Shares of International Business Machines (IBM) suffered their steepest single-day decline in more than 25 years last Monday (23). The company’s market value dropped from $240.16 billion on Friday (20) to $208.59 billion, wiping out more than $30 billion (around R$155 billion) in a single session.
The sharp fall came after Anthropic revealed that its tool Claude Code can analyze and help modernize systems written in COBOL — a language that plays a key role in IBM’s mainframe ecosystem.
Why the news shook investors
Updating legacy COBOL systems has traditionally been a slow and expensive process. Companies often need large consulting teams and multi-year projects just to understand old code before making any changes.
Anthropic’s approach aims to automate much of that early work, such as exploring codebases, identifying dependencies, and generating documentation. If successful, the time needed for modernization could shrink from years to just a few quarters.
For the market, this raises a serious concern. A large part of IBM’s long-term revenue has come from maintaining and upgrading legacy environments. If AI tools make it easier for businesses to move away from these systems, demand for those services could decline.
According to FIAP professor Gustavo Galegale, IBM’s long-standing dominance was built around mainframes and the complex services tied to them. Faster migration to cloud platforms would reduce reliance on proprietary infrastructure.
“If the technical barrier around COBOL disappears, part of IBM’s strategic advantage goes with it,” he noted — a risk investors reacted to quickly.
Market reaction
IBM’s stock fell 13.2%, closing at $223.35, marking its worst daily percentage loss since October 18, 2000. The broader software sector has also been under pressure, with software-focused ETFs down significantly this year.
The ripple effect extended beyond IBM. After Anthropic introduced new security-related capabilities, cybersecurity firms such as CrowdStrike and Datadog also saw their shares decline.
The sell-off reflects growing investor concern that rapid advances in AI could disrupt traditional enterprise technology business models faster than expected.
IBM responded through Senior Vice President and Chief Commercial Officer Rob Thomas, emphasizing that the company’s mainframe value is based on security, reliability, and resilience, not just the COBOL language itself.
What is COBOL?
COBOL (Common Business-Oriented Language) was created in the late 1950s. Although often seen as outdated, it still runs many critical systems around the world.
Today, an estimated 250 billion lines of COBOL code remain active in production. According to IBM:
- About 95% of ATM transactions in the United States rely on COBOL
- Around 80% of in-person credit card transactions are processed through systems built on it
Over time, COBOL became closely tied to IBM’s mainframes, which were designed for centralized, highly controlled environments that are difficult and costly to replace.
The challenge of legacy modernization
One of the biggest obstacles in updating these systems is the shortage of skilled professionals. Mainframe expertise takes years to develop, and COBOL is rarely taught in modern computer science programs.
As a result:
- Understanding old systems is expensive and time-consuming
- Documentation is often incomplete or outdated
- Migration projects carry significant risk
Anthropic argues that in many cases, the cost of simply understanding legacy code has become higher than rewriting it. Its AI tool aims to speed up this phase by automatically mapping relationships across large codebases and generating technical documentation.
A bigger shift underway
Automation at this level raises questions about the future of traditional enterprise infrastructure. As cloud platforms and distributed systems become the norm, companies may feel less tied to proprietary environments.
The key question for the market now is whether mainframes will remain essential for large organizations — or whether AI-powered modernization will make it easier to move away from them.
For IBM and its investors, the answer could shape the company’s long-term role in an industry that is being reshaped by AI faster than many expected.
 suffered their steepest single-day decline in more than 25 years last Monday (23).){kind=link}