当潮水退去，才知道谁在裸泳。 出品 | 新行情 作者 | 松涛 来源：市场资讯 （来源：新行情） 今年10月，英伟达市值首次突破5万亿美元，规模超过了英、法、德等主要经济体的股市总值，也将市 场对AI算力的乐观情绪推至高潮。 然而高峰过后，形势急转直下。 短短一个多月内，英伟达市值连续经历数次大幅下跌，万亿市值蒸发，再次成为舆论焦点。 英伟达市值的剧烈波动，不仅仅是单只股票的涨跌，更反映了市场情绪的集体转变：资本从狂热追逐逐 渐转为冷静审视。 其中也透露出市场的担忧：天价芯片订单是否依赖脆弱的融资循环？激烈的竞争与国际环境波动，是否 在侵蚀其护城河？就连黄仁勋本人关于"未向OpenAI付款"的澄清，也被市场解读出了别样的信号。 一场关于"AI泡沫"的疑虑，正随着英伟达市值的剧烈波动而扩散。风暴尚未停息，它迫使整个行业在狂 飙突进之后，必须直面那些悬而未决的根本问题。 英伟达市值暴跌 12月18日，英伟达股价报收于170.94美元，当日下跌3.81%，市值在一夜之间蒸发1648亿美元（约合 1.16万亿人民币）。 | 170.940 " | 176.130 昨收 177.720 量比 1.29 | | | -- ...

智通财经APP获悉，英伟达(NVDA.US)周一宣布收购AI软件公司SchedMD。这家芯片巨头正通过加大 开源技术投入、强化AI生态布局，以应对日益激烈的行业竞争。 虽然英伟达凭借高性能芯片享誉业界，但其同样提供涵盖物理模拟、自动驾驶等多个领域的自有AI模 型，并将这些模型作为开源软件向研究机构和企业开放。其专有的CUDA软件已成为多数开发者的行业 标准，这正是英伟达芯片的核心卖点，软件实力也成为其维持AI行业领导地位的关键。 英伟达在官方博客中表示："支持最新英伟达硬件的Slurm系统已成为生成式AI关键基础设施的重要组成 部分，被基础模型开发者与AI构建者广泛用于管理模型训练和推理需求。" 值得一提的是，同日早些时候，英伟达发布了全新开源AI模型系列。面对中国AI实验室开源模型的竞 争浪潮，该公司称新系列模型较先前版本速度更快、成本更低且性能更智能。 据SchedMD官网信息显示，该公司由Slurm软件开发者Morris "Moe" Jette与Danny Auble于2010年在加利 福尼亚州利弗莫尔创立，目前拥有40名员工。其客户包括云基础设施公司CoreWeave(CRWV.US)与巴塞 罗那超级 ...

SchedMD提供的软件可帮助调度大型计算任务，这类任务往往会占用数据中心服务器容量的很大一部 分。该公司的Slurm开源技术允许开发者和企业免费使用，通过提供工程与维护支持服务盈利。 英伟达 凤凰网科技讯北京时间12月16日，据路透社报道，英伟达周一表示，已收购AI软件公司SchedMD。随 着竞争的加剧，英伟达正加倍押注开源技术，并加大对AI生态系统的投资以应对挑战。 虽然英伟达以开发高速芯片著称，但它也提供一系列自有AI模型，从物理模拟到自动驾驶汽车，并将 其作为开源软件供研究人员和公司使用。英伟达专有CUDA软件是大多数开发者采用的行业标准，也是 英伟达芯片的一大卖点，使得软件成为其维持在AI行业主导地位的关键。 这笔交易的财务条款未予披露。英伟达表示，将继续以开源方式分发SchedMD的软件。 在宣布这个收购消息并发布新开源（300109）AI模型后，英伟达股价上涨1.35%。(作者/箫雨) ...

英伟达（NVDA）将于 2025 年 11 月 19 日发布 2025 财年第三季度（FQ3）财报。 回顾上一季度，公司调整后每股收益（EPS）为 1.05 美元，公认会计原则（GAAP）下每股收益 1.08 美元，总营收 467.4 亿美元，均超出市场普遍预期。对于即将到来的季度，调整后每股收益 预计将达 1.25 美元，环比增长 19%；营收预计达 548 亿美元，环比增长 17%。值得注意的是，调整后每股收益增速超过营收增速，这意 味着净利润率有望扩大 —— 预计下一季度调整后净利润率将从之前已十分出色的 55.3% 进一步提升至 56.2%。 【如需和我们交流可扫码添加进社群】 英 伟 达 ： 未 来 利 润 率 将 保 持 稳 健 展望未来，分析师对英伟达未来几年的财务前景极为乐观，不仅预期其将实现大幅增长，还将维持强劲的盈利能力。具体而言，根据当前市场 共识， 公司未来 5 年每股收益复合年增长率（CAGR）预计为 22.0%，营收复合年增长率为 20.2%。 按照这一增速，其每股收益和营收在 这段时间内都将实现翻倍以上增长。再次需要注意的是，每股收益增速高于营收增速，这预示着利润率存在扩张潜力。 ...

Core Insights - The article emphasizes Nvidia's pivotal role in the AI revolution, highlighting its rapid growth and strategic investments in various AI sectors since the emergence of ChatGPT [4][5]. - Nvidia's investment strategy is characterized by a broad approach, targeting multiple AI technologies and companies to secure its position as a foundational player in the AI ecosystem [10][24]. Investment Strategy - Nvidia has participated in 50 venture capital deals in 2025 alone, surpassing its total of 48 deals in 2024, indicating an aggressive investment pace [5]. - The company has made significant investments in OpenAI, totaling $100 billion, to develop large-scale AI infrastructure, showcasing its commitment to being a key player in AI development [6][7]. - Nvidia's strategy includes investing in competing AI firms, such as xAI, to diversify its technological bets rather than aligning with a single winner [7][10]. AI Ecosystem Development - Nvidia is building a comprehensive AI ecosystem by investing in companies across various sectors, including data centers, energy solutions, and foundational technologies [16][24]. - The company has invested in firms like Ns c a l e and Cr us oe, which focus on AI data centers, ensuring that Nvidia's GPUs remain integral to AI operations [12][13]. - Nvidia's investments in energy technologies, such as nuclear fusion, aim to address the significant energy demands of AI, positioning the company as a leader in sustainable AI infrastructure [14][15]. Application Areas - Nvidia's investments span multiple AI application areas, including autonomous driving, robotics, healthcare, and content creation, ensuring its technology is embedded across industries [18][22]. - The company has backed various startups in the autonomous driving sector, such as Wa yve and Nu r o, reflecting its long-term vision for AI applications in transportation [18][19]. - In the robotics field, Nvidia has invested in companies like Fi g u r e AI and Bri g ht Ma c h i n e s, indicating its focus on making AI chips the "brains" of robots [20][22]. Strategic Goals - Nvidia's overarching goal is to become the foundational infrastructure for the AI era, akin to how Microsoft and Google established their operating systems [24][25]. - The company aims to create a dependency on its GPUs across all AI companies, ensuring that its technology is essential for AI development and deployment [24]. - By investing in a wide array of AI applications, Nvidia seeks to secure its position as a dominant player in the evolving AI landscape, with a focus on long-term growth and influence [25].

Core Viewpoint - Nvidia has become the first publicly traded company in the world to surpass a market capitalization of $5 trillion, reflecting its dominant position in the AI sector and raising questions about whether this valuation represents a new benchmark or an "AI bubble" [1][6]. Group 1: Market Performance - Nvidia's stock price increased by 3% to $207.04, with a year-to-date gain exceeding 50% [1]. - The company's market capitalization reached $5.03 trillion, marking a significant milestone in its valuation history [1][6]. - Since the launch of ChatGPT in November 2022, Nvidia's stock has surged approximately 11 times, indicating strong investor confidence in its growth potential [6]. Group 2: Demand and Supply Dynamics - The surge in Nvidia's market value is attributed to the explosive global demand for AI, positioning Nvidia as a key supplier in the AI "arms race" [2]. - Nvidia's AI chips, particularly the H100, are in high demand, with companies needing to place orders months in advance due to supply constraints [2][3]. - Major cloud computing companies are projected to increase capital expenditures significantly, indicating sustained demand for Nvidia's products [3]. Group 3: Competitive Landscape - Nvidia's competitive advantage is bolstered by its CUDA software ecosystem, which creates a high barrier to entry for competitors like AMD and Intel [2]. - Despite losing its market share in China, Nvidia is still expected to find growth opportunities in other global markets [3][5]. - The company faces increasing competition from emerging players in the AI chip market, particularly from Chinese firms [7]. Group 4: Valuation Concerns - Analysts have mixed views on whether Nvidia's valuation is justified, with some suggesting that the AI bubble has not yet burst [6][7]. - Nvidia's current price-to-earnings ratio is approximately 33 times its expected earnings, compared to the S&P 500 average of about 24 times, raising concerns about its high valuation [7]. - The potential for challenges in AI commercialization and regulatory issues could impact Nvidia's future growth trajectory [7].

Core Insights - OpenAI and AMD announced a significant computing power infrastructure agreement, involving a total of 6 gigawatts of computing power, which could allow OpenAI to acquire approximately 10% of AMD's shares [1] - Analysts view this deal as a validation of AMD's AI roadmap, positioning the company favorably in the rapidly growing AI market [1] - Following the announcement, Jefferies raised AMD's target price from $170 to $300, while UBS increased it from $210 to $265, reflecting increased market confidence in AMD's future [1][2] Market Dynamics - The agreement between OpenAI and AMD is perceived as a threat to NVIDIA, with retail investor sentiment shifting from neutral to bearish regarding NVIDIA's stock [3] - Some investors predict NVIDIA's stock could drop to $150, a nearly 20% decline from its previous close, due to increased competition and challenges in the Asian market [3] - Despite concerns about NVIDIA's high valuation, analyst Ming-Chi Kuo suggests that as long as the overall AI computing power market grows, the impact of the OpenAI-AMD partnership on NVIDIA should be limited [3] Competitive Landscape - OpenAI's CEO, Sam Altman, reassured that the partnership with AMD is intended to complement existing collaborations with NVIDIA, which includes a separate 10-gigawatt computing power agreement [4] - The deal grants OpenAI warrants that could influence AMD's strategic decisions, contrasting with NVIDIA's relationship where OpenAI is merely a customer [4]

Core Insights - The recent surge in the stock price of Cambricon, known as the "first domestic AI chip stock," is linked to significant changes in the domestic computing power sector, with its market value exceeding 600 billion yuan, surpassing Kweichow Moutai as the new "king of stocks" [1] - The value of domestic computing power is being re-evaluated as a result of new policies and industry trends, particularly in the context of the "Artificial Intelligence +" initiative, which emphasizes the strategic importance of self-controlled domestic computing power [1][2] - The role of computing power is shifting from being merely a resource for the digital economy to becoming a core engine driving the intelligent transformation of society, especially with the rise of generative AI technologies [1][2] Industry Dynamics - The demand for computing power is experiencing exponential growth due to the training and inference requirements of large models, necessitating specialized AI chips like GPUs, as traditional CPUs are inadequate [2] - The transition from "using as much as needed" to "having as much as required" for computing power reflects its strategic scarcity, which is a fundamental reason for the market's re-evaluation of domestic computing power [2] - Domestic companies such as Cambricon, Huawei Ascend, and Haiguang Information are making significant progress in AI chips, servers, and data centers, gradually forming a local industry chain that meets application needs in various sectors [2][3] Challenges and Opportunities - Despite advancements, domestic high-end AI chips still lag behind international counterparts in absolute performance, and critical upstream components like EDA software and advanced manufacturing processes remain challenges [3] - The gap in the ecosystem, particularly in software development, poses a significant hurdle for domestic chip manufacturers, who are striving to build their own ecosystems to compete with established players like NVIDIA [3] - The current international environment highlights the importance of supply chain security and data sovereignty, making the use of self-controlled computing infrastructure a strategic necessity, thus providing a valuable market opportunity for domestic computing power [3] Strategic Considerations - The industry must focus on core technology research, software ecosystem development, and in-depth exploration of application scenarios to avoid resource wastage from low-level repetitive construction [4] - The capital market is encouraged to adopt a patient and visionary approach, identifying and supporting companies with genuine core competitiveness rather than chasing short-term gains [4][5] - The re-evaluation of domestic computing power is timely, but sustainable development in the industry requires a steady and long-term approach rather than a fleeting surge in stock prices [5]

Core Viewpoint - Nvidia has transformed from a leading chip manufacturer to a full-stack AI infrastructure leader, with a 50% stock price increase in three months, driven by strong product offerings and robust financial performance [1][2][9]. Financial Performance - Nvidia maintains a gross margin of over 75% and expects Q2 revenue to reach $45 billion, exceeding market expectations [1][9]. - The company has a free cash flow margin exceeding 60%, indicating strong operational efficiency [1][14]. Product Roadmap - The upcoming GB300 series (Blackwell Ultra) is expected to enhance inference throughput and memory utilization by 50% [4]. - By Q4 2025, the NVL72 will achieve scale in large data centers, becoming a cornerstone for Nvidia's high-margin data center inference workloads, which currently account for over 70% of its data center business [4][9]. - The Vera Rubin architecture, set to launch in H2 2026, will offer over three times the inference computing capability compared to GB300, while maintaining backward compatibility [4][5]. - The Rubin Ultra design, expected by 2027, aims to deliver up to 15 exaFLOPS of FP4 throughput, significantly enhancing Nvidia's position in AI inference cloud [5][9]. Market Position and Competitive Landscape - Nvidia's structural advantages, including dominant platform economics and a deep ecosystem, position it as a core holding in AI infrastructure [2][10]. - The long-term potential market for AI is projected to reach $1 trillion, with infrastructure needs estimated at $300 billion to $400 billion [10][12]. - Despite competitive pressures from AMD and other custom chip developers, Nvidia's established software stack (CUDA, NeMo) and supply chain integration provide a buffer against market share erosion [12][17]. Valuation Metrics - Nvidia's current P/E ratio stands at 54, with a forward P/E of 40, indicating a premium valuation compared to industry averages [12][14]. - The company's PEG ratio is 0.68 (GAAP) and 1.37 (non-GAAP), suggesting that its valuation is at least partially supported by growth [14]. - Nvidia's expected EV/Sales ratio is 21, and EV/EBIT ratio is 34, reflecting a significant premium over industry standards, which reinforces its growth assumptions [14]. Strategic Outlook - Nvidia's roadmap for the next three years includes the launch of Blackwell GB300 in 2025, Vera Rubin in 2026, and Rubin Ultra in 2027, ensuring continued product leadership and predictable profitability [9][17]. - The company plans to invest over $10 billion in next-generation AI research and development, indicating a commitment to maintaining its competitive edge [12][15].

Core Insights - Marvell Technology is advancing its semiconductor technology by transitioning to 2nm and below nodes, utilizing innovative techniques such as gate-all-around transistors and backside power delivery [2] - The company is leveraging modular redistribution layer (RDL) technology to enhance its 2.5D packaging solutions, which can integrate multiple chips and improve power efficiency while reducing costs [2] - Marvell's potential market for data center semiconductors is projected to reach $94 billion by 2028, with a compound annual growth rate (CAGR) of 53% for its custom computing products from 2023 to 2028 [3] Marvell Technology's Innovations - Marvell is utilizing advanced packaging solutions, including 2.5D designs, to develop multi-chip AI accelerator solutions that are 2.8 times larger than existing single-chip solutions [2] - The RDL technology allows for shorter interconnect distances, reducing latency and improving power efficiency, while also enabling seamless replacement of defective chips [2] Competitive Landscape - Broadcom's AI semiconductor revenue is expected to reach $5.1 billion in Q3 FY2025, driven by a 46% year-over-year increase in AI revenue, particularly in AI networking [4][6] - Broadcom's next-generation Tomahawk 6 Ethernet switch, designed for AI-scale architectures, features a transmission rate of up to 102.4 Tbps, addressing network bottlenecks in high-performance AI systems [5] - NVIDIA continues to dominate the AI semiconductor space with its unparalleled GPU performance and scalable AI deployment solutions [6] Industry Trends - The rapid growth in data center infrastructure investments is benefiting companies like Marvell Technology, which is positioned to capitalize on the increasing demand for advanced semiconductor solutions [3] - Intel is advancing its AI strategy with a roadmap aimed at achieving process leadership by 2025, focusing on efficient server chips for high-density AI tasks [6]