作者 | 赵雨荷 编辑 | 硬 AI OpenAI周四发布GPT-5.3-Codex，称其为迄今最强编程代理，并将发布时间安排在Anthropic发布其旗舰升级模型 Claude Opus 4.6的同一时刻。新模型多项基准测试成绩大幅领先，并首次参与自身训练与部署。业内认为，这标志着围 绕企业级软件开发的AI编程大战正式打响。 硬·AI OpenAI周四发布GPT-5.3-Codex。该公司称，这是迄今为止能力最强的编程代理。值得注意的是，此次 发布的时间点被精准安排在Anthropic推出其旗舰模型升级版Claude Opus 4.6的同一时刻。 媒体称，两大 模型同步亮相，被业内观察人士视为"AI编程大战"的第一枪——这是一场围绕企业级软件开发市场展开的 高风险争夺战。 OpenAI首席执行官Sam Altman在模型发布几分钟后就在X上写道： "我非常喜欢用这个模型来开发，它带来的进步感受，远远超过基准测试所显示的幅度。" "看着我们用5.3-Codex来开发5.3-Codex，从而把发布速度提升到这么快，真的令人震撼，这毫无 疑问预示着未来的发展方向。" 媒体表示，模型本身参与了自身的构建，被视为AI ...

Group 1 - The core viewpoint of the article highlights that major US tech companies plan to invest a record $660 billion in AI infrastructure by 2026, which exceeds Israel's GDP and represents a 60% increase from $410 billion in 2025 and 165% from $245 billion in 2024 [2][3][6] - The massive capital expenditure plans have led to a significant market reaction, with Amazon, Google, and Microsoft collectively losing $900 billion in market value following their quarterly earnings reports [3][4][6] - Apple's strategy of minimal capital expenditure has allowed it to avoid the sell-off, reporting record revenue of $144 billion, while its capital spending decreased by 17% to $2.4 billion in the last quarter [10][11] Group 2 - Investors are increasingly anxious about the long-term return on investment from AI, as rising capital expenditures suggest that achieving AI's full potential may require more time and resources [8][9] - The market sentiment has shifted, with the Nasdaq index dropping 4% over five days, reflecting concerns about the sustainability of tech stocks amid rising capital expenditures [9][13] - Despite strong revenue growth, companies like Microsoft and Amazon face pressure to demonstrate that their significant spending will yield attractive returns, as investor patience wanes [9][10] Group 3 - Nvidia, as the highest-valued public company, is expected to face a turbulent market when it releases its earnings report, with investors looking for signs that the escalating capital expenditures will translate into revenue growth [13][14] - Meta has also announced plans to double its capital expenditures to $135 billion, but its stock gains were short-lived amid broader market sell-offs [13][14] - Overall, the combined annual revenue growth of the four companies reached 14% to $1.6 trillion, which is insufficient to alleviate the prevailing market pessimism [13]

在刚刚结束的2025年第四季度财报电话会上，尽管 亚马逊AWS跑出了三年最快增速 ，但市场被一个数字 吓懵了—— 2026年预计约2000亿美元的资本开支 。这个数字不仅比分析师预期的这一水平高出近40%， 更是超过了谷歌刚刚宣布的 1850亿美元上限 。 在美股科技股正经历"AI恐慌"与杀估值的敏感时刻，亚马逊的这一指引引发了盘后股价超11%的剧烈震 荡。然而，在长达一个小时的电话会议中，亚马逊高管团队，特别是CEO Andy Jassy，用了大量篇幅来 论证这种激进投入的必要性，并在自研芯片、AI变现逻辑以及与OpenAI的合作等方面释放了大量增量信 息。 尽管AWS积压订单激增40%至2440亿美元，但市场被亚马逊2026年预计2000亿美元的天量资本开支吓退，股价盘后大 跌11%。CFO预警低轨卫星项目将在Q1带来10亿美元额外成本，严重拖累利润指引。但CEO辩护称巨额投入并非"唐吉 诃德式狂想"，强调AI产能"只要安装就能变现"，并押注AI代理重塑电商体验。 硬·AI 作者 | 龙 玥 编辑 | 硬 AI 01 2000亿美元花在哪？ "我们在拼命地安装产能" Andy Jassy明确表示，这200 ...

Core Viewpoint - The sentiment on Wall Street towards OpenAI has dramatically reversed, with previously favored stocks now facing sell-offs, while Alphabet has emerged as a major winner due to strong financial returns from its AI business [1][4][12]. Group 1: Market Sentiment Shift - Wall Street analysts indicate a narrative shift favoring Google over OpenAI, as concerns grow over OpenAI's financial sustainability and its impact on closely tied companies like Microsoft and Oracle [1][5][6]. - Oracle's stock has plummeted approximately 49% since last October, heavily reliant on OpenAI contracts, while Microsoft's stock has dropped over 20% during the same period [2][5]. - In contrast, Alphabet's stock has surged about 36% since last October, reflecting a positive market sentiment towards its AI capabilities [2][12]. Group 2: Alphabet's Financial Performance - Alphabet has successfully generated returns from its AI investments, with CEO Sundar Pichai stating that AI infrastructure is driving overall revenue growth [9][10]. - As of the end of December, Google Gemini applications had over 750 million monthly active users, and the enterprise version of Gemini reached 8 million paid licenses, contributing to a 48% increase in Google Cloud revenue [10][12]. - Despite initial concerns over a significant increase in capital expenditures projected to reach $175 billion to $185 billion by 2026, strong cloud performance and AI-driven growth have restored investor confidence [10][12]. Group 3: Market Dynamics and Tech Sector Analysis - A Deutsche Bank report highlights a "cleansing" phase in AI investments, indicating that the S&P 500's performance is largely supported by Google, with many tech stocks experiencing significant declines [4][12][14]. - The report emphasizes a shift from a mindset where all tech stocks were seen as winners to a more brutal reality of clear winners and losers in the market [12][14]. - Alphabet's stock has risen nearly 75% over the past six months, translating to approximately $1.7 trillion in market value growth, while most other companies in the tech sector have seen their market values stagnate [14][15].

Core Viewpoint - Google is heavily investing in AI, with projected capital expenditures reaching $175 billion to $185 billion by 2026, primarily for servers and data centers, as part of its commitment to the AI arms race [2][4][30]. Group 1: Capital Expenditure and AI Investment - The significant capital expenditure forecast for 2026 is aimed at supporting the development of Google DeepMind's advanced models and meeting the surging demand from cloud customers [6][30]. - Approximately 60% of the capital expenditure will be allocated to servers, while 40% will go towards data centers and network equipment [6][30]. - The increase in infrastructure investment is expected to lead to a substantial rise in depreciation expenses, which will exert ongoing pressure on profit margins [6][56]. Group 2: AI's Role in SaaS and Market Dynamics - In response to fears that AI could disrupt traditional software business models, Google’s CEO emphasized that Gemini serves as an "engine" for SaaS companies rather than a terminator, with 95% of the top 20 SaaS companies already utilizing Gemini [8][39]. - The Gemini App has surpassed 750 million monthly active users, indicating strong engagement and the potential for monetization through a new "agentic commerce" model [12][29]. Group 3: Partnership with Apple - Google has confirmed a significant partnership with Apple, positioning itself as the preferred cloud provider and collaborating on the development of the next-generation Apple Foundation model based on Gemini technology [9][10][39]. Group 4: Cloud Business and AI Infrastructure - Google Cloud's revenue grew by 48%, driven by strong demand for AI infrastructure, with a notable increase in new customer acquisition and larger customer commitments [19][36]. - The company is employing a dual strategy of using both self-developed TPUs and NVIDIA GPUs to enhance its AI capabilities [19][36]. Group 5: YouTube and Advertising Growth - YouTube's advertising revenue increased by 9%, with significant growth in subscription services, particularly YouTube Music Premium [21][46]. - The platform continues to lead in streaming, with Shorts averaging over 200 billion views daily, indicating a robust engagement trend [21][46]. Group 6: Efficiency and Internal Operations - Google has achieved a 78% reduction in service unit costs for Gemini through model optimization and efficiency improvements [22][56]. - Approximately 50% of the code is now written by coding agents, allowing engineers to focus on higher-level tasks and improve operational efficiency [22][66].

Core Viewpoint - The article discusses the contradictory market narratives surrounding AI and software, particularly how the decline in AI chip stocks, represented by companies like Nvidia, is driven by fears of diminishing returns on AI investments while simultaneously suggesting that AI adoption will significantly increase productivity, which is logically inconsistent [3][7]. Group 1: AI and Software Market Dynamics - The narrative that "AI will defeat software" has led to a broad sell-off in the software sector, which has negatively impacted AI chip stocks as well [4][6]. - Nvidia's stock fell by 3.41%, Broadcom by 3.83%, Micron by 10%, and SanDisk by over 15%, indicating a significant market reaction [4]. - Bank of America analyst Vivek Arya argues that the simultaneous decline in chip and software stocks reflects an overreaction from the market, as the two outcomes cannot coexist [7][11]. Group 2: AI Investment Trends - Arya notes that despite the current market fears, AI spending is expected to increase, with cloud capital expenditures projected to grow by 69% year-over-year by 2025, significantly higher than the initial 20% to 30% growth expectations [9]. - The article emphasizes that even if AI models show impressive performance, it may take years to prove their value in terms of productivity, suggesting that AI investment is unlikely to slow down in the short term [9]. - The chip sector is expected to continue benefiting from the AI investment cycle, as enterprise-level AI adoption is still in its early stages, with more sovereign nations increasing their AI investments [9][10]. Group 3: Implications for Investors - The current market downturn appears to be driven by narrative-based correlations, where the simultaneous sell-off of software and AI chips reflects a risk-off sentiment rather than fundamental changes in AI capital spending and demand [11]. - Investors in Nvidia and similar companies need to discern whether price fluctuations are due to actual changes in AI capital expenditures or merely a reaction to the prevailing narrative of AI's impact on software [11].

Core Viewpoint - AMD's Q4 revenue exceeded $10 billion, achieving a record high with a year-over-year growth rate of over 30%, although EPS growth has slowed down [2][3][8] Revenue and Earnings Performance - Q4 revenue reached $10.27 billion, a 34% year-over-year increase, slightly lower than the 36% growth in Q3 [8] - Adjusted EPS for Q4 was $1.53, a 40% year-over-year increase, surpassing analyst expectations by nearly 16% [8][11] - Operating profit for Q4 was $2.85 billion, a 41% year-over-year increase, exceeding analyst expectations [13] Business Segments - Data center revenue hit a record $5.4 billion in Q4, a 39% year-over-year increase, significantly up from 22% growth in Q3 [16] - Client and gaming segments contributed approximately $3.9 billion in revenue, with a year-over-year growth of 37%, although this was a slowdown from 73% growth in Q3 [21] Market and Competitive Landscape - AMD's Q1 revenue guidance is set between $9.5 billion and $10.1 billion, with a midpoint of $9.8 billion, slightly above consensus expectations [3][24] - CEO Lisa Su indicated strong demand for server CPUs and projected a 60% annual growth in data center revenue over the next three to five years [3][16] Challenges and Concerns - The market is concerned about AMD's ability to maintain growth momentum, particularly in the AI accelerator segment, as any conservative guidance may lead to negative market reactions [4][24] - AMD's sales of the Instinct MI308 chip in China amounted to approximately $390 million, with U.S. export restrictions expected to impact inventory and related costs by about $440 million in 2025 [23]

2月3日，2026年支付宝集福正式启动。今年活动的最大亮点，是首次推出了"健康福"，由蚂蚁集团旗下健康AI"蚂蚁阿 福"发放，用户领取后可获得福气红包。 阿福App今天宣布全新升级 硬·AI 作者 | 硬 AI 编辑 | 硬 AI 因"健康福"寓意吉祥，该活动上线后迅速引发了全网热议和领取热潮。市场数据的反馈则更为直接：这一活动带动"蚂蚁 阿福"App下载量激增，使其迅速冲上苹果商店下载榜第三位。 | | 排行榜 | 所有 Ap | | --- | --- | --- | | | 免费 App 付费 App | | | 1 | 元宝-腾讯旗下 AI助手 DeepSeek+拍题P图搜索工具, 写作学习办公对话问答助手 | 35 EV | | 2 | 抖音精选-看视频 精选作者贺新春 | SACEM App 内购买 | | 3 | 蚂蚁阿福-蚂蚁旗下 Al 健康 助手 健康是福,健康的事就找阿福 | 打开 | | 4 | 豆包 - 字节跳动旗下 AI 助手 AI 全智能聊天对话问答助手,搜索 写作翻译编程全能工具 | 打开 | 根据苹果应用商店的数据显示，蚂蚁阿福此次的排名已经超过了字节跳动旗下的通用大模型应用" ...

以下文章来源于华尔街见闻 ，作者鲍奕龙 华尔街见闻 . 华尔街见闻App是中国领先的金融信息和商业资讯提供商，为用户甄选国内和全球重要资讯，7*24小时全年不间断。 Claude在专业领域的市场地位 在竞争激烈的AI市场中，Claude已成为众多法律和金融专业人士进行分析工作的首选工具。 Anthropic在其插件网站上提供的法律工具，声称可以自动完成合同审查和法律简报等工作。该网站还指出："所有输出结果均应由持证律师审核。" 人工智能领域的一次常规更新引发了法律软件板块的剧烈震荡。 Anthropic上周五为其Claude Cowork AI代理推出的法律插件，在周二触发了相关股票的集体暴跌，多只个股跌幅超过15%。 这家Claude AI聊天机器人的开发商推出的 新工具能够执行多项文书工作，包括追踪合规事项和审阅法律文档。这些功能正是许多法律软件产品的核心所在。 摩根士丹利分析师 Toni Kaplan等人 在一份关于汤森路透的报告中写道： Anthropic为其Claude Cowork AI代理在法律领域推出了新功能，加剧了该领域的竞争。我们认为这是竞争加剧的迹象，因此可能带来负面影响。 周二第一波抛 ...

作者 | 鲍奕龙 编辑 | 硬 AI 英特尔公司首席执行官陈立武警告称，计算机行业的内存芯片短缺问题至少还将持续两年。 2月3日周二， 陈立武在思科系统公司会议上表示，他与两家主要内存厂商进行了沟通，对方明确告知"至 少到2028年之前不会缓解"。 人工智能基础设施的大规模扩张持续推高内存芯片需求，挤压了传统设备可 用的供应量。 陈立武在思科系统公司会议上表示，他与两家主要内存厂商进行了沟通，对方明确告知"至少到2028年之前不会缓解"。 他表示，人工智能将"吸收大量内存"。 硬·AI 陈立武指出，英伟达公司作为人工智能处理器的领先供应商，其最新的Rubin平台及下一代产品将进一步 推高内存需求。 他表示，人工智能将"吸收大量内存"。 与此同时，陈立武透露英特尔计划进军GPU市场，并已聘请首席GPU架构师，该业务将与公司的数据中 心芯片部门及代工服务紧密结合。周二英特尔股价高开低走，一度转跌。 （周二英特尔股价日内走势） 01 英特尔布局GPU市场 陈立武在会议上宣布，英特尔计划制造由英伟达推广普及的图形处理器(GPU)。 陈立武称这次招聘费了一番功夫，他透露： 我刚聘请了首席GPU架构师，他非常优秀。我很高 ...