Shares of International Business Machines tumbled sharply on Monday, marking the company’s worst single-day decline in over 25 years.

The sell-off followed bold claims by AI startup Anthropic that its Claude Code tool could transform legacy COBOL systems widely used on IBM infrastructure.

IBM’s stock fell 13.2 per cent in a single session, its sharpest daily decline since October 18, 2000.

The sudden downturn rattled investors and revived concerns about the future of legacy programming systems that underpin much of the global financial and government infrastructure.

The drop came after Anthropic announced that its AI-powered coding assistant, Claude Code, could significantly modernise COBOL — a programming language heavily relied upon by organisations running IBM mainframes.

COBOL modernisation at centre of concerns

COBOL remains extensively used across banking, insurance and government sectors, particularly on IBM systems.

Anthropic said modernising COBOL systems has traditionally required “armies of consultants” spending years mapping workflows and analysing complex codebases.

“Tools like Claude Code can automate the exploration and analysis phases that consume most of the effort in COBOL modernization,” the company wrote in a blog post.

“With AI, teams can modernize their COBOL codebase in quarters instead of years,” it added.

The announcement appears to have triggered market fears that AI tools could disrupt traditional enterprise software and consulting models tied to legacy systems.

Broader pressure on software

The market reaction was not limited to IBM.

Software stocks have faced pressure in recent months amid growing investor concerns over the expanding capabilities of generative AI tools.

Shares of cybersecurity firms such as CrowdStrike and Datadog also slumped on Monday, as investors assessed the potential impact of Anthropic’s newly introduced security tool on the broader industry.

Anthropic’s recent rollout of plug-ins for its large language model Claude has been widely seen as a push to position the company as an application-layer competitor in enterprise software.

Pentagon talks add to uncertainty

Adding to the turbulence, Axios reported that U.S. Defense Secretary Pete Hegseth has summoned Anthropic CEO Dario Amodei to the Pentagon for discussions over the military use of the Claude AI system.

According to Axios, the meeting is expected to involve potentially tough talks over how the U.S. military can deploy Anthropic’s artificial intelligence tools.

Earlier this month, Reuters exclusively reported that the Pentagon was urging major AI firms, including OpenAI and Anthropic, to make their tools available on classified networks with fewer standard user restrictions.

Axios also reported that the Pentagon had considered cutting ties with Anthropic over its insistence on maintaining restrictions on military usage of its AI models.

Defense officials reportedly said the talks are on the verge of collapsing. One senior official told Axios that this is not a “get-to-know-you meeting.”

An Anthropic spokesperson, however, told Axios that discussions are “productive” and being conducted in good faith.

Reuters said it could not independently verify the Axios report. The Pentagon, the White House and Anthropic did not immediately respond to Reuters’ requests for comment.