Anthropic Just Crashed IBM's Stock 13% by Threatening to Automate Its Most Profitable Business

IBM Just Had Its Worst Day in 25 Years

IBM shares plunged 13% on February 23, wiping over $31 billion from the company's market value in a single trading session. It was IBM's worst single-day loss since 2000. The trigger wasn't an earnings miss or an accounting scandal. It was an announcement from Anthropic that its Claude Code AI tool can now modernize legacy COBOL systems, the exact type of work that generates billions of dollars a year in consulting revenue for IBM.

The sell-off wasn't confined to IBM. Accenture and Cognizant shares also dropped, as investors connected the dots: if AI can automate the painstaking process of analyzing, documenting, and migrating COBOL codebases, a massive chunk of the enterprise consulting industry's revenue is at risk. The S&P 500's technology sector was one of the day's worst performers, and the broader Dow Jones dropped 822 points, with IBM alone responsible for a significant portion of the decline.

Why COBOL Matters More Than You Think

COBOL is a programming language from 1959. Most people assume it's a relic. It's not. COBOL handles an estimated 95% of ATM transactions in the United States. Hundreds of billions of lines of COBOL code run in production every single day, powering critical systems in banking, insurance, airlines, government agencies, and healthcare. The IRS runs on COBOL. Major banks run on COBOL. The systems that process your paycheck almost certainly touch COBOL somewhere in the chain.

The problem is that fewer and fewer people know how to work with it. The average COBOL programmer is well into retirement age, and computer science programs stopped teaching the language decades ago. This has created a massive, lucrative market for consulting firms that specialize in maintaining, documenting, and gradually modernizing these legacy systems. IBM, Accenture, and a handful of other firms have built multi-billion dollar practices around this work.

Modernization projects are notoriously expensive and slow. A typical COBOL migration for a large bank can cost hundreds of millions of dollars and take years to complete. The work involves mapping out complex dependencies between thousands of programs, documenting business logic that was written decades ago with minimal documentation, testing every edge case, and carefully migrating functionality to modern languages. It's exactly the kind of tedious, methodical, high-volume analysis work that AI is increasingly good at.

What Claude Code Actually Does

Anthropic demonstrated that Claude Code can automate the exploration and analysis phases of COBOL modernization. This is the grunt work that traditionally requires large consulting teams: reading through legacy code, mapping dependencies, documenting what each module does, identifying risks, and creating migration plans.

The claim isn't that AI can replace the entire modernization process. Writing new code, testing it, and deploying it still requires human engineers. But the analysis phase, understanding what the existing code does and creating a roadmap for migration, is where consulting firms charge the most and deploy the largest teams. If AI can compress months of analysis into days, the economic value of those consulting engagements shrinks dramatically.

Wall Street's reaction suggests investors believe the threat is real. A 13% drop isn't a knee-jerk overreaction; it's a fundamental repricing of IBM's consulting revenue outlook. The market is saying that a meaningful portion of IBM's projected future earnings from legacy system modernization may not materialize if AI tools can do the same work faster and cheaper.

The Consulting Industry's AI Reckoning

IBM's crash is the most dramatic example, but the pressure is building across the entire consulting sector. Just days before the IBM crash, Accenture announced that it would tie promotions for senior staff to their adoption of AI tools. Associate directors and senior managers were told via email that "regular adoption" of AI tools would be a "visible input to talent discussions." The company is tracking weekly logins to its AI Refinery platform, built with NVIDIA.

This came after Accenture laid off 11,000 employees it determined could not be reskilled for AI roles, spending $2 billion on severance over the past three years. The company trained 550,000 of its roughly 780,000 workers on generative AI fundamentals. The message is clear: adapt or leave.

Meta has also announced that "AI-driven impact" will be a core metric in performance evaluations starting in 2026. Workers need to demonstrate that they've leveraged AI to succeed in their roles and built tools to improve productivity.

These aren't isolated decisions. They reflect a growing corporate consensus that AI competency is becoming a baseline requirement for knowledge workers, not a nice-to-have skill. The companies that employ the largest consulting workforces are simultaneously investing in AI tools that could reduce the need for those workforces.

Who Wins and Who Loses

The IBM crash clarifies the winners and losers in the AI disruption of professional services.

Winners: Companies that build AI tools for enterprise modernization (Anthropic, potentially OpenAI and Google), companies with large legacy codebases that can now modernize faster and cheaper (banks, insurers, government agencies), and a small number of highly skilled engineers who can oversee AI-assisted modernization.

Losers: Consulting firms whose revenue depends on labor-intensive analysis work, mid-career consultants whose primary skill is understanding legacy systems (a skill AI is now acquiring), and any company whose business model is built on the assumption that modernization will remain slow and expensive.

The irony is sharp. IBM has invested heavily in its own AI offerings, including watsonx. But the threat to its consulting business isn't coming from its own AI products; it's coming from a competitor's AI that targets the very work IBM charges its clients millions to perform.

The Broader AI Scare Trade

IBM's crash is part of a larger pattern that's been shaking markets throughout February 2026. Each time an AI company demonstrates a new capability that threatens an established business line, the affected stocks take immediate hits. It's been happening to software companies, cybersecurity firms, and now consulting giants.

Microsoft and CrowdStrike were also notable losers on February 23, as investors broadly repriced the risk that AI tools will compress the value of professional services software. The market is essentially going through a sector-by-sector reassessment of which businesses are AI-enhanced (their work gets more valuable with AI) versus AI-disrupted (their work gets replaced by AI).

For the stock market, this creates a paradox. AI companies are growing rapidly and generating enormous revenue, which pushes indices up. But the companies they're disrupting are often large, established firms with significant index weight, which pushes indices down when they sell off. The net effect has been volatile, sector-specific rotations rather than a clear market direction.

What This Means Going Forward

The COBOL announcement is a proof of concept, not a finished product. Actual enterprise COBOL modernization using AI tools will take time to prove out in production environments. Banks and government agencies aren't going to rip out their consulting contracts overnight based on a product demo.

But the direction is unmistakable. AI tools for code analysis and modernization will get better, faster, and cheaper. The consulting firms that recognize this and restructure their business models around AI-augmented services will survive. Those that try to protect the old model of throwing hundreds of consultants at a legacy modernization project will find their clients increasingly skeptical of the price tag.

The real question for IBM specifically is whether its own AI products can compete with Anthropic and others in the enterprise modernization space. If IBM can't match the capability, it loses consulting revenue. If it can, it cannibalizes its own consulting business. Neither path is comfortable, which is exactly why the stock dropped 13% in a day.

For the broader workforce, the Accenture and Meta announcements are the canary in the coal mine. When the world's largest consulting firm tells its senior staff that AI adoption is a prerequisite for promotion, and then lays off 11,000 people who couldn't adapt, that's not a trend anymore. That's the new reality.