Claude's AI agent crushes Wall Street, wiping out nearly a trillion dollars in market value.

Before AI replaces humans, will it first replace a large wave of software? Last night, Wall Street was gripped by panic. Anthropic's new generation of artificial intelligence tool, "Claude Cowork," and its 11 accompanying functional plugins officially went live. The market viewed this as a watershed moment, marking AI's transition from "copilot" to "independent agent," directly shaking the commercial foundation of SaaS software. From Oracle, Adobe, and Salesforce to Thomson Reuters and NEC, a series of well-known companies saw their stocks sold off. As an AI tool for the era of intelligent agents, Claude Cowork far surpasses the scope of large-scale chatbots. Cowork can work directly on your computer, possessing sufficiently high privileges to take over your mouse, keyboard, and file system, autonomously planning and completing a series of complex tasks according to your vague instructions. On the other hand, unlike the recently popular open-source project ClawdBot, Claude Cowork runs in an isolated virtual machine (VM) environment to prevent AI from accidentally deleting files or performing unexpected operations via the network. This is a more serious intelligent agent, designed from the outset to perform tasks. Anthropic has prepared a series of "professional skill packs" for Claude Cowork, stating that these tools can generate financial statements, research sales leads, prepare sales calls, draft customer support responses, evaluate confidentiality agreements, and review contracts, legal briefs, and build financial models. At first glance, these claims seem similar to those made by AI companies when releasing large models and intelligent agents in the past. When Claude Cowork's "research preview" was first unveiled on January 12, everyone thought it was just a more powerful automation tool. However, by January 31, when Anthropic released 11 official open-source plugins, some realized things weren't so simple. Anthropic has released a series of plugins (https://github.com/anthropics/knowledge-work-plugins) for its "desktop-level all-around digital employee" Claude Cowork agent platform. The company claims these plugins can automate dozens of tasks, including customer service, product management, marketing, legal, and data analysis. Is this similar to what they previously advertised? However, this time, concerns are growing that the software services (SaaS) business model underpinning the technology and digital services industry is facing disruptive risks. Wall Street worries that AI tools like the Claude plugin will challenge existing software companies' data analytics and research offerings—companies leveraging AI for automation and building their own tools may reduce subscriptions to external research and data services, directly harming software companies' profits. Following a broad sell-off on Tuesday, the S&P 500 Software and Services sector fell nearly 4%, marking its sixth consecutive day of decline and wiping out approximately $830 billion in market capitalization since January 28. Whether AI tools will ultimately destroy the software industry remains to be seen. But on Tuesday, investor panic was real, with people selling off stocks of legal and financial software and services companies. After all, if AI can reduce the time required for in-house development, why pay for software at all? It seems that the revolution of AI across numerous industries is becoming tangible. In January of this year, Anthropic CEO Dario Amodei wrote an article arguing that "artificial intelligence could replace half of entry-level white-collar jobs within the next 1-5 years." How confident is Anthropic in this? Yesterday, they emphasized that they wouldn't add ads to their AI, promoted it during the Super Bowl, and even took a jab at OpenAI. OpenAI founder and CEO Sam Altman had to respond. Such fierce competition might soon give rise to even more powerful AI capabilities?