Brief — fact and consequence

On February 23, IBM shares fell by 13% after Anthropic reported that its tool Claude Code can automate updates to programs written in COBOL, CNBC reports. The market interpreted this as a signal: a technology that was previously the prerogative of mainframes and large vendors is now available from an AI-based alternative.

What COBOL is and why it still matters

COBOL is a programming language from the late 1950s, designed for business data processing. According to Anthropic, about 95% of ATMs in the United States still run on systems that depend on COBOL. This explains why even small improvements in code automation can have a large practical effect on financial operations.

"Hundreds of billions of lines of COBOL run every day in real production environments, supporting critical systems in finance, aviation and the public sector. At the same time, the number of people who understand this language decreases every year. Artificial intelligence is ideal for optimizing tasks that made COBOL updates prohibitively expensive before."

— Anthropic (company blog)

Market reaction and risks

The drop in IBM shares is a symptom of a broader reshuffling of expectations: if AI can cheaply maintain or modernize legacy systems, the business model of companies that sold expensive mainframes and modernization services comes under pressure. At the same time, Anthropic released a feature called Claude Code Security, which scans code for vulnerabilities — this is a challenge for the cybersecurity sector, and several companies in that field have already seen share price corrections after the release.

"Modernization of legacy code stalled for years because understanding it cost more than rewriting it. Artificial intelligence changes that balance."

— Anthropic (company blog)

What this means for Ukraine

This story is not only about stock charts in the United States. For Ukraine, the key lessons are these: first, AI tools can accelerate the recovery and modernization of critical systems (banking interfaces, logistics, state registries) without the need for mass hiring of narrowly specialized experts. Second, automation increases speed but also requires new approaches to cyber protection and regulation — especially during wartime, when digital resilience has a direct connection to security.

Conclusion: balance between efficiency and risk

The market is now re-evaluating how quickly AI will change legacy infrastructure. For Ukrainian business and government, this is a signal to invest simultaneously in automation and in cyber resilience, as well as in training personnel who can work with new tools. Whether we will manage to ride this wave of innovation so that it strengthens our economic and national security — that question remains open.