IBM's shares experienced their steepest daily decline since 2000, plummeting 13.2% (and up to 13.1% by market close) and wiping approximately $31 billion off its market capitalization. This significant drop followed an announcement by AI startup Anthropic, stating that its Claude Code tool, part of the Claude large language model, could be used to modernize COBOL, a programming language that underpins critical systems on IBM mainframes in banking, insurance, and government.
Anthropic's blog post detailed how AI tools could automate the exploration and analysis phases of COBOL modernization, potentially reducing project timelines from years to quarters. This is particularly impactful as COBOL, though widely used (handling about 95% of US ATM transactions), faces a decreasing pool of skilled programmers. Investors reacted by selling off IBM stock, along with shares of other cybersecurity and consulting firms like CrowdStrike, Datadog, Accenture, and Cognizant, due to concerns about AI's disruptive potential.
IBM responded by stating that translating COBOL is only a part of modernization, emphasizing that the complex work involves data architecture redesign, runtime replacement, and transaction processing integrity. The company also highlighted its own AI initiatives, such as the watsonX Code Assistant for its Z mainframe, launched two years prior. Despite the sell-off, some analysts suggest the market might have overreacted, noting IBM's strong legacy sales base, reputation for reliability, and internally developed AI tools could help it navigate these shifts. However, the exact impact of AI on IBM's mainframe and software business, which accounted for a significant portion of its revenue last year, remains uncertain.
