公众号记得加星标⭐️，第一时间看推送不会错过。 在现今ASIC与二线AI运算芯片愈来愈强调性价比的大环境下，EMIB是有一些优势存在。 IC设计相 关人士表示，和英特尔针对EMIB进行合作测试的业者有在增加的趋势，当然短时间或许还不会有什 么很大的出货量，但放眼长期，如果合作的成效不错，2.5D的EMIB，甚或是3D封装的Foveros制 程，都有机会稳定接到一些订单。 近期因为台积电的CoWoS先进封装产能高度吃紧，除了几个AI芯片龙头有能力大量"包产能"外，其 他特用芯片（ASIC）和二线的AI芯片业者，难以争取到足够CoWoS产能支援。值得注意的是，市场 陆续传出，英特尔（Intel）的EMIB先进封装制程，成为芯片业者的考量之一，半导体业界更盛传， 网通芯片大厂Marvell甚至联发科，都积极想要尝试，甚至可能有"前段投片台积电、后段找上英特 尔"的新生意模式。 英特尔先进封装实力不俗EMIB有吸引力 据了解，由于英特尔EMIB技术本身价格较为实惠，散热表现也不错，面对一些技术规格需求相对没 那么高的产品，确实是能够支援，部分ASIC厂商如Marvell、联发科传出已经试着透过采用EMIB制 程，来提 ...

同时，EMIB在封装尺寸上也较具优势，相较于CoWoS-S仅能达到3.3倍光罩尺寸、CoWoS-L目前发展至 3.5倍，预计在2027年达9倍；EMIB-M已可提供6倍光罩尺寸，并预计2026年到2027年可支援到8倍至12 倍。价格部分，因EMIB舍弃价格高昂的中介层，能为AI客户提供更具成本优势的解决方案。 11月25日，TrendForce集邦咨询发布的最新研究显示，AI HPC（高效能运算）对异质整合的需求依赖先 进封装达成，其中的关键技术即是TSMC（台积电）的CoWoS解决方案。然而，随着云端服务业者 （CSP）加速自研ASIC（专用集成电路），为整合更多复杂功能的芯片，对封装面积的需求不断扩 大，已有CSP开始考量从TSMC的CoWoS方案，转向Intel（英特尔）的EMIB技术。 TrendForce集邦咨询表示，CoWoS方案将主运算逻辑芯片、存储器、I/O等不同功能的芯片，以中介层 （Interposer）方式连结，并固定在基板上，目前已发展出CoWoS-S、CoWoS-R与CoWoS-L等技术。随 着NVIDIA（英伟达） Blackwell平台2025年进入规模量产，目前市场需求已高度 ...

智通财经APP获悉，根据TrendForce集邦咨询最新研究，AI HPC（高效能运算）对异质整合的需求依赖 先进封装达成，其中的关键技术即是TSMC（台积电）的CoWoS解决方案。然而，随着云端服务业者 （CSP）加速自研ASIC，为整合更多复杂功能的芯片，对封装面积的需求不断扩大，已有CSP开始考量 从TSMC的CoWoS方案，转向Intel（英特尔）的EMIB技术。 TrendForce集邦咨询表示，CoWoS方案将主运算逻辑芯片、存储器、I/O等不同功能的芯片，以中介层 （Interposer）方式连结，并固定在基板上，目前已发展出CoWoS-S、CoWoS-R与CoWoS-L等技术。随 着NVIDIA（英伟达） Blackwell平台2025年进入规模量产，目前市场需求已高度倾向内嵌硅中介层的 CoWoS-L，NVIDIA下世代的Rubin亦将采用，并进一步推升光罩尺寸。 AI HPC需求旺盛导致CoWoS面临产能短缺、光罩尺寸限制，以及价格高昂等问题。TrendForce集邦咨询 观察，除了CoWoS多数产能长期由NVIDIA GPU占据、其他客户遭排挤，封装尺寸以及美国制造需求， 也促使Goog ...

Nov. 25, 2025 根据TrendForce集邦咨询最新研究，AI HPC（高效能运算）对异质整合的需求依赖先进封装达成， 其中的关键技术即是TSMC（台积电）的CoWoS解决方案。然而，随着云端服务业者（CSP）加速自 研ASIC，为整合更多复杂功能的芯片，对封装面积的需求不断扩大，已有CSP开始考量从TSMC的 CoWoS方案，转向Intel（英特尔）的EMIB技术。 TrendForce集邦咨询表示，CoWoS方案将主运算逻辑芯片、存储器、I/O等不同功能的芯片，以中介 层（Interposer）方式连结，并固定在基板上，目前已发展出CoWoS-S、CoWoS-R与CoWoS-L等技 术。随着NVIDIA（英伟达） Blackwell平台2025年进入规模量产，目前市场需求已高度倾向内嵌硅 中介层的CoWoS-L，NVIDIA下世代的Rubin亦将采用，并进一步推升光罩尺寸。 AI HPC需求旺盛导致CoWoS面临产能短缺、光罩尺寸限制，以及价格高昂等问题。TrendForce集邦 咨询观察，除了CoWoS多数产能长期由NVIDIA GPU占据、其他客户遭排挤，封装尺寸以及美国制造 需求，也促使 ...

本文档仅供上海信鱼私募基金管理有限公司18860455898研究使用，请勿外传 ab 20 November 2025 Global Research UBS Global I/O Semiconductors Cloud AI: how tight could N3 foundry and CoWoS be in 2026? [ERRATUM] New analysis of N3 foundry supply-demand In this note, we set out our proprietary analysis of N3 foundry supply. Cloud AI accelerators including Nvidia's Rubin and ASICs like Google's TPUs and Amazon's Trainium are due to migrate from N4 to the N3 process in 2026, all to be carried out at TSMC. Our research in October highlighted TSMC' ...

Core Insights - The rapid development of AI chips is driving the demand for high-performance computing (HPC) components such as GPUs, AI ASICs, and HBM, with advanced packaging technologies like 2.5D/3D becoming crucial in the semiconductor industry [1][3] - Intel's advanced packaging technology EMIB is being evaluated by major tech companies like Apple and Qualcomm, indicating a potential shift in the competitive landscape as these companies consider alternatives to TSMC [1][11] - The advanced packaging market is projected to grow significantly, with revenues expected to exceed $12 billion by Q2 2025 and reach approximately $45 billion by 2024, growing at a compound annual growth rate (CAGR) of 9.4% to around $80 billion by 2030 [3] Advanced Packaging Landscape - TSMC, Intel, and Samsung have established a competitive "three-horse race" in the advanced packaging sector, each playing different roles in the supply chain [3][20] - TSMC's CoWoS technology has become the de facto standard for high-bandwidth packaging, but it faces capacity constraints primarily due to Nvidia's significant demand [7][8] - Intel's EMIB technology offers flexibility and cost advantages, making it suitable for custom ASICs and AI inference chips, positioning it as a viable alternative to TSMC's offerings [11][15] Market Dynamics - The demand for advanced packaging is driven by the strong growth in AI and HPC sectors, with TSMC planning to expand its CoWoS capacity by over 20% by the end of 2026 [8][24] - Samsung's advanced packaging technologies, I-Cube and X-Cube, are designed to leverage its HBM supply chain, aiming to enhance its influence in the AI chip market [20][23] - The competition in AI chip manufacturing is no longer limited to packaging technologies but encompasses computational architecture, supply chain security, capital expenditure, and ecosystem integration [24]

公众号记得加星标⭐️，第一时间看推送不会错过。 在AI芯片快速发展的浪潮中，GPU、AI ASIC等高性能计算（HPC）核心，以及HBM（高带宽内 存），正成为采用 2.5D/3D 封装技术的高端产品的主力军。先进封装平台对于提升器件的性能和带 宽至关重要，其重要性已使其成为半导体领域最热门的话题，热度甚至超越了以往的尖端工艺节点。 近期，有关英特尔的先进封装技术 EMIB 正被科技巨头苹果和高通评估的消息引发了广泛关注：苹 果在相关招聘信息中，寻求熟悉 CoWoS、EMIB、SoIC、PoP 等技术的 DRAM 封装工程师；高通 也在招募资料中心产品管理总监，要求熟悉英特尔EMIB技术。虽然这些动作尚不意味着两大芯片设 计巨头已正式转向，但它们明确透露出全球顶级自研芯片企业正在积极评估英特尔作为台积电之外的 潜在替代方案。 图片来源：高通公司 而在AI芯片的先进封装领域，台积电、英特尔和三星已经形成了"三强鼎立"的格局。由于自身定位不 同，这三家公司在产业链中也承担着不同的封装角色。据Yole Group的分析，短期来看，2025年第 二季度先进封装收入将超过120亿美元。在人工智能和高性能计算强劲需求的推 ...

Core Insights - The market has high expectations for NVIDIA's upcoming earnings, with Q3 revenue projected at approximately $57 billion, a 62% year-over-year increase, surpassing the company's guidance of $54 billion [1][2] - For Q4, market expectations range from $64 billion to $65 billion, reflecting a year-over-year growth of 63% to 65%, also above the company's guidance of $62 billion to $63 billion [2] - The core growth drivers for NVIDIA are the data center segment, particularly the ramp-up of the Blackwell and Rubin chips, with significant demand from major tech companies [3][4] Market Dynamics - Traditional tech giants like Google, Microsoft, Meta, and Amazon have shown a 73% year-over-year increase in capital expenditures in Q3, indicating strong demand for AI infrastructure [3] - Emerging cloud companies such as CoreWeave and Nibiru are experiencing rapid growth in capital expenditures, with CoreWeave's expected capital spending for 2026 projected to be more than double that of 2023 [4][5] - The backlog of orders for these emerging cloud providers is substantial, with CoreWeave reporting over $55 billion in backlog, indicating strong demand for their services [5][10] Supply Chain and Production - Supply chain improvements are evident, with expectations for increased production capacity, particularly in CoWoS technology, which is crucial for NVIDIA's chip production [5][6] - NVIDIA's GTC conference indicated a projected shipment of 20 million chips, valued at approximately $500 billion, highlighting the company's robust supply chain capabilities [6][7] - The overall market is expected to see a significant increase in chip shipments, with major players like TSMC and AMD also raising their production forecasts [10][11] Long-term Outlook - By 2030, the optimistic EPS forecast for NVIDIA could reach $50, reflecting a compound annual growth rate of nearly 60% [2] - The AI infrastructure investment landscape is expected to remain strong, with both traditional and emerging cloud providers experiencing significant growth in demand [5][8] - Concerns regarding the AI bubble and profitability are present, but the long-term outlook remains positive, with a focus on the evolving demand for AI capabilities [8][9]

Summary of Key Points from the Conference Call Industry Overview - **Dominant Theme**: AI remains the dominant theme in the technology sector heading into 2026, with a notable lack of interest in non-AI segments [1][2] - **Investor Sentiment**: Overall sentiment is heavily skewed towards AI-related companies, with discussions focusing on the durability of AI demand and the implications for various companies in the supply chain [1][2] Company-Specific Insights TSMC (Taiwan Semiconductor Manufacturing Company) - **Revenue Growth**: TSMC is expected to see a revenue growth of 21.7% YoY in USD terms for 2026, following a strong 35% growth in 2025 [3][6] - **Utilization Rates**: The utilization rates for TSMC's 5nm and 3nm nodes are running at full capacity, indicating strong underlying demand primarily driven by AI [3] - **Capex Expectations**: There is a debate regarding TSMC's capital expenditure (capex) for 2026, with consensus expectations around US$50 billion, while the company models a more conservative US$44 billion due to cleanroom constraints [6][4] WinWay and MPI - **Market Position**: Both WinWay and MPI are highlighted as top stock picks due to their strong leverage to the Google TPU supply chain and expected revenue growth of 42% and 46%, respectively, into 2026 [11][12] - **Testing Demand**: WinWay is expected to benefit from the adoption of system-level testing (SLT), while MPI is set to ship its vertical probe card (VPC) to Google TPU in 2026 [2][11] Aspeed - **Server Demand**: Aspeed is anticipated to benefit from general server strength, with expectations of above-seasonal guidance for Q1 2026, driven by under-investment in server infrastructure by US cloud service providers [13][27] CoWoS Equipment - **Investor Sentiment**: There is a cautious sentiment around CoWoS equipment names, with limited earnings upside potential into 2026 and a focus on sectors with stronger earnings momentum [7][8] Additional Insights - **MediaTek**: Investor interest in MediaTek is low due to limited bottom-line growth and margin pressures, although potential catalysts exist for a revamp in 2027 [9] - **KYEC**: Elevated interest in KYEC is driven by Nvidia order visibility, although valuation concerns persist [10] - **ASE**: ASE is viewed positively due to continuous margin expansion from TSMC overflow business and high-margin opportunities in Fan-Out Chip-on-Substrate (FOCoS) [10] Risks and Considerations - **Market Risks**: Key risks include potential deterioration in end-demand recovery, slower customer node migrations, and intensifying competition in the semiconductor space [17][21][25][30] Conclusion - The technology sector, particularly in AI and semiconductor manufacturing, is poised for significant growth, with TSMC, WinWay, MPI, and Aspeed identified as key players. However, investors should remain cautious of market dynamics and potential risks that could impact growth trajectories.

| M | | | | --- | --- | --- | | | | Idea | | November 17, 2025 02:13 PM GMT | | | | Greater China Semiconductors Asia Pacific | Morgan Stanley Taiwan Limited+ | | | | Tiffany Yeh | | | | Equity Analyst | | | 2026 CoWoS Expansion to | Tiffany.Yeh@morganstanley.com | +886 2 7712-3032 | | | Charlie Chan Equity Analyst | | | Catch Up After TSMC's 3nm | Charlie.Chan@morganstanley.com | +886 2 2730-1725 | | | Daniel Yen, CFA | | | | Equity Analyst | | | Capacity Increase | Daniel.Yen@morganstanley.com | +886 2 27 ...