IT之家 12 月 15 日消息，ASML 首席执行官 Christophe Fouquet（克里斯托夫・富凯 / 傅恪礼）在公司总部接受彭博社专访时称，他预计 High NA EUV 光刻 机将于 2027~2028 年正式投入先进制程的大规模量产作业中。 Fouquet 表示，High NA EUV 光刻机目前正由英特尔等客户测试，结果显示新设备的成像和分辨率表现良好，该企业 2026 年的任务之一是与客户合作将设 备的停机时间最小化。 Fouquet 还提到，ASML 对未来 10~15 年的大致技术路线已有一定的概念。该企业已启动下一代 Hyper NA EUV 的研究，为本世纪 30 年代的投运打下基 础。 目前在导入新一代图案化技术上最积极的是英特尔代工，其支持 High NA EUV 的 Intel 14A 节点将在 2027 年正式推出。 ...

他进一步指出，从近期的交流情况来看，人工智能数据中心支出所受的"季节性影响十分有限，甚至可 以说几乎不存在"。 智通财经APP获悉，韦德布什证券指出，台积电(TSM.US)11月营收已助其小幅超越市场原先预期。该 行分析师马特·布赖森表示，美元升值带来额外利好：以新台币折算的销售收入及毛利率同步抬升，使 得公司在本地货币口径下的超预期幅度大于美元口径。布赖森维持台积电"跑赢大盘"评级，目标价设定 为1700新台币。 "因此，我们眼下对台积电第四季度的盈利模型可能过于保守了。随着2026年展开，公司还有两股额外 推力：产品均价继续上调，加上产品组合进一步向更先进的新制程(含2纳米晶圆)倾斜，这两点都会带 来比当前预测更高的上行空间。"布赖森补充道。 据了解，台积电于2025年12月10日公布了2025年11月的最新业绩报告。主要数据显示，虽然营收环比略 有下降，但同比依然实现强劲增长。财报显示，台积电11月营收同比增长24.5%，达约3,436.1亿新台 币，但较10月下降6.5%。 这家为苹果、英伟达和AMD等全球顶级科技公司生产芯片的企业，今年10月销售额曾达到约3,674.73亿 新台币。今年1月至11月 ...

摩根大通维持对科磊(KLAC.US)的"增持"评级，给出的目标价为1485美元。在与科磊的高管团队会面 后，小摩发现自公司发布财报以来，订单簿持续改善，预计2026年上半年营收将实现低至中单位数的增 长，此前预期是持平至小幅增长。这主要是由于公司近期订单加速增长，尤其是在DRAM/HBM领域， 且工具的交货期也在延长。 ...

Core Viewpoint - Morgan Stanley maintains an "Overweight" rating on KLA Corporation (KLAC.US) with a target price of $1485, citing improved order books and expected revenue growth in the first half of 2026 [1] Group 1: Orders and Revenue Growth - The order book for KLA has been continuously improving since the earnings report, with a notable acceleration in orders, particularly in the DRAM and HBM sectors [1] - Revenue is projected to achieve low to mid-single-digit growth in the first half of 2026, an upgrade from previous expectations of flat to slight growth [1] Group 2: Delivery Times - The delivery times for tools have been extended, indicating increased demand and potential supply chain challenges [1]

Investment Rating - The report initiates coverage with a "Buy" rating for the company [3] Core Insights - The company is expected to benefit significantly from the ongoing advancements in semiconductor manufacturing processes, particularly in etching and deposition technologies, which are crucial for the production of advanced logic and 3D NAND devices [5][57] - The projected net profits for the company from 2025 to 2027 are estimated to be 21.81 billion, 31.59 billion, and 42.79 billion yuan, reflecting year-on-year growth rates of 35.0%, 44.8%, and 35.5% respectively [3][4] Financial Data and Profit Forecast - Total revenue is projected to grow from 6,263.51 million yuan in 2023 to 20,631.26 million yuan in 2027, with a compound annual growth rate (CAGR) of 41.3% [4][22] - The company's gross margin is expected to stabilize as the thin film deposition segment matures, contributing to overall profitability [31][63] - The company’s return on equity (ROE) is forecasted to improve from 10.02% in 2023 to 14.57% in 2027 [4] Industry Overview - The global semiconductor equipment market is anticipated to grow from 125.5 billion USD in 2024 to 159.9 billion USD in 2026, with a CAGR of approximately 9.1% [40] - The Chinese semiconductor equipment market is projected to expand from 49.1 billion USD in 2024 to 66.2 billion USD in 2027, driven by domestic wafer fab expansions and local replacements [40] Company Positioning - The company is recognized as a leading supplier of front-end semiconductor equipment in China, with a strong focus on etching and deposition technologies [8][18] - The company has established a comprehensive product line that includes advanced etching systems capable of supporting production from 65nm to 5nm nodes and beyond [5][8] - The company is positioned to capture structural growth opportunities in the semiconductor equipment market due to its technological depth and successful customer validations [57]

国芯网[原:中国半导体论坛] 振兴国产半导体产业！ 不拘中国、 放眼世界 ！ 关注 世界半导体论坛 ↓ ↓ ↓ 12月8日消息，据韩媒报道，三星电子近日获得美国AI芯片初创公司Tsavorite的订单，将采用其4nm制程工艺为后者代工AI芯片，订单金额超过1亿美元。 Tsavorite的AI芯片被设计为"全功能处理单元"，旨在成为下一代AI芯片。其特点在于将CPU、GPU与DRAM集成于单一芯片，并能根据不同市场与应用需 求，在功耗、性能与扩展性上进行灵活配置。 今年11月，该公司已宣布获得来自全球财富500强企业、主权云服务商及系统集成商等客户超过1亿美元的预购订单。 报道指出，三星电子持续推进4nm工艺的良率提升，近期已提高至60%-70%水平，重新增强了市场竞争力。今年10月，三星还成功获得了特斯拉自动驾驶 芯片"AI5"的代工订单。 随着市场对3nm及2nm等先进制程的需求增长，三星可能为4nm工艺提供更具竞争力的定价策略，从而进一步巩固其市场地位。与此同时，三星正在加速 推进3nm与2nm工艺的良率改善。据最新消息，其首款采用2nm GAA技术的芯片组Exynos 2600良率已达50%，未来特斯拉 ...

在半导体行业向先进制程持续突破的进程中，电子特种气体作为芯片制造的 "关键材料基石"，其性能 与供应稳定性直接影响制程精度与产业安全。其中，八氟环戊烯（C5F8）凭借优异的刻蚀性能与环保 特性，成为 14 纳米及以下先进集成电路制造中的核心蚀刻气体。基于最新市场调研，我们从产品本 质、市场规模、竞争格局到发展趋势，全面解析这一细分领域的现状与潜力。 什么是半导体用八氟环戊烯（C5F8）？ 八氟环戊烯（化学式 C₅F₈，CAS 号 559-40-0）是一种不饱和全氟烯烃类有机氟化合物，常温常压下呈 无色气体状，具有不易燃、沸点低（约 27℃）、纯度高、高热稳定性等特性。作为含氟电子特种气体 的重要品类，它最核心的应用场景是半导体行业的等离子蚀刻工艺 —— 尤其在先进集成电路（IC）制 造中，针对高深宽比结构（如 FinFET、3D NAND 存储单元）的刻蚀需求，能实现高精度、高选择性的 加工效果。 与传统含氟蚀刻气体（如 CF₄、C₂F₆）相比，它的优势显著：不仅能将硅基底与光刻胶的选择比提升至 15:1 以上，实现 0.1 微米级线宽的精确控制，还能形成 88° 垂直侧壁角度的各向异性刻蚀，同时减少 67% ...

Core Viewpoint - Nvidia is facing significant challenges due to increased competition and potential threats to its business model, particularly from Google and Intel, which could undermine its high profit margins and market dominance [1][2]. Group 1: Nvidia's Current Challenges - Nvidia's crisis began with major investors like Masayoshi Son reducing their stakes, leading to a public dispute over its financial performance and valuation [1]. - The company's high profit margins are under threat from Google's aggressive push with its TPU technology, which is now being offered to clients for on-premises deployment, potentially eroding Nvidia's market share [1]. - The departure of TSMC's senior vice president, who joined Intel, raises concerns about the stability of Nvidia's supply chain and its reliance on TSMC for advanced manufacturing processes [2][5]. Group 2: Impact of Key Personnel Movements - The former TSMC executive, who has extensive experience in advanced process technology, is now tasked with enhancing Intel's manufacturing capabilities, which could shift the competitive landscape [3][4]. - Intel's response to TSMC's legal actions indicates a strategic move to position itself as a viable alternative in the semiconductor manufacturing space, potentially impacting Nvidia's pricing power [5][7]. Group 3: Long-term Implications for Nvidia - The ongoing legal battle and personnel shifts may accelerate the emergence of a second top-tier foundry, which could diminish Nvidia's exclusive advantages in advanced manufacturing [6][8]. - If Intel successfully narrows the technological gap, Nvidia may lose its ability to leverage its manufacturing dominance to justify its high valuation and profit margins [7][8]. - The market's perception of the long-term stability of the Nvidia-TSMC partnership is changing, suggesting that Nvidia's competitive edge may not be as secure as previously thought [8].

Core Insights - The global semiconductor industry is engaged in a strategic competition centered around the establishment of 2nm wafer fabs, seen as a critical threshold for AI computing sovereignty in the AI era [1][18][25] - TSMC plans to expand its 2nm production capacity in Taiwan from seven to ten fabs, with significant investments in overseas projects in the US, Japan, and Germany [2][3][19] - Intel is positioning itself with its 18A process technology, backed by national capital, to compete directly with TSMC's 2nm offerings [4][8][19] - Samsung is ramping up its 2nm production capabilities, achieving a yield of 55-60% and securing significant contracts, including a long-term deal with Tesla [9][10][19] - Japan's Rapidus is also entering the 2nm race, supported by government initiatives, aiming to establish local production capabilities [12][16][19] TSMC's Strategy - TSMC's plan includes ten 2nm fabs in Taiwan, with an estimated cost of approximately NT$300 billion (US$80-100 billion) per fab, totaling around NT$900 billion for the additional three fabs [2][19] - The company emphasizes that advanced processes will primarily serve top-tier clients, particularly in AI and high-end computing sectors [2][3] - TSMC's strategy is a response to both the surging demand for AI chips and competitive pressures from rivals like Intel and Samsung [3][19] Intel's Positioning - Intel's 18A process is designed to compete with TSMC's 2nm technology, with recent reports indicating improvements in yield and production stability [4][6][19] - The company aims to establish itself as a significant foundry player, focusing on collaboration with design partners to enhance its manufacturing capabilities [8][19] Samsung's Developments - Samsung's 2nm process yield has improved significantly, and the company plans to increase its production capacity by 163% by the end of next year [9][10] - The company has secured a major contract with Tesla for AI chip production, which is expected to enhance its standing in the US market [9][10][19] Rapidus and Japan's Strategy - Rapidus is working on establishing 2nm production capabilities with government support, aiming to start mass production by the second half of the 2027 fiscal year [12][16] - The company is also planning to develop 1.4nm products, indicating a long-term commitment to advanced semiconductor manufacturing in Japan [16][19] Market Dynamics - The competition for 2nm fabs is driven by the need for higher transistor density and lower power consumption to support AI applications [18][19] - The establishment of these fabs is not only a corporate decision but also a reflection of national industrial policies, with significant government backing for companies like Intel and Rapidus [19][25] - The high capital expenditure required for 2nm fabs necessitates partnerships with key clients and government subsidies to mitigate financial risks [18][19] Conclusion - The race to build 2nm fabs represents a pivotal moment in the semiconductor industry, with implications for technological leadership, market dynamics, and geopolitical considerations [25] - Companies that successfully navigate this high-capital, high-risk environment will be well-positioned to influence the future of semiconductor manufacturing and AI capabilities [25]

Core Insights - The global semiconductor industry is entering a critical phase with the countdown to the mass production of 2nm technology, intensifying competition among major players like Intel, TSMC, and Samsung [1][7] - Intel has unveiled its Intel 18A advanced process technology, which is positioned to compete with TSMC and Samsung's 2nm offerings, aiming to regain its leadership in process technology [1][4] Group 1: Intel 18A Technology - Intel 18A, equivalent to 1.8nm, will be utilized in the next-generation AI PC platform, Panther Lake, leveraging two key technologies: RibbonFET and PowerVia [1][2] - RibbonFET technology enhances control over current flow, significantly reducing leakage, while PowerVia addresses internal chip congestion by relocating power delivery to the back of the chip [2][3] - Intel 18A demonstrates impressive efficiency metrics, with over 15% performance improvement at the same power consumption, and over 25% power reduction for the same performance, alongside a 30% increase in transistor density [3] Group 2: Panther Lake and Future Developments - Panther Lake, based on Intel 18A, is set to enhance multi-core performance by 50% and graphics performance by over 40%, with an AI computing power of 180 TOPS [4][5] - The first Panther Lake processors are currently in production, with an official launch planned for January at CES [5] Group 3: Competitive Landscape - The competition in advanced semiconductor processes is primarily among TSMC, Samsung, and Intel, with TSMC currently leading in 2nm technology with a yield rate of 60%-65% [7][8] - Samsung faces challenges with its 3nm process yield, which is between 20%-40%, impacting its ability to secure major clients [7][8] - Intel's 18A process, while not yet matching TSMC's capacity, has a potential advantage in power efficiency due to its unique PowerVia technology, which could attract high-performance computing and mobile device markets [8] Group 4: AI Ecosystem Development - Intel aims to build a comprehensive AI ecosystem leveraging the Intel 18A process, transitioning PCs from AI-enhanced to AI-native platforms [9] - The Panther Lake platform integrates CPU, GPU, and NPU architectures, supporting advanced AI models and enhancing performance metrics significantly [9] - Intel emphasizes collaboration with local Chinese software ecosystems to optimize AI applications and improve performance across various use cases [10]