IBM’s stock ended Monday down 13% as Anthropic’s Claude Code threatens to dismantle a critical part of its business.

International Business Machines’ stock is the latest big software loser that can trace its selloff to fears over new artificial-intelligence features.

On Monday, IBM shares fell 13.2% following a newblog postfrom Anthropic sharing that Claude Code can automate parts of the modernization process for common business-oriented language (COBOL), a high-level programming language developed specifically for business data-processing needs. The stock ended the day as the worst-performing member of the S&P 500 index.

Shares of IBM are down 26.8% since the beginning of February, putting the stock on pace for its worst month since December 1992. The stock also staged its worst single-day performance since March 12, 2020, according to Dow Jones Market Data.

COBOL powers many everyday transactions in critical sectors such as banking, travel, insurance and government. The programming language, which was first released in 1960, runs 95% of ATM transactions in the U.S., according to Anthropic’s blog post. Modernizing COBOL has long been a complex, costly process that served as a lucrative moat for IBM — but Anthropic’s latest announcement is putting that part of IBM’s business into question.

Claude Code can “map dependencies across thousands of lines of code” and identify risks that would take human teams months to surface, according to the AI company’s blog post. “With AI, teams can modernize their COBOL codebase in quarters instead of years,” Anthropic wrote.

Until now, IBM had been perceived as a quiet winner within the AI trade, ending 2025 with over $12.5 billion in cumulative generative-AI business. IBM’s stock recently fetched a higher valuation than Microsoft’s, reflecting both the market’s current perception of AI winners and a preference for less capital-intensive tech businesses amid surging AI spending.

IBM’s dominance over COBOL has also been a critical part of its growth in both infrastructure and consulting. The company’s Z mainframe business, which provides core IT infrastructure for COBOL systems, grew 48% year over year in the fourth quarter to achieve its highest revenue in 20 years.

Anthropic’s Monday announcement directly targeted the traditional consulting model of COBOL modernization. “AI makes the economics work by automating what used to require armies of consultants, freeing your engineers to make the migration decisions that require their domain expertise,” the company wrote.

IBM is also developing its own AI tools to help the COBOL modernization process, with CEO Arvind Krishna sharing on its earnings call last month that the company’s Watsonx AI platform can “refactor COBOL into Java” and help people understand existing code on the COBOL platform.