A single AI breakthrough sent shockwaves through Wall Street, with IBM shares plummeting 13.2% to $223.35—the worst one-day loss since the dot-com hangover of 2000.
Down 25% year-to-date, the tech titan grapples with existential threats from Anthropic’s Claude Code, an AI that claims prowess in reviving COBOL, the dinosaur language still animating global finance and infrastructure.
COBOL’s ubiquity is staggering: it drives banking cores, flight systems, policy processing, and public sector backends, including IBM’s mainframe cash cow. Upgrades have been a lucrative nightmare—prohibitively expensive, reliant on vanishing specialists.
Anthropic’s post paints a revolutionary picture. With trillions of COBOL lines live and talent scarce, AI excels at holistic analysis: dependency mapping, risk detection, documentation creation—all in fractions of traditional timelines. U.S. ATMs? 95% COBOL-dependent. Modernization? ‘AI makes understanding legacy code cheaper than rewriting,’ they assert.
IBM’s model—perpetual maintenance contracts—teeters. As AI slashes barriers to modernization, enterprises could ditch costly consultants for self-service tools, eroding IBM’s moat.
The rout underscores AI’s dual edge: innovation accelerator and disruptor of incumbents. For IBM, it’s a call to reinvent amid the legacy code apocalypse.