公众号记得加星标⭐️，第一时间看推送不会错过。 人们普遍认为，半导体产业主要由于劳动力和其他成本高昂而离开欧洲。然而，事实并非如此。真正 的原因是，半导体产业的发展需要巨额投资以及金融机构和大型企业的强力支持。以上世纪90年代的 德国为例，西门子投资兴建工厂，但规模太小。这最终导致其半导体业务分拆为英飞凌科技公司。 全球范围内情况也类似。如果我们回顾过去50年排名前三的DRAM制造商，会发现大约有30家公 司。然而，其中大多数要么已经退出内存市场，要么被收购，要么破产倒闭。仅有的例外是美光科 技、三星电子和SK海力士。不过，这三家公司中，有两家在过去都曾两次濒临破产。 并非这些公司大多因为缺乏必要技能而注定失败。从长远来看，它们面临的竞争环境与那些拥有政府 支持、深谙半导体重要性的公司截然不同。 在动荡的国际局势下，这是"最后的机会"。 半导体产业对全球经济的重要性如今已得到广泛认可。2022年爆发的供应链危机暴露了该行业对半导 体的严重依赖，当时仅仅少数半导体芯片的短缺就导致价值数百万美元的工业设备无法出货。为应对 危机，世界各国纷纷出台《芯片与科学法案》，鼓励在国内建设半导体工厂。 更重要的是，近年来地缘 ...

Core Viewpoint - The company reported a record sales revenue of $659.9 million for Q4 2025, reflecting a year-on-year growth of 22.4% and a quarter-on-quarter growth of 3.9% [2][3]. Financial Performance - Q4 2025 sales revenue reached $659.9 million, with a gross profit of $85.5 million and a gross margin of 13.0%, compared to 11.4% in Q4 2024 [3]. - The company achieved a total annual revenue of $2.402 billion for 2025, marking a 19.9% increase year-on-year [2]. - The shipping volume for 8-inch wafers was 1.448 million pieces, up 19.4% year-on-year [3]. Capacity Expansion - The company is making significant progress in capacity expansion, with the first phase of the second 12-inch production line in Wuxi (FAB9) exceeding expectations [2][9]. - The acquisition of the Shanghai 12-inch manufacturing base (FAB5) is progressing smoothly, laying a solid foundation for future capacity enhancement and process upgrades [2][9]. Revenue Breakdown by Region - Sales revenue from China was $539.3 million, accounting for 81.8% of total sales, with a year-on-year increase of 19.6% [4][5]. - North America contributed $72.8 million, showing a significant growth of 51.3% year-on-year [4][5]. - Revenue from Europe reached $19.3 million, up 35.6% year-on-year, driven by demand for MCU and IGBT products [4][5]. Revenue Breakdown by Technology Platform - Embedded non-volatile memory sales reached $180.2 million, a 31.3% increase year-on-year, primarily due to demand for MCUs and smart card chips [6][7]. - Standalone non-volatile memory sales were $56.6 million, up 22.9% year-on-year, driven by increased demand for flash products [6][7]. - Power devices sales were $168.9 million, reflecting a modest growth of 2.4% year-on-year [6][7]. Revenue Breakdown by Process Node - Sales from 65nm and below process nodes reached $186.7 million, a 44.7% increase year-on-year [8]. - Revenue from 90nm and 95nm nodes was $163.9 million, up 54.0% year-on-year [8]. - Sales from 0.25um nodes decreased by 32.0% year-on-year to $1.9 million, primarily due to a decline in demand for power devices [8]. Revenue Breakdown by End Market - Consumer electronics remained the largest end market, contributing $419.6 million, which is 63.7% of total sales, with a year-on-year growth of 21.8% [9]. - Industrial and automotive products generated $146.7 million, up 18.3% year-on-year [9]. - Communication products sales reached $83.5 million, reflecting a 29.1% increase year-on-year [9].

公众号记得加星标⭐️，第一时间看推送不会错过。 凭借 CAN XL，博世彰显了其在控制器局域网 (CAN) 开发和实施领域的核心领导地位。CAN 最初 由博世于 20 世纪 80 年代开发，现已成为车载通信的骨干，也是现代电子架构的关键要素。自那时 起，博世一直是 CAN 技术的主要推动者，并持续引领着 CAN 技术向更高性能和更高灵活性方向发 展。 CAN技术的演进——由博世提供技术支持 图 1：CAN 从 1990 年到 2025 年的发展历程 2023 年 ， CAN XL 的 问 世 标 志 着 CAN 技 术 发 展 的 一 个 新 里 程 碑 ， 其 开 发 旨 在 弥 合 CAN FD 和 100Base-T1 以 太 网 之 间 的 性 能 差 距 。 当 时 ， 汽 车 OEM 厂 商 主 要 依 赖 于 2 Mbit/s 的 CAN FD 和 100 Mbit以太网。CAN XL通过提供一种经济高效的≤20 Mbit/s解决方案来补充这些技术，该方案保留了 CAN的经典优势，例如仲裁、鲁棒性和长短截线长度，同时引入了增强的安全性和可靠性功能、更 高的服务质量以及对面向服务架构（SOA）的支 ...

上周五，阿里的大模型千问狂撒30亿，请全民喝奶茶，只需要一句话"千问，帮我点杯奶 茶"，就能联动千问、淘宝、支付宝、盒马等多个 App，把原本需要多步操作的流程压缩成 一次指令。多家奶茶店接单到手软，被迫临时"歇业"、外卖骑手爆单、千问登顶 App Store 榜一。可以说，让无数大众近距离地真实体验了AI时代的全新生活方式。 而在前不久，另一个现象在开发者社区发生：原本上市一年多、销量平稳的 Mac mini，在短短十 多天里出现显著增量，甚至出现断货与补货讨论。其背后的推手则是一个诞生不到两个月的开源 AI 助手——OpenClaw。它在社区引发了一波"用 Mac mini 部署 AI 助手"的晒图与教程潮，进而 把一台小主机推成了"AI Agent 试验田"。 两件事表面上风马牛不相及，本质却指向同一个趋势： AI Agent（智能体）负载正在崛起。 大模 型时代沉淀的能力，正在通过"智能体化"加速落地——这也意味着，芯片产业的评价体系可能要变 了。 "一只龙虾"，引爆Ma c mini 两个月前，一位已经退休的奥地利软件工程师开发者Peter Steinberger利用周末随手"攒"出了一个 项目。起 ...

Core Viewpoint - Arm, a company deeply embedded in the semiconductor industry, faces challenges despite its widespread design usage in smartphones and connected devices. The company is optimistic about future growth driven by artificial intelligence, even as its stock price has declined due to weak demand in consumer electronics [2][3]. Group 1: Company Overview - Arm does not sell chips but licenses its designs, allowing clients to modify and produce chips themselves, generating revenue through upfront licensing fees and royalties [2]. - The company has shipped over 300 billion chips, with more than 30 billion shipped in the last year alone [2]. Group 2: Market Dynamics - Despite the growth in the Philadelphia Semiconductor Index by 65% since early 2025, Arm's stock has decreased by 2% due to sluggish demand in smartphones and consumer electronics [2]. - The rise of artificial intelligence is expected to stimulate demand for Arm's chips, particularly in data centers, which are increasingly reliant on both GPUs and CPUs [4]. Group 3: Strategic Choices - Arm may need to evolve from merely selling design blueprints to developing its own chips to capture more value from the AI boom, which could alienate existing clients [3][6]. - The company is exploring new revenue streams through pre-assembled processor modules, which could triple income per chip compared to traditional designs [5]. Group 4: Financial Performance - Analysts project Arm's revenue for the current fiscal year to be around $5 billion, with half coming from patent royalties and the other half from licensing fees, marking a 20% growth from 2025 [4][5]. - The patent royalty per mobile chip was approximately $0.86, representing 2.5% to 5% of the chip's price [5]. Group 5: Competitive Landscape - Arm's revenue is significantly lower compared to major chip manufacturers like Nvidia and Intel, highlighting the need for strategic shifts to enhance profitability [5]. - The company faces competition from China's push for RISC-V architecture, which is seen as a domestic alternative to Arm and Intel designs [7]. Group 6: Future Outlook - Arm's CEO expresses concern over the company's ability to keep pace with the rapid evolution of AI technologies, as chip design and manufacturing can take years, while AI models can evolve in months [7]. - The company's ownership structure, with SoftBank holding over 85% of shares, may influence its strategic direction, particularly as SoftBank seeks to build a competitive chip portfolio [6].

三星电子周四表示，已开始出货其第六代高带宽存储器 HBM4，成为首家开始大规模生产这种对人工智能至关重要的 下一代存储芯片的芯片制造商。 公众号记得加星标⭐️，第一时间看推送不会错过。 在经历了一轮"混战"之后，HBM 4市场终于迎来了第一个赢家，HBM市场也迎来了新的不确定性。 此次发布对三星而言意义重大。在前一轮HBM周期中，三星一直落后于竞争对手，这引发了人们对其在人工智能驱 动型半导体领域竞争力的担忧。凭借HBM4，三星旨在重振旗鼓，并在下一波人工智能内存需求浪潮中抢占先机。 提前一周出货，三星野心勃勃 据韩媒业内人士透露，三星在与客户协商后，将出货计划提前了约一周。据悉，该芯片已提前通过英伟达的质量测 试，这反映出其卓越的性能，正如消息人士所说。 从一开始，三星就立志超越全球半导体标准组织——联合电子器件工程委员会（JEED）制定的基准。为了实现这一 目标，它将最新的1c DRAM与4纳米制程工艺相结合——业内人士称，这种方法此前从未有人尝试过。 三星表示，三星的HBM4显存可提供高达11.7 Gbps的稳定处理速度，比业界标准的8Gbps提升约46%，树立了HBM4 性能的新标杆。这比其前代产品H ...

Core Insights - Nvidia stands out as a leading player in the current AI wave, primarily due to its GPU architecture and CUDA ecosystem, which have positioned it at the center of the computing power competition [2] - The rise of Nvidia is not an isolated event; it is part of a broader evolution involving various components such as wafer manufacturing, advanced packaging, storage bandwidth, and network interconnects [2] - The AI infrastructure competition has shifted from a focus on computing power to addressing connectivity bottlenecks, highlighting the importance of interconnect technologies [10] Group 1: Nvidia and AI Infrastructure - Nvidia's data center revenue has reached new highs, with its latest GPU products being in high demand even before full release [2] - The AI computing power explosion is not solely dependent on GPUs but requires a complex ecosystem of interconnected technologies [2] - The emergence of companies like Credo, which specializes in active electrical cables, illustrates the hidden champions supporting Nvidia's growth [3][4] Group 2: Connectivity Technologies - Credo's active electrical cables have gained significant market share, with an estimated 88% of the active cable market, and are projected to double its revenue to nearly $1 billion by 2026 [6] - Corning's optical fibers are becoming the preferred choice for data center networks due to their superior speed and energy efficiency compared to copper cables [7][9] - The coexistence of copper and fiber technologies in AI data centers reflects a unique symbiotic relationship driven by the specific demands of AI workloads [9] Group 3: Optical Communication and Components - Lumentum is transitioning from a telecommunications provider to a core enabler for AI data centers, with significant revenue growth driven by data center demands [12][13] - Coherent is also innovating in optical transceivers, with a focus on high-speed communication driven by AI and machine learning [16] - The demand for optical components is expected to surge as AI workloads increase, but companies must be cautious of potential demand fluctuations once connectivity bottlenecks are resolved [19] Group 4: Precision Timing and Materials - SiTime specializes in MEMS-based timing devices, which are crucial for maintaining synchronization in AI data centers, showing significant revenue growth [20][23] - T-glass, a critical material for advanced chip packaging, is in high demand but faces supply constraints, highlighting the vulnerabilities in the supply chain [24][26] - The importance of precision timing and reliable materials underscores the foundational elements that support AI infrastructure, often overlooked in discussions about advanced technologies [27][28] Group 5: Supply Chain Dynamics - The supply chain supporting Nvidia's empire is characterized by both strength in technological barriers and vulnerability due to single-source dependencies [30] - Companies like Corning and SiTime have shown resilience and adaptability, indicating a long-term commitment to technological advancement despite market uncertainties [30][31] - The future of AI infrastructure will depend on the ability of these suppliers to meet the exponential growth in computing power demand while managing risks associated with supply chain dependencies [31]

Core Viewpoint - The article discusses a dispute between Taiwanese semiconductor equipment manufacturer FitTech and Sanan Optoelectronics, focusing on equipment quality issues that led to financial losses and operational disruptions for Sanan [1][3]. Group 1: Dispute Background - FitTech announced that the China International Economic and Trade Arbitration Commission ruled that Sanan's subsidiaries must pay approximately 327 million yuan to FitTech due to delays in equipment delivery [1]. - Sanan claims that the reason for withholding payment is due to serious quality issues with the equipment supplied by FitTech, which has caused significant losses for Sanan [3]. Group 2: Equipment Quality Issues - The core of the dispute revolves around the LED die sorter, a critical automation device in semiconductor and LED manufacturing, which is responsible for sorting chips based on performance parameters [3]. - Sanan reported that the equipment delivered by FitTech did not meet contractual technical specifications, citing hardware deficiencies and software issues that hindered production [4]. - The quality defects in the equipment have severely impacted Sanan's production capacity and ability to meet international customer orders, leading to significant order losses and damage to reputation [4]. Group 3: FitTech's Strategic Shift - FitTech has indicated a strategic shift towards focusing on optical communication-related equipment and services, moving away from the LED die sorter business [6][7]. - The company plans to launch silicon photonics-related equipment and expand its applications in laser cleaning equipment, indicating a departure from its previous product focus [7]. Group 4: Domestic Competitors Rising - The article highlights the rise of domestic semiconductor equipment suppliers, such as Silan Microelectronics, which have improved testing efficiency and production capacity, posing a competitive threat to FitTech [10][11]. - Silan's market share in the domestic probe station market has increased from 13% in 2019 to 23.3% in 2024, indicating a shift in customer preferences towards more reliable domestic suppliers [11].

韦斯特恩表示，太空中极高的真空度可以去除这些杂质，从而提高晶体的质量。"例如，如果您担心 氮会干扰晶体生长过程，那么在地球上（真空室中），氮的浓度可能约为10的-11次方（10的-11次 公众号记得加星标⭐️，第一时间看推送不会错过。 去年12月，英国初创公司Space Forge在其ForgeStar-1卫星上启动了轨道熔炉，产生了一股超高温等 离子体流，这有望在未来实现近乎理想的在轨半导体晶体制造。这项里程碑式的成就被誉为轨道制造 领域的突破，也是首次在自由飞行的商业卫星和非载人有效载荷上实现这一目标。 Space Forge成立于2018年，是众多秉持"太空制造材料能够助力打造超高效下一代电子产品、超高速 光网络以及推动药物研发突破"理念的公司之一。Space Forge的飞行熔炉专门用于制造晶种，这些晶 种随后将在地球上用于生产镓、氮化铝或碳化硅衬底，以制造高性能功率器件。 此前，人们曾在太空制造过半导体。上世纪70年代，宇航员在天空实验室空间站上生长了锑化铟和锗 晶体，类似的实验也在国际空间站上进行。根据2024年发表在《自然》杂志上的一项荟萃分析研究， 1973年至2016年间，约有160个半导体 ...

Core Viewpoint - Tower Semiconductor reported strong financial performance for Q4 2025, with all key financial metrics showing comprehensive growth, indicating a continuation of the upward trend seen since the beginning of the year [2][3]. Financial Performance - Q4 revenue reached $440 million, a 14% increase compared to the same period in 2024 [2]. - Gross profit grew by 27% to $118 million, while operating profit increased by 39% to $71 million [2]. - Net profit totaled $80 million, marking a 48% year-over-year increase, with earnings per share at $0.71 [2]. - For the full year, revenue rose by 9% to $1.6 billion, and net profit increased by 6% to $220 million [2]. Future Guidance - For Q1 2026, Tower expects revenue of $412 million, projecting a 15% year-over-year growth with a fluctuation range of 5% [2]. Capital Expenditure and Investments - Tower announced a $650 million capital expenditure plan, with an additional $270 million investment aimed at expanding infrastructure for silicon photonics component production [2][3]. - The total investment of $920 million aims to significantly increase the production capacity of silicon photonics wafers by over five times the actual monthly shipment volume in Q4 2025 by December 2026 [3]. Strategic Partnerships - Tower has entered into a collaboration with NVIDIA to supply silicon photonics components for their products [2]. - However, the partnership with Intel is set to end, as Intel intends to withdraw from a 2023 agreement to produce 300mm wafers for Tower's customers [4][5]. Context of Intel Partnership - The agreement was established shortly after Intel's failed $5.4 billion acquisition of Tower, which was intended to solidify their business relationship [5][6]. - The contract envisioned a $300 million investment from Tower to support the production of next-generation 300mm chips at Intel's facility [5].