在ASIC市场，近日听到的一些言论是相悖的。 "全球众多ASIC项目中，90%会失败"这句话来自英伟达CEO黄仁勋。 黄仁勋对于 ASIC 的评价并不算多。因此相关言论一出，市场上关于 ASIC 成长性的讨论瞬间多了起来，各种猜测层出不穷。 黄仁勋表示，相较于英伟达主打通用GPU架构，ASIC专为特定任务打造，虽在单一用途上效能与效率极高，但缺乏灵活性与扩展性。这种"单点优化"策 略，难以应对AI应用快速演进的现实。黄仁勋评价ASIC时并非否认其价值，而是强调"入场门槛与运维难度都很高"。他以谷歌TPU为例称，其团队是"全 球最强ASIC团队"，但即便如此，谷歌Gemini模型仍同时部署在英伟达GPU上。 但市场上另一种声音同样有依据——在 ASIC 的迅猛发展中，英伟达已嗅到危险的信号。 ASIC，赶超GPU？ 在算力芯片市场，ASIC的"簇拥者"可并不算少。在ASIC芯片大厂、云巨头等助推下，AI算力市场正在迎来新的临界点。 根据野村证券的最新报告，目前英伟达GPU占 AI 服务器市场 80% 以上，ASIC 仅占 8%-11%。 然而，从出货量的角度来看，情况正在发生变化。到2025 年，谷歌的 TPU ...

Group 1 - Nvidia currently holds over 80% of the market value share in the AI server sector, while ASIC AI servers account for approximately 8%-11% [1][3][4] - Major cloud service providers like Meta and Microsoft are planning to deploy their own AI ASIC solutions, with Meta starting in 2026 and Microsoft in 2027, indicating potential growth for cloud ASICs [1][4][10] - The total shipment of AI ASICs is expected to surpass Nvidia's AI GPUs by mid-2026, as more cloud service providers adopt these solutions [4][10] Group 2 - Meta's MTIA AI server project is anticipated to be a significant milestone in 2026, with plans for large-scale deployment [2][13] - Meta aims to produce 1.5 million units of MTIA V1 and V1.5 by the end of 2026, with a production ratio of 1:2 between the two versions [21][22] - The MTIA V1.5 ASIC is expected to have a larger package size and more advanced specifications compared to V1, which may pose challenges during mass production [23][19] Group 3 - Companies like Quanta, Unimicron, and Bizlink are identified as potential beneficiaries of Meta's MTIA project due to their roles in manufacturing and supplying critical components [24][25][26] - Quanta is responsible for the design and assembly of MTIA V1 and V1.5, while Unimicron is expected to supply key substrates for Meta and AWS ASICs [24][25] - Bizlink, as a leading active cable supplier, is poised to benefit from the scaling and upgrading connections in Meta's server designs [26]

Core Viewpoint - Broadcom's Q2 FY2025 performance met market expectations, with revenue of $15 billion, a 20% year-over-year increase, primarily driven by AI business growth and VMware integration [1][6]. Financial Performance - Total revenue for the quarter was $15 billion, aligning closely with market expectations of $14.95 billion [1][5]. - Gross profit reached $10.2 billion, with a gross margin of 68% [1][5]. - The semiconductor business generated $8.4 billion, with AI contributing $4.4 billion, reflecting a sequential increase of $300 million [2][5]. - Infrastructure software revenue was $6.6 billion, showing a slight decline of $100 million due to VMware integration and a shift to subscription models [3][5]. Segment Analysis - AI Business: Revenue of $4.4 billion, with a sequential growth slowdown attributed to Google's TPU product transition. Future growth is anticipated with the ramp-up of TPUv6 [2][10]. - Non-AI Business: Generated $4 billion, experiencing slight declines in wireless and industrial sectors despite growth in enterprise storage and broadband [2]. - VMware Integration: The integration phase is largely complete, with subscription conversion rates exceeding 60%. The software business has seen a decline, indicating the end of high-growth phases post-acquisition [8][10]. Operating Expenses - Core operating expenses totaled $3.77 billion, up $570 million sequentially, primarily due to increased stock-based compensation. Excluding this, core operating expenses were stable at around $2.2 billion [3][5]. Future Guidance - For Q3 FY2025, Broadcom expects revenue of approximately $15.8 billion, with AI revenue projected to grow to $5.1 billion [4][15].

Investment Rating - The report assigns a "Buy" rating for the company with a target price of $133.02 based on a 20X PE for 2026 [4]. Core Views - The company is a leading manufacturer of ASIC servers and Ethernet switches, benefiting from the growth in AI inference demand, particularly from major cloud service providers in North America [2][3]. - The company is expected to recover from a short-term decline in server revenue due to Google's TPU product transition, with anticipated growth resuming in the second half of 2025 [2]. - The company is actively expanding its customer base for ASIC servers, having become a supplier for Meta and secured a project with a leading commercial AI company [2][3]. Summary by Sections 1. Deep Layout in ASIC Servers and Ethernet Switches - The importance of inference computing power is increasing, and the ASIC industry chain is expected to benefit from this trend [14]. - The company is positioned to benefit from the volume growth of ASIC servers and the expansion of its customer base, particularly with Google and Meta [27][31]. - The Ethernet switch business is poised to grow due to the trend of AI Ethernet networking, with increased demand for high-speed switches [32]. 2. Transition from EMS to ODM - The company is shifting from an EMS model to an ODM model, which is expected to enhance customer binding and improve profitability [47]. - The revenue from the hardware platform solutions (ODM) is projected to grow significantly, contributing to overall revenue growth [50][52]. - The company's gross margin and operating profit margin have been steadily increasing due to the growth of its ODM business [52]. 3. ASIC Industry and Company Alpha - The company is well-positioned in the ASIC server and Ethernet ODM switch market, benefiting from industry trends and new customer acquisitions [3][4]. - The company’s net profit is forecasted to grow significantly over the next few years, with expected profits of $593 million, $765 million, and $871 million for 2025, 2026, and 2027 respectively [4][8]. - The company is expected to gain market share as it expands its customer base and increases the complexity of its products [31]. 4. Profit Forecast and Investment Recommendations - The company’s revenue is projected to grow from $7.96 billion in 2023 to $15.89 billion in 2027, with a compound annual growth rate (CAGR) of approximately 14.1% [8]. - The EBITDA is expected to increase from $467 million in 2023 to $1.296 billion in 2027, reflecting strong operational performance [8].