Claude Code、ChatGPT Agent、Manus，再加上一堆大厂的企业级工作流平台，动不动就喊"替代打工人""创造万亿美 元价值"。麦肯锡的报告也在推波助澜：62% 的企业正在试水 Agent。 但把热闹先放一边，真正的问题是——在真实工作场景里，Agent 到底走到哪一步了？ 最近，MIT、哈佛、斯坦福等机构的研究团队发布了一份《2025 AI Agent Index》。 这帮学者干了一件很有价值的事： 进入2026 年，Agent 彻底火了。 他们把市面上最具代表性的30 个代理系统逐一拆开来看了一遍，并设计了45 个维度，仔细扒了扒这些产品的技术细 节、部署情况、设计架构、工具使用和安全机制等信息，得出了当下Agent最硬核的3个真相。 透过这部分报告，我们就能更清晰地了解当下Agent发展的真实情况。 01 产品形态收拢在这3个方向 大多数代理产品集中在2024–2025 年发布。 （蓝色柱状图代表与代理型人工智能产品相关的谷歌新增搜索词，红线则代表谷歌学术中包含"人工智能代理"或"代理 型人工智能"关键词的论文发表数量） 从产品形态看，AI代理基本收拢在了三个方向： 聊天式代理（12个 ...

Artificial intelligence (AI) infrastructure spending is poised to jump significantly once again in 2026. According to market research provider Futurum, the five largest hyperscalers and cloud computing companies in the U.S. could spend between $660 billion and $690 billion on AI infrastructure this year, well above last year's $380 billion. Additionally, the firm notes that pure-play AI companies such as OpenAI, Anthropic, and Perplexity, as well as ambitious infrastructure projects such as Stargate, wil ...

Industry Overview - Artificial intelligence (AI) infrastructure spending is expected to increase significantly, with the five largest hyperscalers and cloud computing companies in the U.S. projected to spend between $660 billion and $690 billion in 2026, up from $380 billion last year [1][2] - The demand for AI infrastructure is driven by both established companies and new entrants in the AI space, indicating a robust growth trajectory for the sector [2] Company Analysis: Micron Technology - Micron Technology's dynamic random-access memory (DRAM) chips, particularly high-bandwidth memory (HBM), are essential for AI data centers, enabling faster data processing for AI models [5][8] - The company has noted that AI servers require six times the amount of DRAM compared to standard servers, leading to a significant increase in demand and a shortage of memory chips [7][9] - Analysts predict a substantial increase in Micron's earnings, with expectations of more than a fivefold increase in earnings per share over two fiscal years, making it an attractive investment opportunity [11] - Micron's current trading multiples are favorable compared to the tech-focused Nasdaq-100 index, suggesting potential for price appreciation [12] Company Analysis: Jabil - Jabil is positioning itself as a key player in AI infrastructure, providing manufacturing and engineering services for AI server racks and related technologies [13][15] - The company plans to invest $500 million to expand its AI data center infrastructure manufacturing capacity, reflecting its commitment to capturing the growing demand in this sector [15] - Jabil is forecasting a 35% increase in AI revenue for fiscal 2026, indicating strong growth potential, and has raised its operating margin guidance due to improved contributions from its AI business [16] - The stock is currently trading at a discount compared to the Nasdaq-100's earnings multiple, suggesting it may still have significant upside potential [18][19]

Core Insights - Anthropic is emerging as a leading AI disruptor, experiencing significant revenue growth and attracting substantial capital investment [1][2] - Amazon serves as the primary cloud computing partner for Anthropic, providing investors a way to gain exposure to Anthropic's growth potential [2][8] Group 1: Anthropic's Growth - Anthropic raised $30 billion in private funding and plans to go public within the year [2] - The company's annualized revenue surged from $1 billion in January 2025 to $14 billion by January 2026, marking a year-over-year growth of over 1000% [3][4] - Tools like Claude Code are enabling software development teams to significantly reduce product development time [4] Group 2: Amazon's Role - Amazon is enhancing its cloud computing capabilities to support Anthropic, including building a custom data center and collaborating on AI model training [5] - Amazon Web Services (AWS) reported a 24% year-over-year revenue growth, reaching $129 billion in sales for 2025, likely influenced by its partnership with Anthropic [6][8] - Projections suggest that if AWS continues to grow at 24% annually, it could reach $250 billion in revenue in five years, generating $86 billion in earnings from AWS alone [9]

大模型市场，直接来了一个「新物种大分叉」。 全球最大AI模型API聚合平台OpenRouter给出数据： 9号到15号这一周，中国大模型的调用量冲到了 4.12万亿Token，第一次超过美国模型的2.94万亿Token。 调用量前五的模型里，中国直接占了四个：MiniMax的M2.5、月之暗面的KimiK2.5、智谱GLM-5、 DeepSeek V3.2。 市场已经彻底分成两拨人：BAT正在「向内收缩」，把模型塞进搜索、电商、办公这些老业务里；新势 力Kimi、智谱、MiniMax则「向外长手」，靠成百上千个Agent死磕开发者生态。 这是商业底层逻辑的彻底分家，大家都在算一笔名为投资回报率的生死账。 01 先得搞清楚一件事：这4.12万亿Token的调用量，到底是谁在用？ 要光看新闻标题，肯定以为中国老百姓突然爱上跟AI唠嗑了；OpenRouter的用户构成很有意 思，上面美国开发者占了47.17%，中国开发者只有6.01%。 这4.12万亿Token，主要是全球那帮程序员「用脚投票」投出来的，跟咱们在抖音刷到的那些AI聊天没 啥关系。 所以，Token调用量比DAU实在多了，日活用户能被红包砸出来，能被 ...

英伟达CEO黄仁勋在2月25日财报电话会上确认，公司已接近与OpenAI达成合作协议，但澄清此前传闻 的1000亿美元投资"从来不是一项承诺"。该协议涉及英伟达向OpenAI提供先进AI芯片及技术支持，以 支撑其大模型训练与推理需求，同时英伟达或将参与OpenAI新一轮融资。黄仁勋形容OpenAI为"一代人 中仅此一家"的企业，强调双方自2016年合作至今，OpenAI所有工作均基于英伟达技术平台。 值得关注的是，OpenAI近期也与AMD达成战略合作，计划部署6吉瓦AMD Instinct GPU，显示其在芯片 供应上正寻求多元化。分析认为，英伟达与OpenAI的协议若最终敲定，将进一步巩固前者在AI算力领 域的霸权地位，同时助力OpenAI推进"星际之门"等大规模AI基础设施建设。 此次合作标志着OpenAI采购模式的重要转变——从通过微软等云服务商间接采购，转向与英伟达直接 合作。与此同时，英伟达正加速AI生态布局，本季度已宣布向Anthropic投资最高100亿美元，并与Meta 签署多年期、价值数十亿美元的芯片供应协议，后者将部署数百万颗Blackwell和Rubin GPU。黄仁勋表 示，公司投资 ...

图源：网络 需要说明的是，"蒸馏"是全球部分AI公司训练自家大模型的常用手段，而Anthropic自己，恰恰是"蒸馏"技术的使用者，甚至存在更激进的数据抓取行为。 值得一提的是，Anthropic如今却成了人工智能领域对中国最不友好的公司之一。2025年9月，Anthropic在官方文件中明确封禁对中资企业的服务。其实指 责中国企业进行数据蒸馏，已经成了美国企业惯用套路。就在2026年2月12日，OpenAI向美国国会提交内部备忘录，明确指控DeepSeek通过复杂的混淆手 段，绕过其安全防护对GPT系列模型实施蒸馏行为。 Anthropic这次直接对中国企业"高调指控"，一个重大背景是进入2026年2月，美国AI概念股出现多轮明显下跌，资本市场对人工智能的未来产生了"颠覆传 统商业模式"的担忧。CNN评论也指出，不少所谓大模型，其实只是精心包装过的搜索引擎。 马斯克嘲讽、资本变脸，中美AI大战升级，中国凭什么破局？ 开工第一天，AI圈就上演"大戏"！美国AI独角兽Anthropic突然发难，指控DeepSeek、月之暗面（Kimi）等3家中国AI企业，靠"蒸馏攻击"抄袭技术。 "蒸馏攻击"是什么？先按下不表 ...

2026 年 2 月 27 日 陈伟光 S0340520060001 电话：0769-22119430 邮箱： chenweiguang@dgzq.com.cn SAC 执业证书编号： S0340521020001 电话：0769-22110619 邮箱： luoweibin@dgzq.com.cn S0340524070002 电话：0769-22119302 邮箱： chenzhanqian@dgzq.com.cn 超配（维持） 散热需求向互连系统延伸，连接器散热成为重要补充 深 度 研 AI 集群互连散热专题报告 证 券 研 究 报 告 资料来源：iFind，东莞证券研究所 相关报告 投资要点： 本报告的风险等级为中高风险。 本报告的信息均来自已公开信息，关于信息的准确性与完整性，建议投资者谨慎判断，据此入市，风险自担。 请务必阅读末页声明。 通信行业 SAC 执业证书编号： 罗炜斌 AI集群功耗上扬，集群散热需求增长。AI算力需求呈指数级爆发直接推 动了AI集群功耗上扬，从单芯片到机柜级别的功耗密度的激增已经超越 了传统数据中心的设计极限。以英伟达产品为例，2026年是其AI硬件产 品将从H100/H ...

AI 编程行业深度报告 2026 年 2 月 27 日 卢芷心 S0340524100001 电话：0769-22119297 邮箱： luzhixin@dgzq.com.cn S0340521020001 电话：0769-22110619 邮箱： luoweibin@dgzq.com.cn S0340520060001 电话：0769-22119430 邮箱： chenweiguang@dgzq.com.cn 超配（维持） 海外价值获验证，国内市场开启高增长周期 深 度 资料来源：iFind，东莞证券研究所 投资要点： 本报告的风险等级为中高风险。 本报告的信息均来自已公开信息，关于信息的准确性与完整性，建议投资者谨慎判断，据此入市，风险自担。 请务必阅读末页声明。 SAC 执业证书编号： 罗炜斌 SAC 执业证书编号： AI Coding向"自主型Agent"跃迁，未来市场空间广阔。AI Coding已 成为众多AI应用领域中落地速度最快、发展潜力最大的细分赛道之一， 其产品或功能演进高度依赖底层大模型能力的发展。当前，国际前沿编 程大模型发展由海外厂商引领，而国产大模型正展现出强劲的追赶态 势。在发展趋 ...

Core Viewpoint - The intense negotiation between the Pentagon and AI startup Anthropic regarding military technology boundaries is causing significant reactions in Silicon Valley, with employees from major tech companies voicing their concerns about ethical implications and potential misuse of AI technology [1][2]. Group 1: Employee Reactions - Over 100 Google AI employees submitted a letter to management demanding clear boundaries in collaboration with the military, specifically opposing the use of their technology for mass surveillance or autonomous weapon systems [1]. - Nearly 50 OpenAI employees and 175 Google employees also published an open letter criticizing the Pentagon's strategy to divide tech companies and urging them to unite against unethical practices [1]. Group 2: Pentagon Pressure and Responses - The Pentagon has exerted significant pressure on Anthropic to allow the military to use its Claude model for "all legitimate purposes," which Anthropic's CEO Dario Amodei has firmly rejected, citing moral objections [2]. - Google employees expressed their desire to prevent any transactions that would cross ethical boundaries, indicating a strong internal push against military collaborations [2]. Group 3: Google Executives' Stance - Jeff Dean, a prominent Google engineer, publicly supported Anthropic's position, emphasizing that mass surveillance violates constitutional rights and can lead to misuse for political or discriminatory purposes [3]. - Google has a complex history with employee activism, having previously faced protests that led to the cancellation of military contracts, highlighting ongoing internal ethical scrutiny [3]. Group 4: Military Strategy and AI Risks - In response to Anthropic's stance, the Pentagon is seeking alternative solutions, having already reached an agreement with xAI to use its Grok model for military applications [4]. - The Pentagon's negotiations with Google are ongoing, and there are threats to invoke the Defense Production Act to compel Anthropic's compliance, indicating a high-stakes environment for AI in military applications [4]. - Concerns about the potential risks of AI in military contexts are underscored by simulations showing that top AI models could opt for nuclear weapon use under pressure, raising alarms about the implications of AI in decision-making [5][6].