MUMBAI– Shares of IBM suffered their steepest single-day drop in more than 25 years after fresh concerns emerged that advances in artificial intelligence could undermine one of the company’s most stable and profitable businesses: legacy software modernization.
IBM stock fell 13.2 percent to close at $223.35, marking its largest one-day decline since October 18, 2000. The shares are now down about 25 percent so far this year as investors reassess how quickly AI could reshape the economics of enterprise software and IT services.
The sell-off was sparked by a blog post from artificial intelligence startup Anthropic, which said its AI tool, Claude Code, can understand and modernize COBOL, a programming language developed in the 1950s that still underpins many of the world’s most critical computer systems.
COBOL remains deeply embedded in banks, airlines, insurance companies, and government agencies, and it is a cornerstone of IBM’s mainframe business. For decades, maintaining and updating COBOL-based systems has been slow, costly, and heavily reliant on large teams of specialized consultants.
That complexity has generated steady revenue for IBM, as many organizations struggle to maintain or upgrade aging systems that few engineers fully understand today. Anthropic argues that AI fundamentally alters that equation by dramatically lowering the cost and effort required to analyze and update legacy code.
In its post, Anthropic said hundreds of billions of lines of COBOL are still running in live systems every day, even as the pool of engineers fluent in the language continues to shrink. The company said AI is particularly effective at handling the complex, time-consuming tasks that historically made modernization projects prohibitively expensive.
Anthropic estimates that roughly 95 percent of ATM transactions in the United States still rely on COBOL, highlighting how deeply the language is embedded in the financial system.
According to the company, its AI tools can scan massive codebases, map dependencies between different software components, generate clear documentation for poorly understood systems, and identify potential risks that would typically take months to uncover through manual analysis.
“Modernisation has been stuck for years because understanding old code often costs more than rewriting it. AI flips that equation,” Anthropic said.
The development has intensified investor concerns that AI-driven automation could erode long-standing revenue streams tied to legacy systems, posing a structural challenge for companies like IBM that have long benefited from the complexity of maintaining outdated but mission-critical software. (Source: IANS)
