IBM’s stock experienced a sharp decline on February 23, 2026 with effects carrying into the 24th, following an announcement from Anthropic about enhancements to its Claude AI model—specifically, capabilities in Claude Code for modernizing legacy COBOL systems.
The drop was significant: IBM shares fell around 13% not exactly 10%, though intraday lows hit near that level before closing lower, marking its worst single-day percentage decline since October 2000. This erased roughly $31 billion in market capitalization in a single session. The stock closed around $223, down from prior levels near $257, and February has seen a broader ~27% monthly slide so far.
What Triggered It?
Anthropic published a blog post on February 23, 2026, titled something along the lines of “How AI helps break the cost barrier to COBOL modernization.” In it, they highlighted how tools like Claude Code can automate key phases of COBOL modernization—such as: Mapping dependencies across massive codebases.
Documenting workflows. Identifying risks and issues that traditionally take human consultants months or years. They claimed this shifts modernization timelines from years to quarters, reducing the need for large consulting teams.
Register for Tekedia Mini-MBA edition 19 (Feb 9 – May 2, 2026).
Register for Tekedia AI in Business Masterclass.
Join Tekedia Capital Syndicate and co-invest in great global startups.
Register for Tekedia AI Lab.
COBOL; developed in the late 1950s still powers critical systems in banking ~95% of U.S. ATM transactions, airlines, insurance, governments, and more—most of which run on IBM mainframes.
IBM has long profited from maintaining these systems, licensing, hardware refreshes, and high-margin consulting and services for modernization projects that are notoriously expensive and complex due to outdated code and scarce expertise.
Investors interpreted this as a direct threat to IBM’s legacy revenue streams, sparking panic selling. Related stocks like Accenture and Cognizant also dropped around 6-7%. IBM executives including Rob Thomas pushed back quickly via blog posts and statements, arguing: Simply translating or analyzing COBOL code isn’t true modernization.
Mainframe value lies in platform architecture, security, uptime, compliance, and reliability—not just the language. New AI tools emerge weekly, but complex enterprise migrations still require IBM’s expertise and hybrid approaches.
IBM has its own AI tools; watsonx Code Assistant for Z for COBOL-to-modern-language translation. Some analysts called the sell-off an overreaction, noting mainframes’ entrenched role in mission-critical environments where reliability trumps speed.
Unlike IBM, Accenture did not issue a direct rebuttal on this specific announcement. The firm has aggressively leaned into AI itself: it has trained over 550,000 employees on generative AI tools, formed deep partnerships with OpenAI and others, and positions itself as the “bridge” helping clients adopt AI safely and at scale.
Executives have repeatedly said AI will augment rather than replace their services, allowing faster delivery and new offerings. Many analysts called the move an overreaction, noting that true modernization still needs human expertise for strategy, integration, testing, compliance, and change management—areas where Accenture’s domain knowledge remains hard to replicate quickly.
Early Feb 24 pre-market trading showed a modest rebound ~0.5% around $202, suggesting some dip-buying. This continues a clear pattern: every major Anthropic Claude Code-related announcement in recent weeks has triggered sector-wide drops in legacy tech and services stocks.
In short, the COBOL news amplified existing investor anxiety about AI commoditizing traditional consulting work, but Accenture’s scale, client relationships, and AI-first pivot give it tools to adapt—much like the broader industry. The stock remains volatile as the market reprices the speed of AI disruption.
This event fits a broader pattern of AI announcements from Anthropic (Claude) pressuring legacy tech and services sectors—similar drops hit cybersecurity stocks after prior Claude features, and Indian IT firms have faced pressure from AI automation fears. It’s a stark reminder of how quickly AI can reprice established business models, even if full disruption takes time.