IBM Plunges as AI Targets COBOL Core

Picture Credit: Inc.

Shares of IBM suffered their sharpest single-day drop in more than 25 years after AI startup Anthropic claimed its Claude Code tool can dramatically accelerate the modernisation of COBOL systems. The stock fell 13.2% to $223.35, wiping out roughly $40 billion in market value and extending its year-to-date decline past 24%. Investors reacted swiftly to the implication that a cornerstone of IBM’s business model legacy system modernisation may be vulnerable to AI disruption.

COBOL, short for Common Business-Oriented Language, was developed in 1959 with input from computing pioneer Grace Hopper. More than six decades later, it still powers vast swathes of global finance and government infrastructure. An estimated 95% of U.S. ATM transactions and 80% of in person credit card payments rely on COBOL based systems.

The Open Mainframe Project has estimated that roughly 250 billion lines of COBOL code remain in active use worldwide, much of it running on IBM’s mainframes, high performance servers optimised for large scale transaction processing.

The challenge has never been that COBOL doesn’t work. It works exceptionally well. The issue is talent scarcity. Fewer developers today are trained in the language, creating an expensive skills bottleneck for banks, insurers and public agencies that depend on it. Migration efforts have historically taken years, required armies of consultants and carried significant operational risk. That complexity has been a steady source of recurring revenue for IBM and major IT services firms.

Anthropic’s claim strikes at that pain point. The company says Claude Code can analyse dependencies across thousands of lines of legacy code, extract business logic and generate translations into modern languages like Java or Python in a fraction of the usual time. If AI can compress multi-year modernisation projects into months, the consulting-heavy model underpinning parts of IBM’s mainframe ecosystem could weaken. The ripple effects were visible beyond IBM, with IT and cybersecurity stocks also sliding as investors reassessed the durability of legacy software revenues.

Whether AI tools can truly handle enterprise scale migrations, with decades of undocumented logic, regulatory constraints and operational complexity, remains uncertain. Large organisations tend to move cautiously when core systems are at stake. Yet markets are pricing in disruption now, not later. IBM’s selloff reflects a broader recalibration across the software industry, where investors are questioning how much of traditional IT services and infrastructure revenue could shrink in an era of increasingly capable AI agents.

This article is proudly brought to you by: