Core Viewpoint - Nvidia reported strong earnings but experienced a significant stock sell-off, losing nearly 10% over two days despite beating estimates and providing positive guidance for the upcoming quarter [1][2]. Financial Performance - Nvidia's earnings per share (EPS) for fiscal 2027 is projected at $8.23, with a forward P/E ratio of 21.5 based on a closing price of $177.19 [6]. - The company achieved a remarkable 73% revenue growth in the fourth quarter, with expectations for a 69% increase in revenue to $364.8 billion this year and a 73% rise in EPS [7]. Market Comparison - Nvidia is now trading at a lower valuation compared to the S&P 500, which has a forward P/E ratio around 22, indicating a potential mispricing given Nvidia's higher growth rate [6][11]. - The semiconductor sector, including Nvidia, is currently undervalued compared to more stable sectors like software, which typically command higher premiums due to their subscription models [10]. Investor Sentiment - The sell-off may reflect investor concerns about Nvidia's valuation and a shift towards undervalued software stocks, despite Nvidia's strong performance [3][10]. - There are ongoing questions regarding the sustainability of AI spending, particularly as major companies are set to invest over $600 billion in capital expenditures, impacting their free cash flow [3]. Historical Context - Nvidia has a track record of exceeding analyst expectations, with revenue growth accelerating unexpectedly in the past year [12]. - The upcoming launch of the new Vera Rubin platform is anticipated to further bolster Nvidia's growth prospects [11].

来源：搜狐财经 2月21日，英伟达公布了2026财年第四季度财报，再次以碾压式的数字震撼市场。总营收高达681.3亿美 元，同比增长73%；GAAP净利润达到429.6亿美元，同比猛增94%——相当于平均每天净赚超过4.7亿美 元。营收、利润、现金流三项核心指标均创历史单季新高。 这份财报的核心逻辑只有一个词：AI。所有增长几乎全部由数据中心业务驱动，其单季收入达到623.1 亿美元，占总营收比例历史性地突破了91.5%。这个数字意味着，英伟达已经从一家"游戏显卡公司"彻 底变为全球AI基础设施的核心供应商。 数据中心业务内部，有一个更值得关注的结构性变化：网络业务（Networking）的收入同比暴增 263%，达到110亿美元。这背后是NVLink互联技术的爆发，英伟达通过"计算芯片+高速网络"的系统级 方案，正在把自己的护城河越挖越深。CEO黄仁勋将其描述为"AI工厂"的基础设施：不只是卖一块芯 片，而是卖一套完整的"算力工厂"解决方案。 在一片叫好声中，市场的目光更聚焦于AI的下一战场——推理（Inference）。这里的推理是指AI模型训 练完成后，真正被用户调用、生成回答的过程，所消耗的算力就是推 ...

当地时间2月26日，美股三大指数收盘涨跌不一，大型科技股多数下跌，英伟达（NVDA）股价大幅跳水，单日跌幅超5%，市值一夜之间蒸发超2700亿美 元（约合人民币1.77万亿元），创下自去年4月16日以来近11个月的最大单日跌幅。截至当日收盘，英伟达股价报184.89美元，较前一交易日下跌10.67美 元，跌幅达5.46%，换手率1.54%，成交量达3.6亿股，成交额675亿美元，创下近期单日成交量新高，总市值降至4.49万亿美元。 英伟达股价表现 | 今开 | 194.270 | | 最高 | 194.290 | | 成交量 | 3.60亿股 | | --- | --- | --- | --- | --- | --- | --- | --- | | 昨收 | 195.560 | | 最低 | 184.315 | | 成交额 | 674.83亿 | | 换手率 | 1.48% | | 市盈(TTM) | 37.42 | | 总市值 | 4.49万亿 | | 分时 | 王日 | 日K | 周K | 月K | 案K | 年K | 更多v | 2月25日，英伟达刚刚发布2026财年第四财季及全年财报，财报数据显示，公 ...

智通财经APP注意到，英伟达(NVDA.US)第四季度财报不仅以显著优势超越预期，其业绩指引甚至让最 看涨的分析师也感到意外。由黄仁勋领导的这家公司预计下一季度销售额将达到780亿美元，远高于华 尔街728亿美元的预测。 然而，该公司股价周四盘前交易中仅上涨约1%。华尔街分析师认为股价还有更多上涨空间：只是需要 更多时间。 摩根大通分析师哈兰.苏尔在给客户的报告中写道："英伟达显然在内部及其庞大供应链的各个环节都火 力全开，在继续提升Blackwell平台(GB200+GB300)产量的同时，也在为其Vera Rubin平台的即将上量(计 划在2026年下半年)做准备。即便如此，股价反应表明投资者仍不满足，我们认为这与英伟达数据中心 业务在2027日历年增长轨迹的持续不确定性有关，考虑到其主要客户(美国五大超大规模企业的资本支 出总额预计在2026日历年同比增长约70%至6500亿美元以上)大幅扩张的资本支出预算以及显著压缩的 自由现金流状况。但退一步看，该股在财报电话会议前的交易价格约为华尔街2027日历年预期每股收益 的19倍，在我们看来，英伟达就像一根被压紧的弹簧，在这份业绩公布后绷得更紧了。" 小摩给 ...

营收681.3亿美元！ 刚刚，英伟达公布了创纪录的2026财年第四季度业绩，并给出了2027财年第一季度780亿美元的业绩预期。 Vera Rubin平台：六款新芯片，一台AI超级计算机 为了把这些惊人的算力需求变现，英伟达还推出了下一代整柜级算力系统——Vera Rubin，将在2026年下半年出货。 英伟达首席财务官Colette Kress透露，公司已经清晰预见到，2025至2026年间Blackwell和Rubin产品组合带来5000亿美元营收。 超预期的四季度业绩 商业世界里，数据永远是最有说服力的语言。 截至周三收盘，英伟达的股价累计上涨了5%，而同一时期，科技股的晴雨表纳斯达克指数下跌了0.4%。 当华尔街和硅谷还在为AI泡沫争论不休时，黄仁勋再次用数据击碎了「AI泡沫论」。 不仅如此，他还提出了自己的「AI经济学」：算力=收入增长，并直言如今流向AI的千亿美元资本开支，最终都会直接转化为收入。 支撑老黄「AI经济学」的，是他看到了智能体拐点已经到来，由此带来了惊人的算力需求。 在2026财年第四季度，英伟达一举冲破了市场的预期线，当季总营收高达681.3亿美元，同比增幅达到了惊人的73%。 公 ...

证券研究报告|行业简评报告 2026 年 02 月 26 日 本季营收与指引均高增，战略备货以满足未来市场需求 英伟达（NVDA.O）FY26Q4 跟踪报告 TMT 及中小盘/电子 事件： 英伟达发布 FY26Q4 季报，本季营收 681 亿美元，同比+73%/环比+20%，创 历史新高；non-GAAP 毛利率为 75.2%，同比+1.7pcts/环比+1.6pcts。综合财 报及交流会议信息，总结要点如下： 评论： 1、FY26Q4 营收高增再超预期，公司开启战略备货以应对未来市场需求。 FY26Q4 营收 681 亿美元，同比+73%/环比+20%，超出此前预期（650 亿美 元），本季营收、营业利润和自由现金流均创历史新高；non-GAAP 毛利率为 75.2%，符合此前指引（74.5%-75.5%），同比+1.7pcts/环比+1.6pcts，环比 增长主要得益于 Blackwell 架构产能持续爬坡；本季库存环比增长 8%，采购承 诺也大幅增加，公司已战略性储备库存并锁定产能，以满足未来数个季度的市 场需求。 2、数据中心营收再创新高网络业务表现亮眼，游戏受供应链影响环比下降。 1）数据中心：营收 ...

Core Insights - Nvidia reported record revenue of $68.127 billion for Q4, a 73% increase from $39.331 billion year-over-year, and a net profit of $42.96 billion, up 94% from $22.091 billion [1] - For the full fiscal year, Nvidia's revenue reached $215.938 billion, with a net profit of $120.067 billion, equating to approximately $328 million per day [1] - Nvidia's performance serves as a barometer for AI demand, indicating that for leading players, there is no downturn, only a resurgence [1] Financial Performance - Nvidia's Q4 revenue of $68.127 billion is a significant milestone, reflecting the ongoing high costs associated with AI [3] - The data center business contributed $62.3 billion in Q4, a 75% year-over-year increase, accounting for over 91% of total revenue [3][4] - Nvidia's full-year revenue surpassed $200 billion for the first time, reaching $215.938 billion [4] Market Dynamics - Nvidia's CEO expressed confidence in the growth of customer cash flows, attributing it to the recognition of the value of Agentic AI across various enterprises [2] - Major cloud providers like Google, Amazon, Meta, and Microsoft are significantly increasing their capital expenditures, with a projected combined spending of nearly $700 billion by 2026 [3] Strategic Initiatives - Nvidia aims to establish a comprehensive AI ecosystem on its platform, encompassing various sectors such as AI, robotics, and life sciences [5] - The company is nearing an agreement with OpenAI for a potential $100 billion AI infrastructure project and has acquired technology from AI startup Groq for approximately $20 billion [5] - Nvidia acknowledges the competitive landscape in China, where local companies are making significant advancements [6] Industry Trends - A McKinsey survey indicates that over 70% of CIOs at large enterprises plan to double their technology spending between 2026 and 2027, with 70% of budgets redirected towards AI [8] - The ROI of AI remains elusive, with clients demanding significant productivity improvements in exchange for large orders [8] - The emergence of Agentic AI is drastically reducing development costs, allowing single individuals to complete tasks that previously required entire teams [9] Future Outlook - Nvidia's inventory is fully booked until 2027, with seamless transitions between product iterations [10] - The company is set to begin mass production of its next-generation Vera Rubin platform in the second half of the year, anticipating widespread deployment among cloud model builders [10]

Core Viewpoint - Meta and Nvidia have established a long-term partnership focusing on local deployment, cloud, and AI infrastructure, marking a significant expansion of their technological collaboration [1][5]. Group 1: Partnership Details - Meta will build ultra-large-scale data centers optimized for training and inference to support its long-term AI infrastructure roadmap [3]. - The collaboration will involve the deployment of millions of Blackwell and Rubin GPUs, as well as Nvidia's Grace CPUs, with Nvidia's Spectrum-X Ethernet switches integrated into Facebook's open switching system [3]. - This partnership represents the first large-scale deployment of Nvidia's Grace and aims to enhance the energy efficiency of AI computing [3][6]. Group 2: Strategic Implications - Meta is expected to allocate a significant portion of its projected capital expenditure of up to $135 billion this year towards expanding Nvidia's data centers [7]. - The large-scale adoption of Nvidia's chips validates Nvidia's "full-stack" infrastructure strategy, which includes both CPU and GPU [7]. - Despite the partnership, Meta is also exploring alternatives, including the potential use of Google's Tensor Processing Units (TPUs) in its data centers by 2027 [7][8]. Group 3: Market Reactions - Following the announcement, both Meta and Nvidia's stock prices rose in after-hours trading, while AMD's stock fell over 4% [3].

Core Insights - Cisco has launched a new networking chip, Silicon One G300, designed to accelerate information transmission within large data centers, potentially competing directly with products from Broadcom and NVIDIA [1][2] - The Silicon One G300 chip can deliver 102.4 Terabits per second, powering AI clusters at gigawatt scale, enhancing GPU utilization, and improving task completion time by 28% [1] - Cisco's new systems, N9000 and 8000, will utilize the Silicon One G300 chip and are tailored for hyperscale cloud providers, emerging cloud services, sovereign clouds, service providers, and enterprise customers [1] - The new systems will feature a 100% liquid cooling design, combined with new optical technologies, aiming to improve energy efficiency by nearly 70% [1] - Cisco has enhanced its Nexus One data center network architecture to facilitate easier operation of AI networks for enterprises, whether on-premises or in the cloud [1][2] Industry Context - As AI training and inference scale up, data movement has become critical for efficient AI computing, with networks now considered part of the computing infrastructure [2] - Cisco's Silicon One G300 chip is designed to provide high-performance, programmable, and deterministic network experiences, enabling customers to fully leverage their computing resources securely and reliably in production environments [2] - NVIDIA recently announced its next-generation AI computing platform, Vera Rubin, which includes key networking and infrastructure components, while Broadcom has begun shipping its Tomahawk 6 series switching chips [2] - Cisco has introduced a series of features to help enterprises securely adopt AI technologies while maintaining the integrity of AI agents and control over agent interactions [2] Security Features - AI Bill of Materials: Provides centralized visibility and governance for AI software assets, including model context protocol servers and third-party dependencies, to ensure AI supply chain security [3] - MCP Catalog: Discovers, inventories, and helps manage risks associated with MCP servers and registries across public and private platforms, enhancing AI governance [3] - Advanced Algorithm Red Team Testing: Expands the scope of AI security assessments [4] - Real-time Agent Safeguards: Ensures the security of agents and applications [4]

Group 1 - The domestic computing power is in a very favorable position, with a shift in focus towards high-performance and cost-effective chips due to changing industry demands [1][5] - The third-generation inference GPU chip, S3, was launched by Xiwang, aiming to reduce the cost of one million tokens to one cent, reflecting the industry's transition from training to inference [3] - By 2030, it is expected that inference chips will account for 80% of the company's resource allocation, indicating a strategic focus on optimizing inference capabilities [3] Group 2 - The integrated training and inference chips face challenges such as high costs, unstable supply, and complex deployment, highlighting the need for a reasonable computing power to memory access ratio [4] - The "memory wall" has become a significant bottleneck in chip performance, as the speed of computing unit enhancements outpaces memory bandwidth improvements, particularly in inference chips [4] - Companies like DeepSeek are driving innovation across the entire technology chain, from model architecture to inference systems, aiming to reduce dependency on NVIDIA's CUDA ecosystem [4] Group 3 - The reduction of costs in AI applications significantly boosts the number of applications in the market, with the domestic computing power positioned advantageously to capitalize on this trend [5]