Ibm Stock plunges after Anthropic boasts Claude can modernize COBOL
IBM shares tumbled more than 13% after Anthropic published a blog post trumpeting Claude's ability to modernize legacy COBOL code, and that sell-off pushed ibm stock into its largest single-day fall since October 2000. The episode underscores how quickly investor sentiment can shift when AI firms claim they can automate long-standing enterprise tasks.
Anthropic's blog post centers Claude and COBOL modernization
Anthropic opened its week by bragging that Claude can modernize old COBOL computer code, writing that "With AI, teams can modernize their COBOL codebase in quarters instead of years. " The post framed Claude as a solution to the shrinking pool of programmers fluent in COBOL and the difficulty companies face maintaining COBOL-based applications.
Ibm Stock reaction: a more than 13% plunge and a historic single-day drop
The grandaddy of tech companies saw its stock take a double-digit licking, plunging more than 13% after the Anthropic post. It was the biggest single-day drop for IBM's stock since October 2000, and market behavior echoed recent patterns: when AI is shown learning a new trick, investors have panicked. The piece captured that mood with the line that the "vibe coding vibe has claimed another victim. "
Why COBOL still matters to IBM mainframes and ATM networks
COBOL, or Common Business-Oriented Language, dates back to 1959 and is described in the coverage as a relic of the Doo-Wop age. It remains in use on some mainframe computers—most notably the giant machines that IBM sells—for tasks where reliability is critical. Anthropic pointed to a specific example, saying 95% of ATM transactions in the U. S. are handled by COBOL.
Maintenance challenges and IBM's existing modernization tools
With the number of programmers fluent in the old language shrinking, companies find it tough to maintain COBOL-based applications and onerous to migrate them to newer languages. The article notes IBM has long offered services and tools, including AI-based tools, to help customers modernize their COBOL applications, but the Anthropic claim nonetheless rattled investors.
Broader AI context: agents, OpenClaw buzz and consulting alliances
The coverage began by posing whether people would hand work to AI agents and noted the buzz over OpenClaw and AI agents in the tech-heavy economy of San Francisco, where a lot of coding work is being delegated to bots. Sharon Goldman is quoted explaining these agents have a lot of promise but require heavy supervision—"like babysitting a toddler"—and advised readers not to quit their day job yet.
OpenAI partnerships and the Frontier platform
Separately, the piece detailed that on Monday OpenAI announced partnerships with Boston Consulting Group, McKinsey & Co., Accenture, and Capgemini to help sell and implement OpenAI's new Frontier platform, a system that lets businesses and organizations build, deploy, supervise, and govern AI agents. The consulting firms will help clients redesign workflows; integrate AI agents with software tools and systems; help with change management; and provide industry-specific expertise OpenAI doesn’t have in-house.
Under those partnerships, which OpenAI calls Frontier Alliances, each consulting firm is investing in dedicated practice groups and building teams certified on OpenAI technology. OpenAI said its own "forward deployed engineers" will work alongside teams from the consultancies in client engagements. That paragraph was attributed to Jeremy Kahn in the context provided.
Additional references and unfinished notes in the coverage
The coverage also pointed readers to a 2023 story by Ben Weiss about the "Cobol Cowboys, " asking whether they "will they ride again?" The article closed with an observation that spending on AI infrastructure now forms a significant part of U. S. GDP growth, unclear in the provided context.
The piece's byline was Alexei Oreskovic, and the reporting captured both the technology-side claims about Claude and the immediate investor response that sent ibm stock sharply lower.
