights several companies involved in the liquid cooling supply chain, including Xiangxin Technology and Kexin New Source, as potential beneficiaries of this growth [3]

Core Viewpoint - TSMC reported a record profit growth of 60.7% in Q2, marking six consecutive quarters of profit increase, driven by strong demand in the semiconductor industry, particularly from AI-related applications [1][2]. Group 1: Financial Performance - TSMC's net profit for the quarter ending June 30 was NT$398.3 billion (approximately $13.53 billion) [2]. - The company's revenue grew by 38.6% in the second quarter [2]. - TSMC's market capitalization is approximately $1.23 trillion, placing it among the top ten highest-valued companies globally [2]. Group 2: Market Demand and Client Relationships - The demand for semiconductors is expected to grow by 17% to 18% this year, with TSMC likely to see sales growth of nearly 30% due to its market leadership [2]. - NVIDIA, TSMC's largest client, is ramping up production of its new Blackwell Ultra chips and is set to resume exports of its AI chip H20 to China [2][3]. - Analysts believe that the resumption of H20 supplies will further boost TSMC's order growth [2][3]. Group 3: Equipment and Supply Chain Impact - TSMC's order volume significantly impacts upstream equipment manufacturers, such as ASML, which is expected to receive substantial orders for its upcoming N2 process equipment [3]. - Current demand from TSMC and Chinese chip manufacturers is exceeding expectations, while demand from Intel and Samsung is below expectations [3]. Group 4: Investment and Expansion Plans - TSMC announced a $100 billion investment plan, including a commitment of $65 billion to build three factories in Arizona, two of which are already completed [4].

Group 1 - The overall revenue of the wafer foundry industry is expected to grow by 17% to 18% this year, driven by the increasing demand for artificial intelligence [1][3] - TSMC's sales are projected to grow nearly 30%, benefiting from its market leadership [1][3] - TSMC reported a net profit of NT$398.3 billion (approximately US$13.53 billion) for the quarter ending June 30, with a profit growth of 60.7%, marking a historical high [3] Group 2 - Major clients of TSMC include chip giants NVIDIA and Apple, with NVIDIA being the largest customer [3][4] - NVIDIA is pushing for mass shipments of its next-generation Blackwell Ultra chips and is expected to resume exports of its AI chip H20 to China soon [3] - Analysts believe that the resumption of H20 supplies to China will boost TSMC's order growth [3][4] Group 3 - TSMC's order volume significantly impacts upstream equipment manufacturers, such as ASML, which is TSMC's largest supplier [4] - TSMC is expected to order equipment for the upcoming N2 process from ASML [4] - Demand and orders from TSMC and Chinese chip manufacturers are currently better than expected, while demand from Intel and Samsung is below expectations [4] Group 4 - TSMC announced a US$100 billion investment in the U.S., with a commitment to invest US$65 billion in building three factories in Arizona, two of which are already completed [5]