IBM shares plunge, steepest drop since 2000, as Anthropic claims AI can modernize COBOL

Anthropic’s announcement that its Claude Code tool can automatically refactor legacy COBOL applications sent shockwaves through the mainframe market, a sector IBM has long championed. In a blog post published Monday, the AI start‑up argued that the scarcity of COBOL‑savvy developers—“a talent pool that has not grown despite training efforts,” the post notes—makes AI‑driven migration both a cost‑saving necessity and a risk‑mitigation strategy for banks, airlines and government agencies that still rely on the language (The Register). By positioning Claude Code as a “hammer” that can safely assess which components are ready for migration and flag those with hidden technical debt, Anthropic implied that IBM’s own Watsonx Code Assistant for Z, which IBM unveiled three years ago to translate COBOL to Java, may already be outdated (The Register).

Investors reacted swiftly. By the close of trading, IBM shares had slumped 13 percent, marking the steepest one‑day decline since the dot‑com bust of 2000 (Reuters). The plunge eclipsed the 6.5 percent dip reported earlier in the day, underscoring how quickly market sentiment can turn on perceived competitive threats. Bloomberg’s coverage linked the sell‑off to broader AI‑risk concerns, noting that research firm Citrini had recently flagged “AI‑related operational risks” across software and payments stocks, a narrative that amplified the anxiety around legacy‑system modernization (Bloomberg). The confluence of Anthropic’s technical claim and the broader AI‑risk discourse created a perfect storm for IBM’s stock.

The fallout also reignited a debate that has simmered in the industry for years: whether AI can truly replace human expertise in mission‑critical legacy environments. Infosys chairman Nandan Nilekani, speaking at a conference last week, argued that AI has lowered the cost of rewriting legacy code to the point where migration is now “affordable and imperative” (The Register). Meanwhile, cloud giants AWS, Microsoft and IBM spin‑off Kyndryl have each launched their own mainframe‑migration services, betting that AI will be the differentiator that convinces risk‑averse enterprises to move off entrenched systems (The Register). Anthropic’s claim, therefore, is not an isolated technical boast but a strategic signal that the AI‑enabled migration market is heating up, and that incumbents may need to accelerate their roadmaps or risk losing relevance.

For IBM, the immediate challenge is twofold. First, it must reassure investors that its Watsonx Code Assistant remains competitive. The company has long marketed its mainframe AI tools as a bridge for existing customers, but Anthropic’s public demonstration of a comparable—or potentially superior—solution forces IBM to either double‑down on its product suite or seek partnerships that can integrate third‑party AI capabilities. Second, IBM must address the broader perception that its legacy‑hardware business is vulnerable to disruption. While the firm continues to generate roughly 10 percent of its annual revenue from mainframe services, analysts cited by Bloomberg have warned that “AI‑disruption fears” could erode that base if rivals can deliver faster, cheaper migration paths (Bloomberg).

The episode also highlights how quickly market narratives can shift when a single blog post touches a strategic pain point. Anthropic’s concise technical note—without any accompanying financial projections—was enough to trigger a 13 percent sell‑off, suggesting that investors are already pricing in the risk that AI could upend the economics of legacy‑system maintenance. As the AI arms race expands beyond generative chat and into specialized engineering tools, the next few quarters will likely see more incumbents like IBM, Microsoft and AWS scrambling to showcase their own AI‑driven refactoring capabilities, lest they become the next headline of a steep market decline.