转自：新华财经 1月5日，黄仁勋再度向全球抛出"重磅炸弹"Rubin，其训练性能是Blackwell的3.5倍，AI软件运行性能飙升5倍；但推理成 本，仅为前代的1/10。但这种惊人的表现是被逼出来的，TPU和可重构数据流架构（RPU）的崛起，正在侵蚀英伟达的统 治地位。 去年11月，Meta拟采用谷歌TPU的传闻一出，英伟达一夜蒸发数千亿美元。焦头烂额的黄仁勋，不得不光速出手，将可重 构数据流芯片公司Groq收入囊中。如今，Groq的并购与Rubin的发布，共同指明了AI芯片的"收敛时刻"：更高性能的通用 芯片，才是所有人要奔赴的终极战场。 疯抢高阶TPU 2025年圣诞节，英伟达豪掷200亿美元（约1400亿人民币），买了一家"非GPU"公司——Groq。英伟达以其持有现金三分 之一的巨资，打包买走了Groq的核心技术、图纸和核心人才。 三个月前，Groq的估值还仅为69亿美元。如今，黄仁勋抛出3倍溢价，对Groq志在必得。而Groq主攻的，是特有的LPU芯 片技术，即用软件定义硬件的"可重构数据流架构"。这种独特的设计，能让LPU在处理大模型时，实现Token"瞬时"+"准 时"的吞吐，超越GPU、TP ...

Group 1: Industry Overview - The CES 2026 showcased over 4,000 companies, focusing on AI and its integration into various products, including PCs, home appliances, and foundational chips [1] - The combined market capitalization of the four major semiconductor companies—NVIDIA, AMD, Qualcomm, and Intel—reaches nearly $5.3 trillion, indicating significant investor interest in AI technologies [1] - Concerns about the over-exuberance of the AI market persist, with 2026 expected to be a year for evaluating genuine value creation rather than mere speculation [1] Group 2: NVIDIA Developments - NVIDIA unexpectedly launched its next-generation AI chip platform "Rubin," which significantly enhances training and inference performance by 3.5 times and 5 times, respectively, compared to its predecessor [2] - The company reported over $500 billion in orders for its Blackwell and Rubin architectures for 2025-2026, reflecting strong demand for computing power in data centers [2] - NVIDIA introduced the open-source autonomous driving model "Alpamayo," which features 10 billion parameters and aims to streamline the processing from sensor input to driving commands [3] Group 3: AMD Initiatives - AMD adopted an aggressive "stacking" strategy, showcasing the Helios AI rack, which integrates 72 MI455X accelerator chips and 31TB of HBM4 memory, claiming it to be the best AI rack globally [5][6] - The new generation of chips is expected to deliver up to a 10-fold performance improvement under various workloads, aided by a hybrid 2nm and 3nm manufacturing process [6] - AMD signed a significant infrastructure procurement agreement with OpenAI, valued at 6 gigawatts, marking a substantial entry into the core supply chain of leading AI model developers [6] Group 4: Qualcomm and Intel Strategies - Qualcomm is diversifying its strategy to reduce reliance on the smartphone market, launching the Snapdragon X2 Plus chip aimed at the mid-range laptop market, with a focus on energy efficiency and 5G connectivity [8] - Intel introduced the "Panther Lake" third-generation Core Ultra processor, its first chip manufactured using the 18A process, which is crucial for regaining manufacturing leadership [9] - The third-generation Core Ultra processors are set to be released in the second quarter of 2026, targeting industrial, medical, and smart city applications [10]

英伟达没有在CES 2026上发布任何新的显卡。 取而代之的，黄仁勋用近两个小时，详细阐述了一个名为Vera Rubin的全新AI超算架构，以及一组足以改写行业规则的数字： 在Rubin上运行AI推理，吞吐量提升十倍，训练万亿参数模型所需的GPU数量，可以减少到上一代Blackwell的四分之一，而每个token的成本降 至十分之一。 这件事可能埋了个信号。 关于没有新显卡与Rubin的商业模式 理解Rubin，关键在于转变视角。 它不是一个更快的GPU，你可以把它视为一套极度垂直整合的AI计算系统。英伟达设计了六颗功能各异、但深度耦合的专用芯片去进行打包： Vera CPU（负责AI数据流调度）、Rubin GPU（核心算力单元）、NVLink 6（超高带宽内部互联）、ConnectX-9 SuperNIC（AI专用网络）、 BlueField-4 DPU（卸载存储与安全任务）、Spectrum-6以太网交换芯片。 六块芯片协同工作，目标是将整个数据中心机柜整合为一台无缝的"巨型AI计算机"。 Rubin解决的，是系统规模化的难题，不是单颗芯片的性能极限。让算力的堆叠，从"手工组装赛车引擎"变为"标准化汽车 ...

NVIDIA Corporation (NASDAQ:NVDA) is one of the AI Stocks Making Waves on Wall Street. One of the biggest analyst calls on Monday, January 5, was for Nvidia Corporation. UBS reiterated the AI chipmaker as “Buy,” heading into the Consumer Electronics Show. “NVDA is hosting a keynote on Monday night, where we expect it may give a little more information about Rubin and the QuantumX/SpectrumX silicon photonics switches, and perhaps further outline how the new Groq technology licensing deal fits into its AI fa ...

January 6, 2026 09:56 PM GMT Asia-Pacific Technology | Asia Pacific AI Supply Chain: CES implications, ASIC production, China AI chips Morgan Stanley does and seeks to do business with companies covered in Morgan Stanley Research. As a result, investors should be aware that the firm may have a conflict of interest that could affect the objectivity of Morgan Stanley Research. Investors should consider Morgan Stanley Research as only a single factor in making their investment decision. For analyst certificati ...

共探新营销，共创新可能 撰文 | 杜 咏 芳 编辑 | 黄 大 路 设计 | 甄 尤 美 拉斯维加斯会议中心始建于1959年，期间经过多次翻修和扩建，但当你走在其中时，依旧会感受到浓浓的时代感，一种特有的美式老派感扑面而 来。 但就是在这个有些"过时"的建筑里，每年都在展示着全球最先进的技术。 2026年1月6日至9日（太平洋标准时间），全球科技界的目光将再度锁定拉斯维加斯，国际消费电子展（CES）正式拉开帷幕。 不过，与往年相比，本届CES略显冷清：参展商数量较高峰时期减少了1000余家，现场人流与声量均有所回落。 本次展会设置了先进制造展示区、全新CTA展区、CES Foundry、创作者展区、移动技术展区及初创展区等多个特色板块，以及制造、技术与标 准、穿戴设备三个核心会议轨道，覆盖科技产业核心领域。 这个曾以电视、电脑与可穿戴设备定义消费潮流的舞台，在过去数年亦成为电动汽车发布的主秀场。 但今年，风向已然转变。一个清晰的信号正在释放：电动化的声量逐渐让位，一个更底层、更具颠覆性的概念走向前台，即"物理人工智 能"（physical AI）。 与处理数据或生成内容的人工智能不同，物理人工智能结合了传感器 ...

Core Insights - Nvidia emphasizes that the AI boom is expanding rather than cooling, with a shift from "AI models" to "AI systems" that can be deployed at scale and cost-effectively in various real-world applications [1] - The company introduced Rubin, its next-generation data-center platform, which is a six-part stack designed for integrated performance rather than standalone upgrades [2][3] AI Market Dynamics - The AI market is transitioning from training large models to focusing on inference, where the ability to run models reliably and cost-effectively at scale is becoming crucial [4] - Nvidia positions itself as the infrastructure backbone of the AI economy, moving beyond being just a chip supplier to creating "AI factories" for industrial-scale intelligence production [5] System Integration and Competitive Edge - The launch of Rubin is accompanied by new networking and data-processing technologies, highlighting the importance of whole-system optimization for competitive advantage [6] - As hyperscalers design their own silicon for inference, Nvidia's strategy is to enhance its offerings to make itself harder to replace [7] Physical AI and Real-World Applications - Nvidia is focusing on "physical AI," which involves systems that can perceive, reason, and act in the real world, emphasizing robotics and autonomous vehicles [8] - The company has been building software platforms around automotive and robotics, anticipating that technology and demand are finally aligning [9] Future Outlook and Investment Implications - Nvidia's confident message at CES indicates a shift in the AI industry towards cost efficiency and sustainable deployment, favoring established players with engineering resources [11] - The next phase of AI growth is expected to resemble industrialization rather than a gold rush, emphasizing competition and the ability to deliver intelligence at lower costs [12] - Nvidia's integrated platforms and focus on physical AI are seen as key to maintaining its central role in the evolving market, with real-world deployment over the next 12 to 18 months being critical for validation [13]

Good morning. Many boards are approving AI strategies without clear visibility into whether the underlying controls actually work, leaving CFOs exposed when regulators, auditors, or investors ask for proof, according to new research. In the private sector, health care appears to face the steepest challenge.Kiteworks, a technology security company, has released its “Data Security and Compliance Risk: 2026 Forecast Report,” based on a survey of 225 security, IT, compliance, and risk leaders across 10 industri ...

Investment Rating - The report maintains an "Overweight" rating for the copper-clad laminate industry, indicating an expectation that the industry index will outperform the market index by more than 10% over the next six months [96]. Core Insights - The copper-clad laminate market is expected to recover in 2024, with global sales projected to reach $15.013 billion, representing a year-on-year growth of 17.90%. This recovery is attributed to the resurgence in consumer electronics demand and increased requirements for high-end copper-clad laminates driven by AI computing power [6][27]. - A new wave of price increases is anticipated in the copper-clad laminate sector, driven by rising raw material costs, high demand from downstream PCB manufacturers, and the concentrated market share of laminate manufacturers, which grants them significant pricing power [6][47]. - AI is driving the high-end product development in the copper-clad laminate market, with expectations for increased value as new platforms from companies like NVIDIA adopt advanced materials [65][71]. Summary by Sections 1. Market Recovery and Applications - Copper-clad laminates are widely used in various sectors, including consumer electronics, computers, communications, and automotive electronics. The market is set to recover in 2024, reversing a two-year decline [14][27]. 2. Price Increase Trends - Historical data shows that the copper-clad laminate industry has experienced two major price increase cycles in the past decade, primarily driven by rising raw material costs and strong end-user demand. Current trends indicate that manufacturers are already implementing price hikes due to increased costs of copper and other materials [38][44][55]. 3. AI-Driven High-End Product Development - The demand for high-performance copper-clad laminates is increasing due to AI computing requirements. The introduction of new materials, such as M8.5+ and M9, is expected to significantly enhance product value and performance [65][71]. 4. Investment Recommendations - The report suggests focusing on two main investment lines: companies benefiting from the price increases of conventional copper-clad laminates and those with high-end laminate or electronic copper foil product reserves. Key companies identified include Shengyi Technology, Nanya New Material, Huazheng New Material, Jin'an Guoji, Defu Technology, and Tongguan Copper Foil [93].

三星电子半导体事业部（DS事业部）的员工将于明年初收到高达年薪一半的绩效奖金。 30日，三星电子公布了2025年超额利润激励（OPI）的预计发放比例。DS事业部的发放比例预计为年薪的 43%至48%，是去年14%的三倍多。 这被解读为通用DRAM价格上涨和第五代高带宽内存（HBM3E）全面供应带来的性能显著提升。 公众号记得加星标⭐️，第一时间看推送不会错过。 移动体验（MX）部门的薪酬比例设定为45-50%，较去年（40-44%）略有提高。负责电视业务的视觉显示 （VD）部门的薪酬比例设定为9-12%，较去年的27%有所提高。去年薪酬比例均为9 %的数字家电（DA） 部门、网络部门和医疗设备部门，今年的薪酬比例均维持在9-12%的预期水平。 三星电子采用两种基于绩 效的奖金制度。 今天公布的OPI奖金最高，当所属部门的年度业绩超过目标时，奖金最高可达年薪的 50%，但需在超额利润的20%以内。 此外，业绩目标奖励（TAI）分上下半年发放，根据绩效差异而定， 最高可达月基本工资的100%。 本月22日公布的下半年TAI方案中，存储器业务部门和半导体研发中心的 TAI为100%，系统LSI和代工部门的TAI为2 ...