Future and its affiliate partners may earn a commission when you purchase through links on our articles.
What do airlines, banks and insurance companies have in common? In addition to being a pain to deal with, they all relied on COBOL and IBM mainframe computers as their core infrastructure. However, the computing giant’s control of these markets may finally start to unravel. Anthropic announced COBOL-specific functionality for its Claude AI robot, sending IBM investors’ shares down 13%.
A closer look at TH Premium: Artificial Intelligence and Data Center
Anthropic published its thoughts in a blog post , and the company seems to know its target market well. There is a code modernization playbook available for download and an existing YouTube video of Claude Code playing around with COBOL illustrating the concept. Mixing “COBOL,” “AI robots,” and “YouTube” in the same sentence is a complete abomination of common logic…and yet here we are.
For practical purposes, COBOL ran on only one type of system and was supported by one group of people: IBM’s mainframes and their engineers. That means the company has been telling customers for decades how many zeros will be added to their hefty bills in the next period. The ruthless nature of this stranglehold meant that any attempt to break it would be unpopular with existing customers and pose a serious threat to IBM’s business.
If you’ve ever dealt with Social Security, Public Administration, Healthcare, Government, Finance, Insurance, Automotive, Retail, or Airlines, you’ve been exposed to a COBOL system at some point in your transaction, even if it’s 30 layers deep. Like gravity, language is invisible yet affects every part of the modern world.
A cynical view of the situation would say that a system running COBOL should be 100% accurate 100% of the time, a concept that does not lend itself well to the “probability of being correct” that an LL.M. can provide. Still, as I’ve proven myself, a good bot is a force multiplier for capable developers, and can also lower the barrier to entry for younger people trying to tackle legacy systems.
The language dates back to the 1960s, pitching itself as a human-readable language targeted at business transactions, using full decimal point math as the default, in contrast to the default floating point math in other languages. As it was proposed, it revolutionized business computing, became entrenched in nearly every field worth noting, and was never truly replaced.