Claude Code's COBOL Skills Cost IBM Dearly: A $26 Billion Collapse
Originally published in Turkish. Read the original →
Anthropic's coding engine Claude Code has gained the ability to read and modernize COBOL, the 66-year-old veteran programming language. The news rattled the shares of IBM, which has earned billions of dollars for years maintaining legacy systems, and wiped billions of dollars off its value overnight.
The AI revolution is not only speeding up the development of new applications; it is also starting to seep into the oldest "fortresses" of the software industry. Claude Code, Anthropic's AI coding engine, which recently drew attention for the security vulnerabilities it uncovered, has now set its sights on enterprise infrastructure inherited from the 1950s.
According to Anthropic's latest announcement, Claude Code has gained the ability to successfully read and analyze systems written in COBOL (Common Business-Oriented Language), one of the oldest and perhaps hardest-to-replace programming languages in the world. What's more, it can convert these massive chunks of legacy code into a format that can be modernized. While this news is exciting for developers, it turned into a small-scale nightmare for one of the tech giants.
Why Is COBOL So Important?
With a history of a full 66 years, COBOL still carries a huge share of the backend systems of banking, aviation, insurance and government institutions today. Moving a COBOL system to a modern architecture (for example, a cloud-based Java or Go architecture) is one of the most painful processes in the software world.
Modernizing legacy systems requires hundreds of expert developers coming together to answer the question "what has been coded here", years of workforce planning and enormous consulting budgets. This is where Claude Code promises to bring incredible speed to years of effort by almost completely automating the "code discovery and business logic analysis" phases of these massive migration projects.
IBM's $26 Billion Loss
When it comes to COBOL and legacy system modernization, the first enterprise player that comes to mind is undoubtedly IBM. Having provided the mainframe infrastructure of large companies for decades, IBM also earns enormous revenue from maintaining and consulting on these COBOL systems.
The expectation that Claude Code would shorten this complex and lucrative modernization process caused major panic among investors. IBM shares lost 13% of their value in a single night after the announcement, which meant $26 billion wiped off the company's market value. According to analysts, this sharp drop went down on record as IBM's biggest overnight plunge since 2000 and its most dramatic monthly collapse since 1968.
IBM's Response: "Modernization Is Not Just Translating Code"
Following this violent earthquake in the markets, statements quickly came from IBM as well. To calm investors, company officials emphasized that their services consist of much more than simply translating old code into a new language.
According to IBM's argument, enterprise infrastructure transformations go far beyond code translation and involve complex engineering processes such as platform performance, strict security protocols and hardware compatibility. The company argued that AI can only analyze code, and that the integrated quality offered by IBM Mainframe systems, built on years of experience and requiring high reliability, cannot be matched by a language model alone.
IBM, a seasoned player in the tech world, still continues to sell advanced "mainframe" packages to giant customers seeking high operational security and uninterrupted operation (uptime). However, this shocking development on the Claude Code front proves that AI can not only produce next-generation tools but also seriously threaten companies' "most profitable and untouchable" lines of business.
