Anthropic’s blog post revealed its Claude AI can decode and upgrade COBOL systems, the 1950s-era language that underpins mission-critical operations from finance to aviation. This year, IBM stock is down 25%, reflecting investor panic over AI’s potential to accelerate the obsolescence of legacy software maintenance—a goldmine for IBM’s consulting and mainframe divisions.

For decades, firms have grappled with COBOL’s upkeep: costly, consultant-dependent, and hampered by a shrinking pool of programmers. Banks process billions in transactions daily on COBOL; U.S. ATMs rely on it for 95% of operations. Anthropic argues AI flips the script, scanning massive codebases, untangling interdependencies, documenting the undocumented, and flagging hazards swiftly.

‘Legacy modernization is notoriously expensive—often more so than a full rewrite,’ the startup explained. Hundreds of billions of COBOL lines persist in production, but AI could render human-led overhauls unnecessary. IBM, long profiting from this inertia, now confronts a future where AI democratizes updates, potentially slashing demand for its services.

The market’s verdict was swift and brutal. As AI reshapes software engineering, legacy players must pivot or perish, with IBM’s tumble marking a pivotal moment in tech’s evolution.