Gen AI provider Anthropic today launched a tool that helps financial institutions modernize their archaic coding language COBOL.

AI code translation directly competes with IBM‘s legacy modernization consulting. IBM’s stock had dropped by 13% to $222 at at the end of trading today following Anthropic’s announcement.

As legacy systems built on COBOL continue to underpin critical functions in finance, government and transportation, the high cost of modernizing these decades-old applications has long been a barrier for many organizations. But AI is upending that calculus, making modernization more affordable and achievable, according to Anthropic’s release today.

COBOL, a language developed in the 1950s, still runs hundreds of billions of lines of code worldwide and handles the vast majority of ATM transactions in the United States. However, the pool of developers who deeply understand it is shrinking as people retire and fewer newcomers learn the language. This gap has historically forced organizations to hire expensive consultants to map and rewrite legacy code.

AI technologies are changing that by automating work that once required large teams and lengthy timelines. Tools such as Anthropic’s Claude Code can ingest and analyze COBOL codebases, tracing dependencies, documenting workflows and identifying risks that would take months for humans.

With AI, organizations can focus their limited human expertise on strategy and oversight while machines handle repetitive and error-prone tasks like dependency mapping and initial translation.

Register here for the inaugural FinAi Banking Summit, taking place March 2-3 in Denver. View the full event agenda here.