在 February 11, 2026，文章描述了一次由 AI 驱动的急剧市场重置：在 Anthropic 于上周发布新的 Claude 工具后，投资者迅速抛售软体、法律资料、媒体、行销与商业资讯股票，形成广泛的「先卖再说」行情。其核心定价逻辑是，AI 可能以更低成本或免费替代方案取代付费工作流程，特别是在无程式码法律审查外挂让颠覆看起来是立即发生而非理论风险之后。结果是既有商业模式被快速重估，而投资者难以区分拥有耐久资料资产的公司与经济结构脆弱的公司。
证据重点在于情绪扩散有多快，以及损害持续了多久：Anthropic 先推出聚焦法律的外挂，接著又很快推出金融产品，扩大了多个垂直领域的风险感知。即使在 1 week 之后，许多跌势仍未修复，尽管分析师对部分标的仍维持建设性看法，这显示在主动管理人较少的市场中，资金流主导的过度反应。Relx 被视为关键案例：其估值倍数在 2021-2025 期间因 AI 上行预期而扩张，但在过去一年股价大约下跌 50% 后，该重评被大幅逆转，显示当盈余风险上升时，成长溢价假设可以多快反转。
文章主张，更有吸引力的一段可能是 AI 采用者而非 AI 赋能者：如果供应商端估值被下修，客户端应可透过更低营运成本和/或更高营收取得对应收益。UBS 强调具可衡量落地潜力的产业，包括金融服务、零售、医疗保健与运输，并举出 BNP Paribas、Tesco、Siemens Healthineers 与 DSV 等例子。具体基准是 BNP 在 2026 的「value-from-AI」目标 €750 million ($893 million)，同时也提醒投资者以硬性营运指标验证主张，例如每位员工销售额上升与毛利率扩张；主要不确定性在于时点，也就是生产力提升是否会比经济学家目前预测更快到来。
On February 11, 2026, the article describes a sharp AI-driven market reset: after Anthropic released new Claude tools last week, investors rapidly sold software, legal-data, media, marketing, and business-information stocks in a broad “sell first, ask questions later” move. The core pricing logic is that AI may replace paid workflows with lower-cost or free alternatives, especially after no-code legal-review add-ons made disruption feel immediate rather than theoretical. The result is a fast repricing of incumbent business models, while investors struggle to separate firms with durable data assets from firms with fragile economics.
The evidence centers on how quickly sentiment spread and how long damage persisted: Anthropic first launched legal-focused add-ons and then quickly introduced a finance product, widening perceived risk across multiple verticals. Even 1 week later, many declines remained despite analysts staying constructive on some names, suggesting flow-driven overreaction in markets with fewer active managers. Relx is presented as a key case: its valuation multiple expanded during 2021-2025 on AI-upside expectations, but that rerated heavily as the share price fell by roughly 50% over the past year, showing how quickly growth-premium assumptions can reverse when earnings risk rises.
The article argues the more attractive leg may be AI adopters rather than AI enablers: if vendor-side valuations are marked down, customers should capture offsetting gains through lower operating costs and/or higher revenue. UBS highlights sectors with measurable implementation potential, including financial services, retail, health care, and transport, and names examples such as BNP Paribas, Tesco, Siemens Healthineers, and DSV. A concrete benchmark is BNP’s 2026 “value-from-AI” target of €750 million ($893 million), while investors are urged to verify claims via hard operating metrics such as rising sales per employee and expanding gross margins; the main uncertainty is timing, specifically whether productivity gains arrive faster than economists currently project.