Group 1 - The core viewpoint of the article emphasizes that Xiaomi's self-developed 3nm chip marks the beginning of its chip development journey, with the "Xuanjie O1" chip winning the highest award at the 2025 Xiaomi "Ten Million Technology Award" [1] - Lei Jun expressed the importance of continuous technological innovation to ensure Xiaomi's sustained success and commitment to investing 100 billion yuan in core technology research over the past five years, with actual investment reaching approximately 105 billion yuan [1] - Starting this year, Xiaomi has pledged to invest 200 billion yuan in research and development over the next five years, focusing on core technologies such as chips, AI, and operating systems to build a comprehensive ecosystem [1] Group 2 - Lei Jun hinted at a potential breakthrough in technological innovation in 2026, with hopes of achieving a "grand convergence" of self-developed chips, operating systems, and AI models in a terminal product [1]

Core Viewpoint - Nvidia's CEO Jensen Huang's speech at CES impacted Credo Technology's stock negatively, but market analysts believe the stock may have been unfairly punished [1]. Group 1: Market Reaction - Credo Technology, a leader in Active Electrical Cables (AEC), saw its stock drop by 5.2% on January 6 and further decline by over 6% on January 7 due to concerns that Nvidia's new Vera Rubin system would shift away from copper cables to optical interconnect solutions [1]. - After hitting a low of $125 on January 7, Credo's stock experienced a dramatic turnaround, closing up 6.05% at $141, even surpassing its closing price from January 5 [1]. Group 2: Future Outlook - Analysts clarified that Huang still favors expanding the use of copper cables within cabinets to maximize connection speeds between GPUs [1]. - Despite Nvidia's movement towards more optical interconnect solutions, Credo is actively developing its own optical and photonic solutions, which could mitigate concerns about its future [2]. - Analysts recommend investors consider buying Credo's stock if it falls below $140 [2].

Core Viewpoint - The article discusses the recent resurgence of Borui Jinxin, a chip company that previously faced financial difficulties but has now received new investments and cleared debts, indicating a potential turnaround in the Chinese chip industry, particularly for ARM CPU manufacturers [1][9]. Group 1: Industry Trends - The Chinese chip design industry has seen significant growth, with the number of chip design companies increasing from 1,780 in 2018 to 2,218 in 2020, and projected to reach 3,901 by 2025, reflecting a compound annual growth rate [2]. - Factors driving this growth include the rise of artificial intelligence, the maturity of RISC-V, and the electrification and intelligence of vehicles, with ARM high-performance chips being a standout segment [2][10]. - ARM architecture has gained recognition across various applications, including embedded systems, mobile devices, PCs, and servers, particularly in the context of AI and cloud computing [2][10]. Group 2: Market Dynamics - IDC predicts a 70% increase in server shipments based on ARM architecture by 2025, with ARM aiming for nearly 50% of computing power in top-tier data centers to be ARM-based by the same year [3]. - Major cloud service providers like AWS, Google, and Microsoft are increasingly adopting ARM server CPUs, indicating a favorable market environment for ARM chips [4][10]. - Despite the challenges faced by domestic companies like Borui Jinxin, the demand for ARM server chips is on the rise in China, driven by the need for domestic alternatives and advancements in AI technology [10]. Group 3: Challenges and Opportunities - The design and production of advanced SoCs (System on Chips) present significant challenges, particularly as the industry moves towards sub-5nm processes, which dramatically increase design costs and complexity [6][8]. - The cost of designing chips has escalated, with design costs rising from $90 million for 16nm to $249 million for 7nm, and projected to reach $725 million for 2nm, making it difficult for startups to secure funding [8][9]. - Despite past difficulties, Borui Jinxin's recent investment and strategic support from state-owned enterprises may provide a unique opportunity to leverage its ARM instruction set authorization and compete in the server chip market [11].

如果您希望可以时常见面，欢迎标星收藏哦~ 现代汽车集团机器人实验室开发出一款利用人工智能技术在设备本地运行任务的处理器。 实验室此前曾与他人合作开发了一款人工智能控制器，目前该控制器应用于首尔东部Factorial Seongsu工厂的人脸识别系统Facey以及DAL-e送货机器人。 （来源 ：半导体芯闻综合 ） 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。 推荐阅读 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 该处理器功耗低于5瓦，能够利用提供的数据进行实时检测、识别和决策。 该芯片无需云端或网络连接，因此可以在网络连接不稳定或缺失的地下停车场或物流中心等场所使 用。实验室表示，由于它是设备端芯片，因此比依赖云计算的芯片速度更快、安全性更高。 现代汽车集团机器人实验室负责人兼副总裁玄东镇表示："现代汽车集团机器人实验室不仅仅是制 造机器人，我们更致力于构建一个可持续的机器人生态 ...

Group 1 - TSMC announced a consolidated revenue of approximately NT$335 billion for December 2025, representing a month-over-month decrease of 2.5% but a year-over-year increase of 20.4%, marking a record high for the same period and exceeding NT$300 billion for six consecutive months [1] - The company's revenue for 2025 is projected to exceed NT$3 trillion, reaching approximately NT$3.8 trillion, with a year-over-year growth of 31.6%, setting a new annual revenue record [1] - In Q4 2025, TSMC's revenue surpassed NT$1 trillion, reaching NT$1.04 trillion, with a quarter-over-quarter increase of 4.8% and a year-over-year increase of 16.3%, also exceeding financial forecasts [1] Group 2 - TSMC is accelerating production capacity expansion in advanced processes and advanced packaging due to insufficient capacity amid strong market demand, while reallocating resources from less urgent mature processes [3] - The company is transferring some mature process equipment from Taiwan to its 12-inch facility in Singapore to create more space for advanced process equipment [3] - TSMC's CEO confirmed that Qualcomm is facing capacity shortages, leading the company to negotiate with Samsung for the production of its next-generation processors using Samsung's 2nm process [3] Group 3 - TSMC's board approved an investment of over US$179.25 million to acquire land for a new factory in Phoenix, Arizona, covering approximately 365.27 hectares, which, combined with an existing site, totals over 810 hectares, larger than the Hsinchu Science Park [4] - The land acquisition is expected to support an investment of approximately US$165 billion in building three additional 12-inch wafer fabs, two advanced packaging plants, and a research center to meet market demand [4] - TSMC is also selling machinery to World Advanced Semiconductor Manufacturing Corporation (VSMC) for an estimated US$71 million to US$73 million to enhance its 12-inch wafer capacity [4] Group 4 - VSMC, in collaboration with NXP Semiconductors, plans to establish a 12-inch wafer fab in Singapore with a total investment of approximately US$7.8 billion, enhancing geographical resilience and accelerating the semiconductor ecosystem in Singapore [5] - TSMC stated that the newly acquired land in Arizona is primarily for operational and production use, but the company is currently in a quiet period and cannot comment on the transfer of equipment to Singapore [5]

Core Viewpoint - DISCO's shipment value continues to decline, although it remains at the third-highest level in history, leading to a drop in stock price despite a year-to-date increase of approximately 14% [1][2]. Group 1: Shipment Performance - For the third quarter of 2025 (October-December), DISCO's non-consolidated shipment value decreased by 0.8% year-on-year to 90.1 billion yen, marking the second consecutive quarter of decline, yet it is still the third-highest quarterly figure recorded [1]. - Compared to the previous quarter (July-September 2025), the shipment value increased significantly by 16.3% [1]. - The demand for precision processing equipment, particularly for generative AI applications, remains high, showing growth compared to the previous quarter [1]. Group 2: Revenue Performance - DISCO's non-consolidated revenue for the third quarter of 2025 increased by 13.8% year-on-year to 88 billion yen, and rose by 3.1% compared to the previous quarter [1]. - The estimated consolidated shipment value for the third quarter of 2025 is projected to decrease by 8% year-on-year to 101.6 billion yen, although it will still be the third-highest quarterly figure in history [2].

如果您希望可以时常见面，欢迎标星收藏哦~ 针对美国从原本封锁高端芯片输中到开放英伟达H200芯片销售，是否将影响中国科技自主进程， 中国半导体行业协会副理事长魏少军表示，中国半导体产业必须对此保持高度警惕，坚决不被其表 像所迷惑，更不能因此动摇在先进制程等领域坚持国产化道路的信心与决心。 据环球时报报导，H200芯片是英伟达仅次于Blackwell系列的次顶级芯片，在拜登政府时期被禁止 对中出口。川普政府在限制出口最先进Blackwell系列芯片与"完全不出口反而助长中国更快实现国 产替代"之间寻求平衡，最终允许有条件放开H200芯片对中出口，但需获得美国商务部出口许可审 批。 报导表示，在这个角度上，美国变幻不定的态度，在高端芯片上时而解禁时而施压，让使用者难以 确认其真实战略意图。最近的所谓放松究竟是推动良性互动的讯号，还是意图扰乱中国发展节奏、 让中国放松警惕的新策略，魏少军说，中国半导体产业必须对此保持高度警惕，坚决不被其表像所 迷惑，更不能因此动摇在先进制程等领域坚持国产化道路的信心与决心。 但魏少军也指出，"从行业发展角度看，高端算力资源的合理流动有助于促进人工智慧等前沿技术 的应用探索。科技无国 ...

Tesla 将会有一个2 纳米晶圆厂，而且我可以在晶圆厂里吃起司汉堡和抽雪茄。 当Elon Musk 说出这句话时，访谈者的反应是：「哦，拜托。」但Musk 是认真的。他不只打赌 Tesla 会建造颠覆传统的晶圆厂，更重要的是，他判断全球将遭遇「芯片墙」，即使台积电2025 年砸下400 亿美元、计划兴建10 座新晶圆厂，他仍认为这不够激进。 核心矛盾在于：AI 算力需求呈指数增长，但芯片产能和电力基础设施的扩张都是线性的。这不是 技术问题，而是产业逻辑的根本性转变。 第一阶段： 从制程竞赛到产能竞赛 过去二十年，半导体产业遵循一个清晰的逻辑：制程领先就能赢得市场。谁能率先量产7nm、 5nm、3nm，谁就能垄断高端市场。这套逻辑塑造了整个产业的投资决策，大量资金投入研发，追 求每一代制程的极致微缩。 但在这场访谈中指出了一个被忽视的事实：这套逻辑在AI 时代已经不适用了。 Musk 说：「制程微缩已经到了收益递减阶段」。 从3 纳米到2 纳米，性能改进只有10% ，不是3:2 的比例改进，只有10%。 Jensen Huang 说过， 摩尔定律已死。你不能只靠把东西做得更小就能做得更好，原子数量是离散的，你 ...

Core Viewpoint - NVIDIA's CEO Jensen Huang emphasized the ongoing necessity of HBM (High Bandwidth Memory) despite the advantages of SRAM (Static Random Access Memory) and other cost-saving solutions for AI workloads, indicating that each memory type has its own strengths and limitations [1][2]. Group 1: Memory Technology Insights - Huang noted that while SRAM can significantly enhance efficiency for certain workloads, its capacity limitations may hinder its effectiveness in large-scale AI production environments compared to HBM, which offers better bandwidth and density [1][2]. - The transition of workload pressures among different memory types and interconnects is dynamic, with varying demands based on the architecture of AI models, such as mixed expert models and multi-modal models [2]. Group 2: Market Dynamics and Cost Considerations - Despite customer complaints regarding the costs of HBM and GPUs, Huang believes there is no inherent conflict between these costs and the demand for advanced memory solutions, suggesting that the market for NVIDIA's products remains robust [3].

Core Viewpoint - TSMC's advanced 3nm process is experiencing high demand, leading to a temporary halt in new project kick-offs for this technology, as existing capacity is fully utilized and cannot meet the influx of orders [2] Group 1: TSMC's 3nm Process - TSMC has raised prices for its 3nm process and has stopped new project kick-offs due to full order books and limited capacity [2] - The company is strategically encouraging clients to shift their product planning towards the 2nm process, which offers better cost structures and efficiency [2] - The 3nm family of products is currently filled with orders from AI GPUs, cloud data center ASICs, and high-end mobile processors [2] Group 2: Transition to 2nm Process - The 2nm process is seen as a critical turning point for TSMC, introducing nanosheet transistor architecture that significantly improves performance, power consumption, and density [3] - The introduction of atomic layer deposition (ALD) in the 2nm process does not significantly increase the number of EUV exposure layers, making the overall manufacturing cost more competitive compared to the 3nm process [3] - The GAAFET process represents a significant upgrade in wafer manufacturing, requiring advanced techniques to overcome challenges in material layering and etching [3]