IBM Stock Records Steepest Drop in Decades Following Anthropic’s COBOL AI Announcement

On February 23, 2026, International Business Machines (IBM) experienced its most significant single-day market decline since the peak of the dot-com era. Shares of the technology giant cratered by 13.15%, closing at $223.35 and erasing more than $31 billion in market capitalization. The sell-off was triggered by a strategic announcement from AI research firm Anthropic, which detailed new capabilities for its "Claude Code" tool specifically designed to automate the modernization of COBOL, the aging programming language that remains the backbone of global financial infrastructure.

A Decisive Pivot in the AI Arms Race

The market reaction underscores a growing investor anxiety regarding the vulnerability of established enterprise services to specialized generative AI models. COBOL, or Common Business-Oriented Language, is a nearly 70-year-old language that powers an estimated 95% of ATM transactions in the United States and manages hundreds of billions of lines of code within government, airline, and insurance sectors. For decades, the high cost and complexity of migrating these systems—combined with a shrinking pool of proficient programmers—created a lucrative and defensible consulting niche for IBM.

Anthropic’s announcement directly challenged this economic model. According to the firm’s "Code Modernization Playbook," Claude Code can now automate the exploration and analysis phases of COBOL modernization, identifying dependencies and mapping workflows that typically require months of human labor. Anthropic claims that with these new AI capabilities, organizations can modernize legacy codebases in quarters rather than years, effectively commoditizing the high-margin advisory services traditionally provided by IBM’s consulting arm.

The $31 Billion Market Reassessment

Wall Street’s response was immediate and focused on the potential disruption of recurring revenue streams. While IBM has pivoted heavily toward hybrid cloud and AI in recent years, its mainframe business and associated consulting services remain critical pillars of its profitability. Analysts noted that the 13% drop represents more than just a fluctuation; it is a fundamental repricing of IBM’s legacy moat. The decline also rippled through the broader IT services sector, with competitors like Accenture and Cognizant seeing significant share price reductions as investors weighed the future of the "billable hour" model in an era of automated code refactoring.

Market data indicated that the selling pressure was driven by both institutional outflows and a technical breach of the stock’s previous support levels. Before this week, IBM had been trading near multi-year highs, buoyed by strong fourth-quarter earnings and growing adoption of its own AI platform, watsonx. However, the prospect of a third-party AI tool capable of unseating IBM’s dominance in the mainframe ecosystem prompted a swift reversal in sentiment.

Competitive Dynamics: Watsonx vs. Claude

The rivalry between IBM and Anthropic is not a new development, but it has entered a more aggressive phase. IBM previously launched the "watsonx Code Assistant for Z," which utilizes generative AI to help developers translate COBOL to Java. IBM’s strategy has focused on keeping clients within the mainframe ecosystem by offering tools that modernize code without necessitating a full migration to distributed cloud environments. In contrast, Anthropic’s messaging suggests that AI can now facilitate the safe migration of these systems to more modern, vendor-neutral platforms.

Industry experts suggest the primary threat lies in the discovery phase. A large portion of the cost in legacy modernization is not the coding itself, but the years of institutional knowledge buried in undocumented COBOL modules. If Claude can reliably reverse-engineer these logic flows, the barrier to switching away from IBM hardware and services drops significantly.

IBM’s Defense: Complexity Beyond Code

In response to the market volatility, IBM spokespeople emphasized that the challenge of legacy modernization extends far beyond simple code translation. The company argued that enterprise-grade modernization requires a deep understanding of data architecture, transaction processing integrity, and hardware-accelerated performance—areas where IBM has over half a century of expertise.

• Data architecture redesign remains a manual-intensive strategic task.
• Runtime replacement requires rigorous testing for transaction consistency.
• Hardware-software coupling on IBM Z provides security features that AI cannot yet replicate on generic cloud servers.

Despite these defenses, the market appears more concerned with the speed of AI evolution than with historical precedents. As Anthropic and its backers, including Amazon and Google, continue to push into specialized enterprise applications, the premium placed on legacy system expertise is likely to remain under intense scrutiny.