IBM shares suffered their steepest single-day fall in over two decades on Monday, February 23, 2026, after artificial intelligence startup Anthropic unveiled a disruptive new tool aimed squarely at the heart of the tech giant’s legacy business. The tool, Claude Code, promises to modernize COBOL—the decades-old programming language underpinning much of the world’s financial infrastructure—with an efficiency and speed that has left Wall Street rattled and forced investors to rapidly rethink the future of legacy software consulting.
COBOL, which stands for Common Business-Oriented Language, is a relic of the late 1950s and 1960s but still quietly runs the show behind the scenes. According to Anthropic’s own blog post, "COBOL is everywhere. It handles an estimated 95% of ATM transactions in the US. Hundreds of billions of lines of COBOL run in production every day, powering critical systems in finance, airlines, and government. Despite that, the number of people who understand it shrinks every year." The language’s staying power is legendary, but its very age has become a liability as the pool of engineers who can read and maintain it dwindles with each passing year.
Anthropic’s announcement immediately sent shockwaves through the markets. IBM’s stock tumbled 13% by the closing bell, ending the day at $223.39—a loss of over $31 billion in market value, according to data compiled by Bloomberg and Forbes. This marked IBM’s largest single-day percentage drop since the dot-com bubble burst in October 2000, and capped a brutal February in which the company’s shares have fallen 27%, on track for their worst monthly performance since at least 1968.
The reason for the panic? Anthropic’s Claude Code tool directly threatens the lucrative consulting and modernization business that IBM, along with firms like Accenture and Cognizant, has built around helping organizations update their aging COBOL systems. Traditionally, modernizing COBOL codebases involved massive teams of consultants painstakingly mapping out dependencies, documenting workflows, and searching for risks—a process that could drag on for years and cost a fortune.
Anthropic claims its AI tool can do this grunt work in a fraction of the time and cost. In its Monday blog post, the company explained, "Claude Code can help modernize COBOL codebases by mapping dependencies across thousands of lines of code, documenting workflows and identifying risks that would take human analysts months to surface." The tool can even trace execution paths, map data flows between modules, and identify program entry points, automating what has long been the most labor-intensive and expensive part of legacy code modernization.
"Legacy code modernization stalled for years because understanding legacy code cost more than rewriting it. AI flips that equation," Anthropic argued. The company also released a Code Modernization Playbook to help organizations navigate the transition, signaling its intent to make the process as accessible and rapid as possible. For companies weighed down by "technical debt"—the mounting future costs of maintaining and patching old software—this could be a game changer.
IBM, for its part, has long been synonymous with COBOL. The company helped popularize the language and still sells mainframe systems optimized for large-scale transaction processing, where COBOL remains king. But as Anthropic’s tool threatens to automate much of the value IBM provides in this space, investors are questioning whether the old guard can keep up with the rapid pace of AI-driven change.
The impact of Anthropic’s announcement wasn’t limited to IBM. Shares of Accenture and Cognizant Technology Solutions also declined, as both companies generate significant revenue from legacy system modernization practices. The broader tech sector was already under pressure Monday due to tariff uncertainty and a research report circulating online warning of far-reaching negative impacts from AI. The result was a volatile "sell first and ask questions later" mood on Wall Street, with software and cybersecurity stocks—already battered by Anthropic’s earlier AI plugin releases—taking another hit.
Anthropic’s recent flurry of updates has targeted a range of business processes, from legal and customer service automation to cybersecurity. Just last week, the startup introduced Claude Code Security, a feature designed to scan codebases for vulnerabilities, which led to a sharp drop in cybersecurity stocks like CrowdStrike and Zscaler. The cumulative effect has been to fuel investor fears that AI could rapidly disrupt not just software development, but the entire ecosystem of companies built around maintaining and upgrading legacy technology.
Still, not everyone is convinced that the sky is falling. Several economists and analysts have urged caution, warning that the market may be overreacting to the latest AI developments. LPL Financial analyst Adam Turnquist wrote that the volatility reflected a "market narrative" shift rather than any actual decline in revenues or profits. JPMorgan, for its part, called the idea that AI companies would upend the software industry "broken logic," while WedBush Securities analyst Dan Ives told CNBC the broader selloff was "the most disconnected trade I’ve ever seen in my career on Wall Street." Ives argued that, in the long run, AI advancements could actually boost companies like IBM by making their services more efficient and valuable.
Yet the numbers are hard to ignore. As of February 23, IBM’s market value had dropped from $240.8 billion to about $208.7 billion in just a few days. The company’s shares have been battered by a one-two punch of AI disruption and broader market jitters, and there’s little sign that the turbulence will subside soon. The shrinking pool of COBOL experts—taught by only a "handful" of universities, as Anthropic pointed out—means that the need for modernization is only becoming more urgent.
For now, the story of COBOL is a reminder that the past is never really past in the world of technology. The systems built decades ago still run the world’s money, flights, and government services. But as Anthropic’s AI tool demonstrates, the way we maintain and upgrade those systems is changing—fast. Whether companies like IBM can adapt to this new reality, or whether AI startups will eat their lunch, is the billion-dollar question hanging over Wall Street today.
As the dust settles, one thing is clear: the battle for the future of legacy code modernization has only just begun, and the stakes—for Wall Street, Main Street, and the millions of people whose daily lives depend on invisible lines of code—couldn’t be higher.