服务器CPU制造商面临的一大机遇在于，超大规模数据中心已进入"升级周期"。这意味着他们正在 寻求替换老旧的CPU架构，并转向更新的产品，例如AMD的第五代EPYC"Turin"处理器和英特尔 的Xeon Granite Rapids处理器。更具体地说，预计仅今年一年，服务器CPU的出货量就将增长高 达25%，这表明AMD/英特尔等公司通过服务器CPU产品拓展数据中心业务具有巨大的潜力。 我们知道，仅今年一年，AMD 就凭借 EPYC 系列处理器的强劲表现，成功占据了服务器市场的 主导地位。然而，根据 KeyBanc 的预测，服务器 CPU 的收入前景对所有市场参与者来说都非常 乐观。更重要的是，AMD 和 Intel 都计划在下一代服务器 CPU 产品线上采取"积极进取"的策 略，尤其是 AMD 计划在 2026 年下半年推出其 EPYC Venice 2nm 处理器，这款处理器被认为具 有颠覆性。 据报道，AMD 和英特尔正在考虑将服务器 CPU 的价格提高至多 15%，因为两家公司今年剩余时 间的现有库存均已售罄。 人工智能热潮如今已席卷服务器CPU市场。据KeyBanc（通过Jukan）估计，AMD和英 ...

如果您希望可以时常见面，欢迎标星收藏哦~ 外交部：对于美国输华芯片问题，中方已多次表明立场 针对美国政府称将会允许向中国出口英伟达H200芯片，但将会对此出售进行25%的关税征收，外交 部发言人毛宁1月15日在例行记者会上说，对于美国输华芯片问题以及关税问题，中方都已经多次表 明了立场。 美国或将很快提高芯片关税 白宫于 1 月 14 日表示，唐纳德·特朗普总统可能很快会扩大对半导体及其衍生产品的进口关税， 此举因其可能对韩国科技巨头三星电子和 SK 海力士以及该国消费电子产品制造商产生的影响而受 到密切关注。 10万亿，投向半导体 芯片巨头，市值大跌 黄仁勋：HBM是个技术奇迹 Jim Keller：RISC-V一定会胜出 全球市值最高的10家芯片公司 据白宫消息，特朗普总统签署了一项公告，对进口到美国并随后转口到第三国的半导体产品（例如 英伟达的H200芯片）征收25%的关税。去年12月，特朗普曾表示将允许H200芯片出口到中国，但 要求中国将25%的销售收入上缴美国。这项关税旨在通过对经美国从台湾运往中国的H200芯片征 收关税，将英伟达部分与中国相关的收入转化为美国政府收入。 此举源于美国商务部去年1 ...

Group 1 - TSMC has informed Nvidia and Broadcom that it cannot meet their large capacity demands due to overwhelming demand from the AI wave [1] - Broadcom has been requesting additional capacity from TSMC, primarily for the production of Google's Tensor Processing Units (TPUs) [1] - TSMC's recent expansion in Arizona will not provide immediate relief as the new facilities will take years to become fully operational [1] Group 2 - Intel's foundry business is becoming increasingly important as TSMC struggles to meet demand [1] - KeyBanc analyst John Vinh noted that while Intel may not challenge TSMC's leadership with its 18A process, it has the opportunity to surpass Samsung Electronics to become the second-largest foundry [1] - Intel's 18A process yield has improved to over 60%, indicating significant progress in its foundry operations [2] Group 3 - Apple is preparing to outsource the manufacturing of low-end processors for MacBook and iPad to Intel starting in 2027 [2] - Discussions are ongoing between Intel and Apple regarding the potential for Intel to manufacture low-end A-series mobile processors for iPhones using its next-generation 14A process by 2029 [2]

Core Viewpoint - The article discusses President Trump's announcement of a 25% tariff on certain advanced computing chips, including NVIDIA's H200 and AMD's MI325X, while exempting chips that support the U.S. technology supply chain [1][8]. Group 1: Tariff Announcement - President Trump announced a 25% tariff on specific advanced computing chips, which includes NVIDIA's H200 and AMD's MI325X [1][8]. - Chips imported to support the U.S. technology supply chain will be exempt from these tariffs [1][8]. - The White House indicated that broader tariffs on semiconductors and their derivative products may be imposed in the future [2][8]. Group 2: National Security and Economic Implications - The tariffs are justified under national security concerns, invoking Section 232 of the Trade Expansion Act of 1962 [2][9]. - The U.S. aims to restore domestic semiconductor production capacity to reduce reliance on foreign sources, which is critical for economic and national security [9]. - The current import levels of semiconductors are deemed insufficient to meet domestic demand, posing a threat to the U.S. industrial and military capabilities [9]. Group 3: Industry Reactions - NVIDIA expressed appreciation for the decision, highlighting the balance it strikes for U.S. chip industry competitiveness and high-paying jobs [2]. - AMD stated its compliance with all U.S. export control laws and policies [2]. - Both companies previously agreed to pay 15% of their sales from chips sold in China to the U.S. government, indicating their involvement in the ongoing trade dynamics [3].

Core Viewpoint - The "Cooperative Innovation Forum from Devices to Networks" aims to facilitate cross-field collaboration in the semiconductor industry, focusing on the entire semiconductor supply chain and promoting technological breakthroughs and ecosystem development in the context of domestic substitution [2][3]. Forum Information - The forum will take place on March 18, 2026, at the Shanghai New International Expo Center [6]. - It is co-hosted by Semiconductor Industry Observation and Munich Shanghai Optical Expo [4]. Agenda Overview - The event will feature a series of presentations and discussions, including expert speeches and company showcases, starting from 9:00 AM to 2:50 PM [9][8]. Industry Context - The Munich Shanghai Optical Expo serves as a premier platform for the global optoelectronics, laser, semiconductor, and communication technology sectors, emphasizing "technology leadership and ecosystem co-construction" [12]. - The forum is designed to address the collaborative challenges within the semiconductor and communication industries, creating a platform for targeted engagement and deep exchanges among key industry players [12]. Sponsorship Opportunities - The event offers various sponsorship packages, including conference packages, individual presentation opportunities, and tea break sponsorships, aimed at enhancing brand visibility and showcasing company capabilities [13].

在 2025 年末的 A 股市场，一场关于"硬核科技"的盛宴正进入高潮。 如果说国产GPU独角兽的接连上市是令人振奋的"前菜"，那么随着2025年12月30日长鑫科技正式 递交招股书，这场 AI 算力时代的"顶级主菜"终于正式登场。 作为今年科创板最受瞩目的 IPO，长鑫科技代表着中国 AI 算力最深层的"护城河"。 在 AI 大模型狂飙的今天，英伟达的GPU无疑是舞台的中心。但业界共识是：算力的上限不仅取决 于 GPU 计 算 有 多 快 ， 更 取 决 于 DRAM （ 动 态 随 机 存 取 存 储 器 ） ， GPU 的 每 一 次 计 算 都 必 须 由 DRAM内存芯片提供实时数据传输。 国产GPU大多属于Fabless（无晶圆厂）模式，即只负责设计，生产制造则外包给代工厂。这一模 式的优势在于轻资产，投入成本较低，但劣势在于生产环节不受控，在产能紧缺、地缘政治等情况 下，无法保障产品生产。 相比之下，长鑫采用的是 IDM（设计、制造一体化）模式。这意味着巨大的供应链安全性，在全 球地缘政治不确定的环境下，拥有自己的工厂意味着生产不再依赖外部。同时也具有设计-制造深 度耦合的优势，让长鑫能够更深层 ...

Core Viewpoint - The article discusses the critical shortage of high-end glass fiber, particularly from Nittobo, which is impacting the production of printed circuit boards (PCBs) essential for electronic devices, including iPhones and AI chips. This shortage is expected to create significant bottlenecks in the electronics manufacturing and AI industries by 2026 [2][4][6]. Group 1: Supply Chain Challenges - Apple is competing with major companies like Nvidia, Google, and Amazon for the limited supply of high-end glass fiber, which is crucial for chip substrates and PCBs [2][4]. - The demand for high-performance PCBs has surged due to the growth of artificial intelligence, leading to supply shortages that affect not only Apple but also Qualcomm [2][4]. - Apple has taken proactive measures, including sending employees to Japan to secure more materials and even seeking assistance from the Japanese government to increase supply from Nittobo [4][6]. Group 2: Industry Impact - The shortage of glass fiber is described as one of the biggest bottlenecks facing the electronics and AI industries in 2026, according to industry insiders [4][6]. - Companies like Qualcomm and AMD are also feeling the pressure, with Qualcomm exploring alternative suppliers to mitigate the supply constraints [6][7]. - The specific type of glass in demand is low thermal expansion coefficient (CTE) glass, known for its stability and high-speed data transmission capabilities, which are critical for AI computing and advanced processors [7].

Core Viewpoint - SK Hynix has introduced "StreamDQ" as a solution to address the upcoming era of customized High Bandwidth Memory (HBM), enhancing data processing performance by offloading certain tasks from GPUs to HBM [1][2]. Group 1: StreamDQ Technology - StreamDQ technology aims to commercialize customized HBM, with a focus on promoting it to major clients like NVIDIA at CES 2026 [2]. - The technology transfers some controller functions from existing GPUs to the HBM substrate, potentially improving the performance and efficiency of system semiconductors [2]. - StreamDQ allows real-time dequantization of data as it flows through HBM, which can significantly reduce the bottleneck traditionally faced by GPUs during the dequantization process [3]. Group 2: Performance Enhancements - The implementation of StreamDQ is expected to enhance the inference speed of large language models (LLMs) by approximately seven times, addressing the memory bottleneck that previously consumed up to 80% of the overall inference time [3]. - By processing data closer to memory, the overall efficiency of semiconductor systems is anticipated to improve significantly, aligning with the concept of processing near memory (PNM) [3]. Group 3: Manufacturing and Integration - SK Hynix is leveraging advanced processes from Taiwan's TSMC for adding GPU controllers and other components to the base chip without facing major challenges [2]. - The integration of UCIe interfaces on the chip substrate enhances chip integration, allowing for the segmentation of chips into functional units for manufacturing [2].

Core Viewpoint - The global memory market is experiencing a severe shortage due to explosive demand for high-performance memory from AI servers, which is expected to drive price increases throughout 2026 [1][3]. Group 1: Market Dynamics - Major memory manufacturers, including Samsung, SK Hynix, and Micron, have initiated capacity expansion plans, but these efforts are insufficient to alleviate the current memory shortage [1]. - Samsung plans to increase its DRAM wafer input to nearly 8 million pieces by 2026, a 5% increase from 2025 [1]. - SK Hynix's DRAM production is expected to rise from 5.97 million pieces in 2025 to 6.48 million pieces in 2026, an 8% increase, driven by expansion at its Cheongju M15X plant [2]. Group 2: Production Challenges - Despite increased wafer input, actual chip output faces technical bottlenecks, particularly for Samsung as it transitions to 10nm 6th generation DRAM, which may temporarily reduce production capacity [2]. - The supply-demand imbalance is critical, with DRAM suppliers meeting only about 60% of customer demand, and server-specific DRAM satisfaction rates falling below 50% [3]. Group 3: Price Trends - DRAM contract prices are expected to rise sharply, with a forecasted increase of 55% to 60% in Q1 2026 [3]. - NAND Flash contract prices are also projected to increase by 33% to 38% during the same period, particularly driven by server demand [3]. - Despite a weak PC market, DRAM prices for PCs are anticipated to continue rising due to reduced supply allocations to PC manufacturers [4]. Group 4: Long-term Outlook - Experts believe that the current supply crisis will not see fundamental relief until Samsung's P4 factory becomes operational, which is not expected until after 2027 [4][5]. - SK Hynix also requires the launch of its Yongin semiconductor cluster to significantly enhance production capacity [5].

如果您希望可以时常见面，欢迎标星收藏哦~ 2026年1月，名古屋大学低温等离子体科学研究中心肖新安教授和堀胜教授领导的研究团队与东 京电子宫城株式会社合作，宣布他们已阐明了使用冷却晶片和氟化氢（HF）等离子体的反应离子 刻蚀（RIE）工艺的机理。与传统工艺相比，二氧化硅（ SiO2）薄膜的刻蚀速度提高了五倍。此 外，使用HF作为刻蚀气体也降低了对环境的影响。 在制造结构精细复杂的器件时，例如 GAA（环栅）晶体管和 3D NAND 闪存，传统 RIE 工艺的 蚀刻速率会显著降低。为了解决这个问题，我们提出了一种采用 HF 等离子体并冷却晶圆的新工 艺。 当衬底温度保持在较低水平，例如-60℃时，刻蚀气体HF和反应产物水（H2O ）会吸附在SiO2表 面。研究发现， H2O起到催化剂的作用，将SiO2的刻蚀活化能降低到几乎为零。 点这里加关注，锁定更多原创内容 *免责声明：文章内容系作者个人观点，半导体芯闻转载仅为了传达一种不同的观点，不代表半导体芯闻对该 观点赞同或支持，如果有任何异议，欢迎联系我们。 10万亿，投向半导体 芯片巨头，市值大跌 黄仁勋：HBM是个技术奇迹 Jim Keller：RISC-V一定 ...