We witnessed history this week. Not the impressive-demo kind of history. The $300-billion-evaporating-in-24-hours kind.
On February 3, 2026, Anthropic released a set of plugins for Claude Cowork. Within hours, software stocks worldwide entered freefall. Thomson Reuters dropped 16%—its worst day ever. RELX (LexisNexis) plunged 14%. ServiceNow, Salesforce, Intuit—all cratered.
By the time the dust settled, approximately $300 billion in market value had vanished from American software, finance, and data stocks.
This isn’t just market news. It’s a signal that the AI disruption everyone’s been talking about has finally arrived at Wall Street’s doorstep.
TL;DR — The Quick Take
Anthropic released AI plugins that automate legal contracts, sales, marketing, and data analysis. Markets panicked because foundation AI companies are now competing directly with enterprise software vendors—not just supplying them. Jensen Huang calls this “illogical.” The truth is somewhere in between. Either way, the playbook has changed.
What Actually Happened?
On January 30, Anthropic quietly released 11 open-source plugins for Claude Cowork on GitHub. These plugins transform Claude from a general AI assistant into specialized enterprise software that can:
- Review contracts and flag risks against your company playbook
- Triage NDAs automatically by approval category
- Track compliance and generate legal briefings
- Automate sales with CRM integration and prospect research
- Handle marketing campaigns and content drafts
- Analyze data that used to require expensive proprietary databases
The legal plugin got the most attention. Here’s why: it does what companies pay Thomson Reuters and LexisNexis millions of dollars annually to provide. And Anthropic open-sourced it on GitHub. For free.
The Market Carnage: By the Numbers
The selloff was brutal and global:
| Company | Drop | Context |
|---|---|---|
| Thomson Reuters | -16% | Worst single-day drop in company history |
| RELX (LexisNexis) | -14% | Biggest decline since 1988 |
| Wolters Kluwer | -13% | Legal analytics leader crushed |
| ServiceNow | -7% | YTD losses now 28% |
| Salesforce | -7% | 2026 decline approaches 26% |
| Intuit | -11% | TurboTax parent battered |
| London Stock Exchange Group | -8.5% | Data services unit targeted |
Morgan Stanley analysts summed up investor sentiment: “Most of the investors we have spoken with recently are overwhelmingly bearish on [Thomson Reuters] as the consensus opinion worries that the company will be unable to maintain the same level of growth within its legal segment given increased competition from specialized AI tools.”
Software’s forward price-to-earnings ratio compressed from 33.1x to 23.2x—a 30% contraction. That’s not a correction. That’s a repricing of an entire sector.
Why This Time Is Different
We’ve had AI hype cycles before. This selloff is different for three reasons:
1. Foundation Models Moved Up the Stack
Until now, companies like Anthropic and OpenAI sold APIs. They were infrastructure. The unspoken agreement was: “We provide the brains, you build the products.”
That agreement is dead.
By shipping complete workflows—contract review, compliance tracking, sales automation—Anthropic declared war on the software vendors who assumed they’d be the ones adding value on top of AI. (For context on which Claude model powers what, see our Claude Opus vs Sonnet comparison.) As one legal tech analyst noted: Anthropic is shifting “from model supplier to application layer and workflow owner.”
2. It’s Open Source
Anthropic didn’t just release enterprise software. They open-sourced it. The plugins are on GitHub. Any company can take them, customize them, and deploy them.
This isn’t a proprietary moat play. It’s a platform strategy. Anthropic wants enterprises building on Claude, not bolting Claude onto legacy software.
3. Wall Street Finally Believed It
For years, analysts have written about AI disruption as a future risk. “In 5-10 years, AI might…” That comfortable distance collapsed on Tuesday.
When Thomson Reuters loses 16% of its value in a single session, investors aren’t pricing in hypothetical future disruption. They’re pricing in disruption that’s happening now.
The Counter-Argument: Jensen Huang Says Relax
Not everyone thinks the sky is falling.
Hours after the selloff, Nvidia CEO Jensen Huang weighed in at an AI event in Japan: “There’s this notion that the tool in the software industry is in decline, and will be replaced by AI… It is the most illogical thing in the world, and time will prove itself.”
Huang’s argument: AI doesn’t replace software—it needs software. More AI means more infrastructure, more tools, more integration layers. The software industry will grow, not shrink.
He has a point. AI agents still need CRMs to update, databases to query, and applications to control. Someone has to build and maintain all that plumbing.
But here’s the nuance Huang glosses over: the kind of software that thrives will change dramatically. Simple data aggregation and lookup services—exactly what Thomson Reuters and LexisNexis sell—are most vulnerable. Complex infrastructure and integration? Probably safe, maybe even growth stories.
Winners and Losers: Who Gets Disrupted?
Most Vulnerable
Legal data and analytics: If Claude can review contracts, triage NDAs, and track compliance using your own documents, why pay for expensive database subscriptions? Companies like RELX, Thomson Reuters, and Wolters Kluwer face existential questions about their core business models.
Enterprise workflow wrappers: Dozens of startups raised money on the thesis “foundation model + our proprietary wrapper = valuable product.” Harvey, Legora, and other legal AI startups now compete directly with Anthropic—for free.
Simple SaaS automation: Any SaaS whose value proposition is “we automate X task” faces the same competitive pressure. If Claude Cowork can automate that task with a plugin, what’s the differentiation?
Probably Safe
Infrastructure providers: Nvidia (chips), cloud providers (compute), networking companies. More AI means more infrastructure spending, not less.
Integration specialists: Complex system integrators who handle messy enterprise deployments. AI makes this harder, not easier.
Proprietary data with network effects: Companies whose data improves with scale and can’t be replicated by training on public sources.
Potential Winners
Anthropic: Obviously. They’re positioning Claude as the platform everything else connects to.
Companies that embrace AI agents: Businesses that adopt these tools get efficiency gains. Their competitors who don’t fall behind.
AI-native startups: New companies built around AI workflows from day one, without legacy product lines to protect.
What This Means for Your Career
If you work in a field that Anthropic just released a plugin for—legal, sales, marketing, data analysis—this is your wake-up call.
If You’re in Legal Tech
The writing is on the wall. Routine contract review, NDA triage, and compliance tracking are automating now. Not in five years. Not in ten. Now.
What to do: Move up the value chain. Complex negotiations, novel legal questions, strategic advice—these require human judgment AI can’t replicate. Position yourself as the person who directs AI tools, not the person whose tasks AI tools replace. For a full breakdown of what’s available, see our best AI tools for lawyers guide.
If You’re in Software Sales/Marketing
Sales CRM automation and marketing campaign tools are next. The plugin is already on GitHub.
What to do: Focus on relationship-building and strategic thinking. AI can draft outreach emails and analyze campaign data. It can’t build genuine human connections or navigate complex political situations.
If You’re a Developer
Paradoxically, this might be good news. More AI deployment means more integration work, more customization, more debugging of AI-generated outputs.
What to do: Learn MCP (Model Context Protocol). Understand how AI agents interact with systems. Be the person who makes AI tools work in production, not just demo. Our guide to the best MCP servers and tools is a practical starting point.
If You’re an Investor
The repricing isn’t done. Software P/E ratios dropped 30% in one day, but the disruption will play out over years.
What to do: Distinguish between software companies with genuine moats (network effects, proprietary data, complex integration) and those selling commoditizable workflows. The latter group has a lot further to fall.
Our Take: The Honest Assessment
Was the selloff an overreaction? Probably, in the short term. Markets are emotional. A 16% drop in Thomson Reuters implies massive revenue collapse that hasn’t actually happened yet.
Is the selloff directionally correct? Absolutely. The competitive dynamics have permanently shifted. Foundation model companies competing directly with software vendors changes every assumption about market structure.
Will software companies adapt? Some will, some won’t. The smart ones are already building AI into their products. The slow ones are the next quarter’s casualties.
Should you panic about your job? No, but you should prepare. The tasks AI automates today become table stakes tomorrow. The professionals who thrive are the ones who work with AI, not against it. (For practical advice, see our guide on how to use AI tools without losing your job.)
Here’s the uncomfortable truth: This is the first time AI disruption showed up in a Bloomberg terminal. Wall Street traders aren’t AI researchers. They don’t read Arxiv papers. But they understand market cap destruction.
When $300 billion evaporates because of a GitHub repository, you’ve crossed from “interesting technology” to “market-moving force.” There’s no going back.
The Bigger Picture
Jensen Huang calls the selloff “illogical.” Anthropic keeps shipping plugins. Investors keep selling software stocks.
Who’s right?
Everyone, depending on your time horizon.
Short-term: The selloff is probably overdone. These companies still have revenue, customers, and time to adapt.
Medium-term: Significant disruption is coming. Business models built on data aggregation and simple workflow automation face real competitive pressure.
Long-term: The software industry won’t disappear. It will transform. New categories will emerge. Some incumbents will adapt. Others will become case studies in business school courses about disruption.
The one thing that’s certain: Tuesday, February 3, 2026, will be remembered as the day AI disruption stopped being a hypothetical and started being a market force.
What you do with that information is up to you.
📬 Get weekly AI tool reviews and comparisons delivered to your inbox — subscribe to the AristoAIStack newsletter.
Keep Reading
- Claude Cowork Plugins: The Market Collapse
- Best AI Agents 2026
- MCP Protocol Explained
- ChatGPT vs Claude: Which Should You Use?
- Claude 4 Opus Deep Dive
- How to Use AI Tools Without Losing Your Job
Last updated: February 2026
