NVIDIA NVLink Fusion是英伟达推出的一项关键技术，旨在通过开放其NVLink生态系统，允许第三方 CPU、ASIC等芯片与英伟达的GPU或Grace CPU通过NVLink实现高速互联，形成异构计算架构。在此 次合作中，双方计划在2030年前，将各自的半导体连接在同一块基板上，实现GPU、CPU等多颗芯片 像一颗芯片一样的超高速互联。 节能成关键指标 针对此次合作，英伟达CEO黄仁勋表示，通过与富士通CPU的连接，"能够实现全新层级的节能与高 效"。 10月5日，近日，英伟达与日本电信及计算机制造商富士通达成协议，将共同构建集成AI智能体的全栈 AI基础设施。 资料显示，富士通是全球领先的信息通信技术（ICT）企业，成立于1935年6月，总部位于日本。该公 司在全球拥有约11.3万名员工，业务涵盖、半导体、云计算、人工智能等领域，曾被《财富》杂志评 为"全球最受尊敬的公司"之一。 根据富士通公告，此次合作将专注于开发面向医疗保健、制造业和机器人等领域的行业专用AI智能体 平台，并通过NVIDIA NVLink Fusion技术无缝集成FUJITSU-MONAKA CPU系列与NVIDIA G ...

Core Viewpoint - Broadcom has reported impressive earnings, driven by strong demand for AI-related products, particularly custom ASIC chips, positioning itself as a significant competitor to Nvidia in the AI chip market [3][4]. Financial Performance - For Q3 of FY2025, Broadcom achieved revenue of $15.952 billion, a 22% year-over-year increase, slightly above the previous guidance of $15.8 billion [3]. - Adjusted net income reached $10.702 billion, reflecting a 30.15% year-over-year growth [3]. - AI business revenue was $5.2 billion, marking a 63% increase year-over-year, surpassing the previous quarter's guidance of $5.1 billion [3][4]. AI Chip Market Dynamics - Broadcom's XPU business accounted for 65% of its overall AI revenue in Q3, with continued growth in demand for custom AI accelerators from major clients [5]. - The company anticipates AI semiconductor revenue growth to accelerate, projecting $6.2 billion in AI revenue for Q4, contributing to 11 consecutive quarters of growth [4][11]. - Broadcom has received a significant order exceeding $10 billion for XPU chips, potentially from OpenAI, which could enhance its AI performance expectations for FY2026 [6]. Competitive Landscape - The ASIC chip market is expected to grow as it caters specifically to AI inference needs, which are becoming increasingly important as AI applications expand [8]. - Nvidia and AMD are also exploring opportunities in the ASIC chip market, indicating a competitive environment where both GPU and ASIC chips will coexist and develop [4][10]. - Major cloud service providers are investing in both Nvidia's GPUs and developing their own AI chips, reflecting a dual strategy to meet diverse AI infrastructure needs [9][10]. Market Outlook - The ASIC chip market is projected to grow at a CAGR of 52% from 2023 to 2028, potentially surpassing GPU shipments by 2028 [11]. - Broadcom's strong Q3 performance and optimistic Q4 projections have led to increased market confidence, with analysts raising target prices for the stock [11].

Core Insights - Marvell Technology (MRVL) is enhancing its position in AI infrastructure by expanding its custom chip capabilities, integrating new components to improve performance and scalability across large-scale systems [1][6] Financial Performance - In Q1 FY26, Marvell reported record Data Center revenues of $1.44 billion, representing a 76% increase year over year, driven by the rapid scaling of custom AI silicon [2][10] - Marvell's forward price-to-sales ratio is 7.36X, which is lower than the industry average of 8.15X [13] Strategic Developments - Marvell partnered with NVIDIA in May 2025 to offer NVLink Fusion technology, enhancing the flexibility of its custom cloud platform silicon for next-generation AI infrastructure [3] - The introduction of a new multi-die packaging solution based on proprietary interposer technology aims to improve die-to-die interconnect efficiency, reduce power consumption, and lower product costs [4] - Marvell launched the Ultra Accelerator Link (UALink) scale-up solution, providing an open-standards-based interconnect platform that enhances compute utilization and reduces latency [5] Competitive Landscape - Advanced Micro Devices (AMD) is advancing its AI solutions through the acquisition of ZT Systems, which will reduce deployment time for hyperscalers [7] - Broadcom (AVGO) reported a 170% year-over-year increase in AI networking revenues, now comprising 40% of its total AI semiconductor revenues, and introduced the Tomahawk 6 switch to enhance AI cluster performance [8] Market Outlook - The Zacks Consensus Estimate for Marvell's fiscal 2026 and fiscal 2027 earnings indicates year-over-year growth of 77.71% and 27.73%, respectively, with recent upward revisions in earnings estimates [16]

Summary of Key Points from the Conference Call Industry and Company Overview - The conference focused on the AI server industry chain, highlighting key players such as NVIDIA and Qualcomm, along with the Taiwanese server supply chain's strengths and weaknesses [1][2]. Core Insights and Arguments - **NVIDIA's Focus on AI Servers**: NVIDIA's announcements at COMPUTEX emphasized the AI server industry, including AI factories, agentic AI, robotics, and enterprise AI transformation. The introduction of NVLink Fusion aims to strengthen interconnectivity advantages [1][2]. - **Qualcomm's Entry into Server CPU Market**: Qualcomm announced its entry into the server CPU market, aiming to disrupt the X86 ecosystem and establish a closed ARM ecosystem. The Snapdragon E-LITECOM product line was showcased for enterprise applications [1][5]. - **Taiwan's Server Supply Chain**: Taiwan's server supply chain is robust, featuring major companies like TSMC and Foxconn. However, innovation in AI applications is lagging compared to global standards [1][6][7]. - **AI Token as a Future Tool**: AI Tokens are viewed as crucial for the future, with data centers expected to produce these tokens, marking the realization of AI applications [3][14][15]. - **Market Trends**: The Middle East market is anticipated to see increased shipments, potentially replacing Singapore, which could impact NVIDIA and AMD [3][17]. Financial Performance and Forecasts - **NVIDIA's Q2 Financial Guidance**: NVIDIA's Q2 revenue guidance is projected between $4.5 billion and $4.7 billion, factoring in a $550 million HRC usage compensation. The first quarter is expected to perform well due to the flexibility of RTX replacing RWS and increased shipments to the Middle East [3][11][19]. - **Supply Chain Challenges**: NVIDIA's supply chain is facing challenges, particularly with the GB200 NFL72 product, which has a complex assembly process involving over 200 components and multiple suppliers [12][18]. Additional Important Insights - **Product Releases**: NVIDIA's key product releases include NVLink Fusion, which allows third-party integration, and RTX Pro Server, which can replace some H20 Server functionalities, especially in the Chinese inference service market [4][10]. - **Taiwan's Technological Position**: Taiwan is home to five globally significant tech companies, but it still needs to enhance its AI application innovation compared to competitors [6][7]. - **Future Product Launches**: The GBE300 MV72 is expected to be a significant product in the second half of the year, with major hardware showcases from ASUS, Acer, and MSI in the AI PC sector [9].