Core Viewpoint - Nvidia's CEO Jensen Huang projected that data center infrastructure capital expenditure (capex) will exceed $1 trillion by 2028, indicating significant growth potential for the company [1] Group 1: Data Center Infrastructure - The anticipated $1 trillion in data center infrastructure capex by 2028 represents a continued acceleration in spending, which is favorable for Nvidia [2] - Nvidia estimates that data center infrastructure spending will reach approximately $400 billion in 2024, capturing around 25% to 30% of this market [3] - If Nvidia maintains its current market share, it could generate between $250 billion to $300 billion in data center infrastructure revenue by 2028 [4] Group 2: Product Innovations - Nvidia introduced the Blackwell Ultra GPU, which is expected to be more powerful and beneficial for time-sensitive services, with revenue projections exceeding those from the previous Hopper architecture [4] - The company plans to launch the Vera Rubin chip, which combines a GPU with a custom-designed CPU, promising double the speed of previous models [5] - Nvidia also unveiled the open-source software system Nvidia Dynamo, aimed at enhancing inference throughput and reducing costs across thousands of GPUs [6] Group 3: Expansion into New Markets - Nvidia is entering the robotics and autonomous driving markets, introducing the Isaac GROOT N1, an open humanoid robot foundation model [7] - The company is collaborating with General Motors to develop autonomous driving systems, providing GPUs and custom AI systems for manufacturing [9] - This partnership follows a recent deal with Toyota to supply chips and software for advanced driver-assistance features [9] Group 4: Investment Potential - Nvidia continues to innovate and expand beyond data centers, positioning itself as a leader in AI infrastructure and inference [10] - The stock is currently attractively valued, trading at a forward P/E ratio of under 26 and a PEG below 0.5, indicating potential for long-term investment [11]

撰 文 / 钱亚光 设 计 / 赵昊然 此次GTC大会上，黄仁勋继续表达对算力需求增长前景的看好。虽然大型语言模型能提供基础知 识，但推理模型能给出更复杂、更具分析性的回答。黄仁勋表示，借助该公司新推出的开源软件 Nvidia Dynamo和Blackwell芯片，将使DeepSeek R1的运行速度提高30倍。 他在主题演讲中，强调了英伟达系统所支持的人工智能应用的广度。他详细阐述了英伟达在自动驾 驶汽车、更优无线网络和先进机器人技术开发方面的贡献，并公布了公司未来两年的产品路线图。 他说，来自四大云服务提供商对GPU的需求正在飙升，并补充说，他预计英伟达的数据中心基础设 施收入到2028年将达到1万亿美元。 3月19日晚间，身着标志性的黑色皮装的英伟达首席执行官黄仁勋（Jensen Huang）在英伟达GTC大 会上占据了中心位置。 此次活动吸引了超过25000人来到美国加州圣何塞SAP中心，黄仁勋在主题演讲开始时向观众抛出 印有"AI 超级碗大赛"字样的T恤，并宣布今年的GTC（全球人工智能大会）为"AI 超级碗"大赛。 "去年我们在这里办GTC，被描述为'AI的摇滚音乐节'（AI Woodstock ...

Core Insights - The rapid advancement of AI models is accelerating, with improvements in the last six months surpassing those of the previous six months, driven by three overlapping expansion laws: pre-training expansion, post-training expansion, and inference time expansion [1][3]. Group 1: AI Model Developments - Claude 3.7 showcases remarkable performance in software engineering, while Deepseek v3 indicates a significant reduction in costs associated with the previous generation of models, promoting further adoption [3]. - OpenAI's o1 and o3 models demonstrate that longer inference times and searches yield better answers, suggesting that adding more computation post-training is virtually limitless [3]. - Nvidia aims to increase inference efficiency by 35 times to facilitate model training and deployment, emphasizing a shift in strategy from "buy more, save more" to "save more, buy more" [3][4]. Group 2: Market Concerns and Demand - There are concerns in the market regarding the rising costs due to software optimizations and hardware improvements driven by Nvidia, potentially leading to a decrease in demand for AI hardware and a symbolic oversupply situation [4]. - As the cost of intelligence decreases, net consumption is expected to increase, similar to the impact of fiber optics on internet connection costs [4]. - Current AI capabilities are limited by cost, but as inference costs decline, demand for intelligence is anticipated to grow exponentially [4]. Group 3: Nvidia's Roadmap and Innovations - Nvidia's roadmap includes the introduction of Blackwell Ultra B300, which will not be sold as a motherboard but as a GPU with enhanced performance and memory capacity [11][12]. - The B300 NVL16 will replace the B200 HGX form factor, featuring 16 packages and improved communication capabilities [12]. - The introduction of CX-8 NIC will double network speed compared to the previous generation, enhancing overall system performance [13]. Group 4: Jensen's Mathematical Rules - Jensen's new mathematical rules complicate the understanding of Nvidia's performance metrics, including how GPU counts are calculated based on chip numbers rather than package counts [6][7]. - The first two rules involve representing Nvidia's overall FLOP performance and bandwidth in a more complex manner, impacting how specifications are interpreted [6]. Group 5: Future Architecture and Performance - The Rubin architecture is expected to deliver over 50 PFLOPs of dense FP4 computing power, significantly enhancing performance compared to previous generations [16]. - Nvidia's focus on larger tensor core arrays in each generation aims to improve data reuse and reduce control complexity, although programming challenges remain [18]. - The introduction of the Kyber rack architecture aims to increase density and scalability, allowing for a more efficient deployment of GPU resources [27][28]. Group 6: Inference Stack and Dynamo - Nvidia's new inference stack and Dynamo aim to enhance throughput and interactivity in AI applications, with features like intelligent routing and GPU scheduling to optimize resource utilization [39][40]. - The improvements in the NCCL collective inference library are expected to reduce latency and enhance overall throughput for smaller message sizes [44]. - The NVMe KV-Cache unload manager will improve efficiency in pre-filling operations by retaining previous conversation data, thus reducing the need for recalculation [48][49]. Group 7: Cost Reduction and Competitive Edge - Nvidia's advancements are projected to significantly lower the total cost of ownership for AI systems, with predictions of rental price declines for H100 chips starting in mid-2024 [55]. - The introduction of co-packaged optics (CPO) solutions is expected to reduce power consumption and enhance network efficiency, allowing for larger-scale deployments [57][58]. - Nvidia continues to lead the market with innovative technologies, maintaining a competitive edge over rivals by consistently advancing its architecture and algorithms [61].

Core Viewpoint - Nvidia is positioned as a leading computing infrastructure company, with significant developments highlighted during its annual GTC event, showcasing both bullish and bearish analyst perspectives on its stock performance and future potential [2][5][7]. Company Overview - Nvidia, founded in 1993 and headquartered in Santa Clara, California, specializes in graphics, compute, and networking solutions, operating globally with two primary segments: Compute & Networking and Graphics [3][4]. Analyst Perspectives - Bullish analysts view Nvidia's Q4 performance as a catalyst for positive investor sentiment, with expectations that the stock may approach or exceed its 52-week high of $150 [5]. - Bearish analysts express concerns over potential overvaluation and risks associated with increased competition and demand reductions, suggesting a "sell" rating at current levels [7]. GTC Event Highlights - CEO Jensen Huang emphasized AI's transformative impact on technology, stating that generative AI has fundamentally changed computing, necessitating more computing power [8][9]. - Huang announced that the top four cloud service providers have purchased 3.6 million Blackwell GPUs, a significant increase from the previous year's 1.3 million Hopper GPUs [11]. Strategic Partnerships - Nvidia has formed strategic partnerships with major tech companies, including Cisco and T-Mobile for building full-stack radio networks, and GM for developing self-driving autonomous fleets [10]. Hardware Developments - Nvidia introduced new GPU generations, including Blackwell Ultra and Rubin, aimed at enhancing memory capacity and computational power, with Blackwell Ultra expected in the second half of 2025 and Rubin in the second half of 2026 [12][14]. Future Outlook - Analysts suggest that visibility into large customer spending commitments could serve as a positive catalyst for Nvidia's stock, similar to Broadcom's recent performance [13].

Core Insights - Nvidia Corporation has announced a partnership with General Motors to enhance self-driving technology, indicating a significant step towards the adoption of autonomous vehicles [1] - GM's CEO emphasized the long-standing collaboration with Nvidia, highlighting the role of AI in optimizing manufacturing and vehicle innovation [2] - The partnership will expand to include plant design and operations, showcasing a deeper integration of AI in GM's manufacturing processes [3] Group 1: Partnership Details - The partnership will involve GM building next-generation vehicles on Nvidia's Drive AGX platform, utilizing the Nvidia Blackwell architecture [1] - Key areas of collaboration include factory planning, robotics, in-vehicle hardware for advanced driver-assistance systems, and in-cabin safety experiences [1] - GM has been investing in Nvidia GPU platforms for AI model training, which will now extend to plant design and operations [3] Group 2: Market and Technology Insights - Nvidia's CEO highlighted the growing demand for Blackwell GPUs driven by advancements in generative AI, agentic AI, and the emerging field of physical AI [6] - Huang expressed optimism about the future of AI across various sectors, indicating a shift in focus from generative AI to physical AI [5] - The introduction of Nvidia Dynamo, an open-source distributed inference-serving library, was also announced, aimed at enhancing AI capabilities for partners [6] Group 3: Stock Performance - Nvidia's stock was trading at $115.83, down 3.1% on the day, with a 52-week trading range of $75.61 to $153.13, and a year-to-date decline of 15% [7] - GM's stock experienced a brief recovery following the announcement, trading at $48.37, with a 52-week range of $38.96 to $59.39 [7]