Anthropic’s Claim on Modernizing COBOL Triggers Rout in IBM Stock
The selloff reflects a deeper fear that AI could erode the long timelines and specialist labor that have long underpinned legacy modernization revenues.
Topics
News
- McKinsey, OpenAI Team Up on Enterprise AI Rollout
- TCS Partners ServiceNow to Embed Agentic AI Across Global Businesses
- Altman’s AI Energy Comparison to Human Development Draws Flak
- Why Open Networks, DPIs Are Key to Scaling Social Impact
- Anthropic Expands Claude Into Code Security With AI Reasoning
- IBM Expands GovTech Push in India With Lucknow AI Center
AI firm Anthropic’s latest claim that its artificial intelligence can untangle decades-old COBOL systems in months rather than years sent International Business Machines Corp. (IBM) shares down 13%, the company’s steepest one-day decline in more than a quarter century, as investors reassessed the economics of one of enterprise technology’s most entrenched businesses.
In a blog post published on Monday, the AI startup argued that legacy modernization has long stalled because understanding old code cost more than rewriting it.
Tools such as its Claude Code platform, the company said, can automate much of the exploration and analysis that historically required large consulting teams, allowing organizations to modernize systems “in quarters instead of years.”
COBOL, developed more than half a century ago, still runs critical infrastructure across finance, airlines and government.
Anthropic estimates the language handles roughly 95% of ATM transactions in the US, with hundreds of billions of lines in production globally.
Yet the engineers who built many of those systems have retired, documentation is often incomplete and few new programmers are trained in the language. That scarcity has helped keep modernization projects slow, expensive and dependent on specialist expertise.
For IBM, whose mainframes continue to underpin many of those systems, the issue is more about the potential compression of time and labor. Modernizing legacy environments has typically meant years of consultants mapping how programs connect, tracing data flows and identifying hidden dependencies before safely translating them into newer architectures.
Those long timelines have supported steady revenue from servicing and upgrading installed systems, particularly in regulated industries where caution and continuity matter.
Anthropic’s argument targets that first phase, or the understanding. The company said its system can read entire COBOL codebases, map relationships across thousands of files, surface risks that might take analysts months to uncover and generate documentation for workflows that organizations no longer fully remember building.
In its framing, AI handles the heavy groundwork while human engineers focus on strategy, oversight and validation.
The company did not present independent third-party verification of large-scale deployments, and migrating mission-critical systems remains complex.
Regulatory scrutiny, operational risk and internal approvals can slow projects regardless of how quickly code is analyzed. Still, markets appear to be reacting to the possibility that one of enterprise IT’s most time-consuming bottlenecks could shrink.
IBM’s decline came amid broader volatility in software and technology shares as investors weigh how rapidly advancing AI tools may reshape traditional revenue streams. But the move stood out for its size, reflecting how closely the company is associated with the world of legacy systems and modernization services.
Analysts and investors will likely watch closely for early evidence of customer adoption of Claude Code or similar tools in large modernization efforts, as well as any commentary from IBM on how it views the competitive landscape.
