On February 23, 2026, IBM experienced its sharpest single-day stock drop in more than twenty years, tumbling over 13 percent after Anthropic introduced a new coding tool called Claude Code. In a matter of hours, billions of dollars in market value vanished, not because IBM stumbled operationally, but because investors suddenly saw a future arriving faster than expected.
The anxiety wasn’t about a product recall or an earnings miss. It was about relevance.
For decades, IBM’s quiet strength has been its mainframe business: the sprawling backbone systems that run banks, airlines and government agencies. Many of them written in aging programming languages like COBOL. Maintaining and modernizing that code has been complex, specialized and highly profitable work, and few companies could do it at IBM’s scale.Claude Code claims to reshape that equation. Anthropic says it can navigate and update dense legacy systems with far less human intervention. If AI can modernize legacy mainframe code autonomously, IBM’s high-margin services engine could start to crack.
Markets trade on perception as much as performance. And on this day, perception shifted. Investors weren’t reacting to what IBM is, they were reacting to what AI might soon make unnecessary.