来源：光大证券微资讯 英伟达发布重磅财报，科技行业又迎机遇！ 美国当地时间2月25日，全球投资者都在关注的英伟达财报重磅发布，2026财年第四季度英伟达取得丰 硕成果，营业收入高达681亿美元，刷新历史纪录；GAAP净利润达到429.6亿美元，毛利率高达75%。 在发布大超预期的财报之前，英伟达相关负责人展示了Vera Rubin机架的细节，涉及130万个组件、80 多家供应商。 展望未来，生成式AI带来算力需求的大幅增长，叠加英伟达不断提升技术水平，全球算力产业链依然 维持较高景气度。 1、英伟达财报大超预期，营收指引高达780亿美元 无论对于总市值高达4.75万亿美元的英伟达能否站稳当前位置，还是市场担忧人工智能出现较大泡沫， 英伟达四季报都是关键影响因素。 在NVLink技术的加持下，英伟达Grace Blackwell芯片继续占据推理算力的大部分市场，该产品也让算 力成本快速下降。 中国市场一直是英伟达的核心区域，在H20获得出口许可后，该芯片的销售收入为6000万美元。H200芯 片尽管已被允许向中国客户出口，但是仍未产生收入。美国的出口政策让英伟达在中国市场处于被动局 面，再加上国产GPU厂商快速 ...

2026财年全年，英伟达营收达到2159.4亿美元（约合人民币14834.0亿元），同比增长65%；净利润则达 到1200.8亿美元（约合人民币8248.9亿元），同比增长65%。 据美国消费者新闻与商业频道（CNBC）、南华早报、英伟达网站消息，当地时间2月25日，英伟达公 布了截至2026年1月25日的2026财年第四季度财报以及2026财年全年业绩。 本财季，英伟达营收和净利润双双超越华尔街预期，营收达681.3亿美元（约合人民币4680.2亿元），同 比增长73%，环比增长20%；净利润（非GAAP）达395.5亿美元（约合人民币2716.9亿元），同比增长 79%，环比增长25%。 其中，数据中心营收达623亿美元，环比增长22%，同比增长75%。报道称，该公司超过91%的销售额来 自数据中心部门，该部门包含其市场领先的人工智能芯片。 克雷斯还称，"我们在中国的竞争对手，在近期IPO的推动下正在取得进展，从长期来看，有可能颠覆 全球AI行业的格局。" 据悉，英伟达对2027财年第一季度的展望为预计实现营收780亿美元，上下浮动2%；预计GAAP和非 GAAP毛利率分别为74.9%和75.0%，上下浮 ...

Core Viewpoint - Nvidia's transition to the new Blackwell AI chips has faced significant deployment challenges, particularly for major clients like OpenAI and Meta, but the company has managed to maintain its market position and address many technical issues [2][3][4]. Group 1: Deployment Challenges - Nvidia's CEO Jensen Huang indicated that the complexity of the new Blackwell AI chips would make the transition from the previous generation challenging for clients, requiring adjustments across various system components [2]. - Major clients, including OpenAI and Meta, struggled with the deployment and operation of Blackwell servers, which contrasted sharply with the quicker deployment of previous Nvidia AI chips [2][3]. - Despite these challenges, Nvidia's business has not been severely impacted, maintaining a market capitalization of $4.24 trillion and resolving many technical issues hindering client deployment [2][3]. Group 2: Client Reactions and Adjustments - Clients like OpenAI and Meta have expressed private dissatisfaction regarding the inability to build chip clusters at the expected scale, which limits their capacity to train larger AI models [3][4]. - To address client dissatisfaction, Nvidia provided refunds and discounts related to issues with the Grace Blackwell chips [3][4]. - Nvidia has collaborated closely with leading cloud service providers to improve the deployment process, indicating a commitment to joint engineering development [4]. Group 3: Product Improvements - Nvidia has learned from the deployment challenges and has optimized the existing Grace Blackwell systems while also improving the upcoming Vera Rubin chip servers [5]. - An upgraded version of the Grace Blackwell chip, named GB300, has been introduced to enhance stability and performance, addressing issues encountered with the first generation [5]. - Some clients have adjusted their orders to the upgraded products, indicating a shift in demand towards improved chip versions [5]. Group 4: Financial Implications - Delays in chip deployment have led to financial losses for cloud service partners of OpenAI, who invested heavily in Grace Blackwell chips expecting quick returns [9][10]. - Some cloud service providers negotiated discount agreements with Nvidia to alleviate financial pressure due to delayed chip usage [9]. - Oracle reported significant losses in its AI cloud business due to the slow deployment of Blackwell chips, highlighting the financial risks associated with new technology launches [10].

Core Insights - Jensen Huang, CEO of Nvidia, emphasizes that the company's success is not based on GPU production volume but rather on its unique corporate culture and innovation capabilities [1][24] - Huang predicts that AI investments will fundamentally change computing, leading to computers that can learn autonomously under human guidance, resulting in a transformation of job roles rather than a reduction in employment [2][41] - Nvidia's management structure is designed to foster a safe environment where mistakes are tolerated, allowing for innovation and growth without fear of termination [1][25] Group 1: Company Philosophy and Culture - Nvidia has cultivated a culture where no one is fired for making mistakes, fostering a safe environment for innovation [1][25] - The company has a unique management structure with nearly 61 individuals acting as "CEOs," each deeply committed to the company's mission [1][18] - Huang believes that the essence of Nvidia's success lies in its corporate character and the ability to unite the team in adversity [24] Group 2: Vision for the Future - Huang asserts that in five years, AI will enable computers to handle problems a billion times larger than current capabilities, fundamentally altering the nature of work [38][39] - The future will see an increase in productivity and efficiency across industries, with AI solving previously insurmountable challenges [40][41] - Huang anticipates that while job roles will evolve, the overall number of jobs will not decrease, and AI will provide new opportunities for those currently unemployed [41][44] Group 3: Historical Context and Personal Insights - Nvidia's journey has spanned 33 years, with a consistent focus on reshaping the computing industry since its inception [5][16] - Huang reflects on the importance of learning from past decisions and maintaining a flexible approach to leadership and strategy [14][15] - The company has a history of making bold decisions, such as the early adoption of CUDA technology, which laid the groundwork for its current success [6][8]

Core Viewpoint - Jensen Huang, CEO of NVIDIA, emphasizes that the company's success is not solely based on production volume but rather on its unique corporate culture and the ability to innovate and adapt in the tech industry [2][33]. Group 1: Company Philosophy and Leadership - NVIDIA fosters an environment where mistakes are accepted, and no one is fired for errors, which contributes to a culture of learning and resilience [34]. - Huang describes the role of CEO as fragile and emphasizes the importance of humility and continuous learning within the company [2][22]. - The company has a unique management structure with nearly 61 individuals acting as "CEOs," reflecting a collaborative leadership approach [17][27]. Group 2: Technological Vision and Future Trends - Huang predicts that AI investments will fundamentally change how computers operate, evolving from being programmed by humans to learning autonomously under human guidance [3][49]. - The future will see a significant increase in productivity and efficiency across industries, with AI enabling the resolution of complex problems that were previously deemed unsolvable [50][52]. - Huang believes that while job roles will change, there will not be a significant loss of jobs; instead, AI will create new opportunities for those currently unemployed [52][54]. Group 3: Historical Context and Company Evolution - NVIDIA has been on a 33-year journey to reshape the computing industry, with a focus on innovation and market strategy since its inception [8][9]. - The company has consistently prioritized technological advancement and product innovation, which has allowed it to maintain a competitive edge despite being a smaller GPU manufacturer [33][34]. - Huang reflects on the importance of foresight and strategic planning in the company's success, highlighting the need to be ahead of technological trends [11][12].

Core Insights - NVIDIA plans to build 20 AI factories in Europe and establish the world's first "industrial AI cloud" in the region [1] - The AI computing capacity in Europe is expected to increase tenfold within two years [1] - Quantum computing technology is at a turning point, with potential applications to solve significant global issues in the coming years [1] Group 1: AI Infrastructure Development - NVIDIA's CEO Jensen Huang announced the construction of 20 AI factories in Europe [1] - The first industrial AI cloud, equipped with 10,000 GPUs, will be established in Germany [1] - NVIDIA is forming alliances with various European companies, including the French startup Mistral AI [1] Group 2: Quantum Computing Advancements - Huang expressed optimism about the rapid advancement of quantum computing technology, which has been in development for decades [1] - Quantum computers can process information at speeds significantly higher than traditional computers due to their ability to perform parallel computations [1] - Following Huang's positive outlook, stocks of companies involved in quantum technology saw a rise, with Quantum Computing stocks increasing by 12.5% [1]

Core Insights - Nvidia is expanding its AI infrastructure projects in Europe, including a partnership with French startup Mistral AI to enhance local AI computing capabilities [1][2] - The company aims to address the lack of infrastructure in Europe, which is lagging behind the US in AI development and investment [2] - Nvidia plans to establish over 20 AI factories across Europe in the next two years, significantly increasing the region's AI hardware capacity [2] Group 1 - Nvidia's CEO Jensen Huang announced the need for data centers in Europe to facilitate AI technology deployment [1] - The collaboration with Mistral AI will utilize 18,000 new Grace Blackwell chips in a service called Mistral Compute, which will be developed in Mistral's data center in France [1] - Other countries, including the UK, Italy, and Armenia, are also installing new Nvidia hardware to enhance their AI capabilities [1] Group 2 - Nvidia is collaborating with 1.5 million developers, 9,600 enterprises, and 7,000 startups in Europe to build AI infrastructure [2] - The company plans to increase Europe's AI computing capacity by tenfold, with an estimated tripling of AI hardware production in the region next year [2] - Major companies like Microsoft and Meta contribute approximately half of Nvidia's sales, indicating a strong market presence [2] Group 3 - Nvidia's Lepton service will assist AI developers in connecting with necessary computing hardware, with participation from companies like AWS and Mistral [3] - The company emphasizes the need for AI models based on local languages and data, providing software and services to accelerate these initiatives [3] - Vehicles equipped with Nvidia's chips and software, such as models from Mercedes-Benz, Volvo, and Jaguar, are beginning to hit the roads [3]

Group 1 - The core point of the article is that the AI lab Mistral plans to utilize 18,000 Nvidia Grace Blackwell chips for its computing equipment [1] Group 2 - Mistral is focused on advancing its capabilities in artificial intelligence through the deployment of high-performance computing resources [1] - The use of 18,000 chips indicates a significant investment in technology infrastructure, which may enhance Mistral's competitive edge in the AI industry [1] - Nvidia's Grace Blackwell chips are designed to optimize AI workloads, suggesting that Mistral is aligning its hardware choices with the demands of modern AI applications [1]

Core Viewpoint - Nvidia is shifting its focus towards the low-end market in the telecom sector, promoting its ARC-Compact chip for distributed RAN, which is less powerful than its previous offerings but is marketed as cost-effective and energy-efficient for low-latency AI workloads [1][2]. Summary by Sections Nvidia's Strategy - Nvidia has not abandoned its efforts to sell AI chips to the telecom industry, despite limited interest so far [1]. - The ARC-Compact is designed for installation at cell sites, contrasting with the previous ARC servers aimed at centralized RAN [1]. Technical Specifications - The main components of ARC-Compact include the Grace CPU and L4 Tensor Core GPU, which are lightweight and suitable for edge video processing but lack the capability for large language model training [2]. - Nvidia describes ARC-Compact as an "economical and energy-efficient" option for low-latency AI workloads and RAN acceleration [2]. Market Competition - Major RAN suppliers like Ericsson, Nokia, and Samsung have invested in virtual RAN technology but show limited interest in adopting Nvidia's CUDA for RAN development [4]. - These suppliers prefer a "lookaside" virtual RAN model to maintain hardware independence, keeping most software on the CPU [4]. Supplier Insights - Ericsson has successfully migrated software for Intel x86 CPUs to Grace with minimal changes, indicating potential for GPU use only in specific tasks like forward error correction (FEC) [5]. - Samsung has tested its software on Grace but denies the need for inline accelerators, suggesting that CPU capacity will suffice as technology advances [5]. Nokia's Position - Unlike Ericsson and Samsung, Nokia has invested all its virtual RAN resources into inline acceleration but acknowledges that its first-layer accelerator comes from Marvell Technology, not Nvidia [6]. Industry Perception - A survey by Omdia revealed that only 17% of respondents believe most AI processing will occur at base stations, with 43% favoring end-user devices [8]. - The telecom industry appears to be in a challenging position between device capabilities and large-scale cloud platforms, with low demand for ultra-low latency services in medium-sized countries [9]. Future Outlook - The emergence of Grace is timely as doubts about Intel's future as a virtual RAN CPU provider grow, allowing RAN suppliers to demonstrate independence from underlying hardware [9]. - There is a potential shift in AI processing focus from GPUs to more powerful CPUs, as model sizes decrease and machines handle critical AI workloads [10].