Shares of International Business Machines (IBM) saw their steepest daily drop in over 25 years on February 23, 2026, falling 13.15% to close at $223.35. This significant decline, the most severe since October 18, 2000, and its worst performance since March 12, 2020, resulted in a loss of over $31 billion in market value for the company in a single trading session. The sell-off was triggered by anxieties among investors regarding the disruptive potential of artificial intelligence on legacy technology services.
The catalyst for this dramatic fall was a blog post from AI research firm Anthropic, creators of the Claude family of large language models. Anthropic claimed its "Claude Code" tool could significantly streamline, or even partially automate, the modernization process for COBOL (Common Business-Oriented Language). COBOL is a decades-old programming language fundamental to mission-critical systems across major banks, insurance companies, government systems, and global airlines, many of which run on IBM mainframes.
Investors are now reassessing the long-term viability of IBM's mainframe and consulting franchises in an era of rapid AI advancement. The concern is that if AI tools can meaningfully reduce the cost and effort involved in COBOL modernization, enterprises might increasingly opt for AI-driven solutions rather than costly, multi-year consulting engagements, thereby pressuring IBM's long-term services revenue streams. While IBM's financial results for the fourth quarter ended December 2025 showed revenue of $19.7 billion, up 12% year-on-year, with software revenue up 14% and consulting revenue up 3%, the market reacted strongly to Anthropic's announcement, suggesting a re-evaluation of future revenue streams.
