Wall Street’s anxiety over artificial intelligence is spreading, and the latest target is IBM (IBM). The legacy tech giant’s stock suffered its worst single-day drop in more than a quarter-century on Monday, tumbling over 13% and wiping roughly $31 billion off its market capitalization.
The trigger for the sell-off wasn’t an earnings miss or a guidance cut—it was a blog post from AI startup Anthropic.
In a blog post yesterday, Anthropic said that AI tools could help modernize systems based on COBOL code. The company, which is behind the Claude AI platform, noted that its coding tools could be used to automate the experimentation, analysis, and implementation phases associated with modernizing COBOL-based systems. The post also pointed out that COBOL is taught at only a handful of universities today, and the pool of engineers capable of reading the code is shrinking.
That was enough to put investors on edge. COBOL, the programming language born in the 1950s, still forms the bedrock of IBM’s mainframe business. The market fears that if Anthropic and other AI leaders can easily replace COBOL systems, IBM’s mainframe infrastructure and the consulting services built around it could face an existential threat.
This isn’t a niche part of IBM’s business. Mainframe sales accounted for 23% of the company’s overall revenue last year, while mainframe-related software represented roughly 29% of total software sales. Any signal that this revenue stream could erode is bound to make investors nervous.
IBM was quick to respond to the concerns. A company spokesperson argued that translating COBOL is actually “the easy part” and that the real work is “data architecture redesign, runtime replacement, transaction processing integrity, and hardware-accelerated performance built over decades of tight software and hardware coupling.”
The company also noted that new AI tools emerge every week, and pointed to the launch of its watsonX Code Assistant for its Z mainframe two years ago as an example of its own initiatives to use AI for code modernization.
Analysts have also rushed to IBM’s defense. Evercore ISI analyst Amit Daryanani called Monday’s sell-off “unwarranted.” “The idea of shifting off of mainframe is not a new concept,” he said in a note to clients. “IBM’s customers have had ample opportunities to migrate off of mainframe and are sticking with the platform given inherent advantages.” He noted that customers gravitate toward benefits like 100% uptime, quality encryption and cost efficiencies.
Jefferies analyst Brent Thill echoed that view, suggesting the market was taking an oversimplified view of IBM’s business. “Modernization on mainframe is rarely just code translation and documentation; it requires deep integration with operational resilience, performance tuning, and change management, areas where IBM already sits at the center,” Thill wrote in a note to clients. He views the mainframe business as “resilient” given examples of how customers are growing, not shrinking, their workloads. Mainframes are “deeply entrenched” across industries, he added.
Thill also noted that IBM “is already disrupting itself by productizing AI-native modernization tools that directly address the core challenges in its install base, including a shrinking COBOL talent pool and the need to reverse-engineer decades of business logic.”
After Monday’s bloodbath, IBM shares saw a modest recovery on Tuesday, gaining 2.7% as of 3 p.m. ET. The stock is now down roughly 22% across this year’s trading.
From a fundamentals perspective, IBM isn’t in trouble. The company generated $67.5 billion in revenue last year, up 8% annually—or 6% on a currency-adjusted basis. It’s projecting currency-adjusted revenue growth of roughly 5% for the current fiscal year. Free cash flow is expected to come in at approximately $15.7 billion, representing about 7% annual growth.
The stock offers a dividend yield of roughly 2.9%, which should appeal to value investors, and shares are currently valued at about 18.5 times this year’s expected earnings—hardly an outrageous valuation.
The problem is relative performance. Compared to other tech giants riding the AI wave, IBM’s sales and earnings growth look decidedly modest. Even if the market overreacted to Anthropic’s blog post, the potential threat that AI poses to some of IBM’s consulting and software businesses is real. As one market observer put it, the actual impact of tools like Claude Code on COBOL and IBM’s mainframe business is difficult to estimate right now. Perhaps the market did overreact. IBM’s entrenched legacy business, reputation for reliability, and in-house modernization tools may well help it weather the AI-driven shift.
But the risks can’t be completely dismissed either. With growth already looking relatively muted compared to other big players in the tech space, any uncertainty is amplified by the market. That’s why a single blog post from Anthropic was enough to send Big Blue into its worst tailspin in more than 25 years.
