2025年1月，AI初创企业Anthropic正筹备推出Claude 4聊天机器人，意欲与ChatGPT、Gemini等产品展开竞争。Anthropic同时还透露，正以超600亿美元的 估值进行融资。这笔交易让这家仅成立四年的企业一跃成为全球估值最高的人工智能企业之一，公司的7位联合创始人也尽数跻身亿万富豪行列。 Anthropic由此进入了美国公司的另一个梯队：亿万富豪制造工厂。 对大多数大型企业而言，如果能有一位创始人成为亿万富豪，已是万幸。部分成功企业还能让联合创始人或是早期投资者迈入亿万富豪俱乐部，少数企业 甚至能让一两位早期员工实现巨额财富积累。而仅有寥寥几家幸运的顶尖巨头，能培育出众多亿万富豪。在美国680位白手起家的亿万富豪中，有54位的 财富积累主要得益于8家企业。 | | 亿万富豪 | 亿万富豪财富总额 | | --- | --- | --- | | 公司 | 人数 | (亿美元) | | 谷歌/Alphabet | 10 | 6182 | | Facebook/Meta | 8 | 3140 | | AppLovin | 8 | 766 | | Anthropic | 7 | 259 | ...

英伟达CEO黄仁勋 CES 2026 演讲，我们整理了 10 个非常具有远见的短句，未来是 agent 和物理 bot 的世界。 1. 关于行业大洗牌 "The entire computer industry, the entire five stack of the computer industry is being reinvented." "整个计算机行业，从底层的每一层 堆栈，正在被彻底地重新发明。" 解读： 以前的经验正在失效。如果你还在守着旧的 IT 架构（CPU 时代）不放，你所在的行业地基都已经变 了。这不是升级，这是推倒重来。 2. 关于职业生涯的终结与新生 "You no longer program the software, you train the software." "你不再编写软件，你是在训练软件。" 解读： 程序员的 定义被改写了。未来的"编程"不是写代码，而是通过数据和反馈来"调教"智能体。如果你不会训练 AI，你将失去对软件的控制权。 7. 关于软件交互的革命 "The enterprise user interface will no longer be Exce ...

Core Viewpoint - Microsoft is expected to deliver solid but unspectacular growth in 2026, ranking as the No. 4 stock in the "Magnificent Seven" for that year [1][4]. Company Overview - Microsoft is a computing powerhouse, offering a wide range of products including personal computers, operating systems, tablets, gaming consoles, and services like LinkedIn, Edge, and Bing [2]. - The company has made significant investments in artificial intelligence, particularly through its partnership with OpenAI, enhancing productivity and automation for users [2][6]. Financial Performance - Microsoft has experienced substantial growth over the last decade, with revenue increasing by over 230% and earnings per share and net income rising by more than 500% [8]. - For the first quarter of fiscal 2026, Microsoft reported revenue of $77.7 billion, an 18% increase year-over-year, with net income of $27.7 billion, up 12%, and earnings per share of $3.72, up 13% [10]. Segment Performance - The company operates in three primary segments: - **Productivity and Business Processes**: Revenue of $33.02 billion, up 16.6% year-over-year, with operating income of $20.41 billion, up 23.5% [11]. - **Intelligent Cloud**: Revenue of $30.89 billion, up 28.2%, with operating income of $13.39 billion, up 27.5% [11]. - **More Personal Computing**: Revenue of $13.75 billion, up 4.4%, with operating income of $4.16 billion, up 17.8% [11]. - The Intelligent Cloud segment is growing rapidly and is expected to surpass the productivity software tools as the most lucrative segment if the growth trajectory continues [11]. Investment Perspective - Microsoft is viewed as a solid and reliable investment option for 2026, with a stronger growth engine than Apple and a more effective business model than Amazon, while being less volatile than Tesla [12][13]. - The company is positioned in the middle of the pack among the "Magnificent Seven," with dynamic growth opportunities seen in competitors like Nvidia, Alphabet, and Meta Platforms [13].

知名科技博主犀利指出，各大实验室通过RL（强化学习），耗资数十亿美元让大模型"排练"Excel、网页操作等技能，恰 恰暴露其距真正AGI仍远。若AI真接近类人智能，就该像人类一样从经验中自主学习，而非依赖"可验证奖励训练"。而真 正突破在于"持续学习"能力，这一过程或需5-10年才能完善。 硬·AI 作者 | 龙 玥 编辑 | 硬 AI 在人工智能迈向通用智能（AGI）的竞赛中，当前最受追捧的强化学习（RL）路径，可能正将我们引向一 条歧路——RL越强，距离真正的AGI或许越远。 12月24日，知名科技博主、Dwarkesh Podcast博客主持人Dwarkesh Patel今日发布了一则引发行业深思 的视频，直击当前大模型发展的痛点。在硅谷普遍对Scaling Law（缩放定律）和RL（强化学习）持极度 乐观态度的背景下，Patel提出了一个反直觉的犀利观点： 对RL（强化学习）的过度依赖和投入，可能非 但不是通往AGI的捷径，反而是其远未到来的明确信号。 Patel的核心论点在于，当前顶尖AI实验室正耗费巨资，通过基于可验证结果的强化学习，为大模型"预 制"大量特定技能，例如操作Excel或浏览网页。然 ...

自今年 9 月将部分职责移交以来，微软首席执行官已成为公司最具影响力的产品经理，他尤其专注于让公司的 Copilot 人工智能助手对客户来说变得更加不 可或缺。 几周前 ，微软首席执行官萨提亚 ·纳德拉向公司负责开发消费者版 Copilot 的工程负责人发送了一封邮件。 Copilot 是嵌入微软多款产品中的 AI 助手。邮件 讨论中提到，一位微软经理指出，谷歌的 Gemini 聊天机器人近期在连接谷歌云端硬盘方面表现有所提升，例如能够总结文件夹中存储的照片内容。 与此同时，纳德拉对微软该项技术的实际运行效果并不满意。根据《信息》查阅的邮件内容， 他指出微软用于连接 Copilot 与 Gmail 、 Outlook 的程 序"大部分情况下根本没法用"，且"不够智能"。 过去几个月里，纳德拉已转型为微软最具影响力的产品管理者。去年 9 月，他向微软员工表示计划将部分职责下放，以便更专注于微软人工智能产品的开 发，以及监管数据中心建设和提升 AI 模型智能水平等其他雄心勃勃的技术工作。 作为转型的一部分，他将许多业务职能移交给了微软销售主管贾德森 ·阿尔托夫，后者获得了新任命的商业首席执行官头衔。例如，在 11 ...

Group 1 - The core argument of the article is that software is transitioning from being an "auxiliary tool" to an "executive entity," marking a paradigm shift in its commercial attributes [4][7][12] - In the past, software was strictly defined as a tool dependent on human operation, with its value released only through human input [4][5] - The emergence of AI has transformed software into a digital workforce capable of independent task execution, thus changing how businesses evaluate software value [7][8][11] Group 2 - The traditional pricing model based on per-user subscriptions is becoming obsolete, necessitating a fundamental adjustment in monetization strategies for entrepreneurs [12][13] - The proposed "Goldilocks Zone" pricing strategy aims to find an optimal arbitrage space between software costs and human labor costs, ensuring pricing is significantly lower than hiring real employees while still being higher than traditional software subscription fees [15][16][17] - Entrepreneurs are advised to leverage the "Gold Brick Theory" to identify structural gaps that giants strategically overlook, shifting the focus from homogeneous model capabilities to deep understanding of specific industry contexts [18]

Excel几乎是所有使用电脑的人都会接触的一款办公软件，一般会用于做数字表格。令人意外的是，这个办公软件居然有官方举办的电竞比赛—— Excel世界锦标赛（Microsoft Excel World Championship，以下简称Excel世锦赛）。 张云翼。受访者供图 Excel世锦赛不比做表格 张云翼是一位小有名气的清华大学毕业生：曾是辽宁省丹东市中考第一名；2011年考入清华大学，一路从本科读到博士毕业；在清华大学土木水利 学院官网上，还有他在导师带领下参与福建省泉州市欣佳酒店"3•7"坍塌事故调查的文章；2022年，他博士后出站，目前从事施工管理软件的研发 工作。 据多家媒体报道，近日，第五届Excel世锦赛全球总决赛在美国拉斯维加斯落下帷幕，毕业于清华大学土木工程系的博士张云翼，一路从预赛杀入全 球总决赛，还在每5分钟淘汰一人的残酷赛制下位列第九名。 Excel世锦赛到底比什么？参赛需要什么能力？带着这些问题，12月10日，上游新闻记者采访了刚从美国回国的张云翼。 随后张云翼参加了线上淘汰赛，这是一种1V1的激烈对决，256名选手通过抽签捉对厮杀，结果他发挥出色，成功杀入最后的八强，直接"保送" ...

2025 年 12 月 13 日，a16z最新一期播客《AI Eats the World》上线。 科技趋势分析师 Benedict Evans（前a16z合伙人）与 a16z 合伙人 Erik Torenberg，开场就问：AI 到底是 又一轮平台迁移，还是像电力和互联网那样，成为重写世界的底层通用技术？ 听起来很宏大。但真正的矛盾在另一边：巨头们抢算力、调团队、建超级工厂，普通人却连每天拿 AI 做点什么都想不清楚。 这不是懒或不懂技术。 Evans 认为：不是 AI 不够强，而是它还没变成日常工具。就像电梯，你得主动走进去，按个键，它才 带你上楼。 这场对话要回答三个递进的问题： 为什么 9 亿人注册，却想不出要做什么？ 为什么能力很强，却没有变成日常工具？ 谁会赢得下一个入口？ 最后，Erik 问出那句很多人都在回避的问题：要发生什么，AI 才能被真正认为比互联网更具变革性？ 第一节｜9亿用户，95%的人想不出要做什么 打开 ChatGPT，不等于用上 AI。 一个被忽视的事实：AI 工具使用率远落后于注册量。 ChatGPT 已经拥有 8–9 亿周活用户，是全球访问量最高的 AI 工具。但真实数据更 ...

Event & Competition - Microsoft participated in the Excel World Championship in Las Vegas [1] - Satya Nadella participated in the Microsoft 365 Copilot Challenge, using Agent Mode [1] Product & Technology - Showcasing Microsoft 365 Copilot and Agent Mode [1] Social Media & Community - Encouraging viewers to subscribe to Microsoft's YouTube channel [1] - Promoting Microsoft's presence on LinkedIn, Twitter, Facebook, and Instagram [1]

Core Insights - The tech sector is experiencing heightened investor anxiety due to skyrocketing AI valuations and record data center expenditures, with potential repercussions if the $10 trillion AI boom loses momentum [2] Company Analysis Google: The Utility - Google operates on recurring human behavior, maintaining relevance in both economic booms and downturns, with a diversified revenue stream and high switching costs [5][8] - As one of Nvidia's largest clients, Google could impact Nvidia's revenue if it reallocates workloads to its own TPUs, which are reportedly improving [8] Microsoft: Workflow Subscription - Microsoft provides essential infrastructure for daily operations, ensuring stable revenue through its subscription model, which includes tools like Outlook, Excel, and Azure [6][8] - The integration of AI functionalities across its services enhances average revenue per user (ARPU) and promotes cloud consumption [8] Apple: Consumer Ecosystem - Apple relies on consumer loyalty and device cycles rather than heavy investments in data centers, with a strong position in digital services exceeding $100 billion [7][9] - The company could benefit from AI advancements due to its vast user base of over 2 billion [9] Oracle: Legacy Lock-In - Oracle's established presence in sectors like banking and government creates a legacy lock-in, providing a reliable revenue base despite lacking glamour [7][9] - Significant investments in AI, with a capital expenditure of $35 billion in FY'26, are backed by substantial take-or-pay contracts, ensuring high-visibility revenue [9] Risk Exposure Exposed Models - Companies heavily reliant on capital cycles and non-recurring orders are most vulnerable to downturns, experiencing significant gains during booms but suffering first when conditions soften [10] - Nvidia, with 60% revenue growth and a concentration of sales from a few customers, faces risks if hyperscalers reduce GPU purchases [14] Valuation-Dependent Growth - Companies with diversified demand, steady revenue streams, and strong customer lock-in are better positioned to withstand market corrections [12] - Established software firms like Palantir may face steep re-evaluations if IT budgets tighten, lacking the long-term customer lock-in that Oracle enjoys [14]