American banks, government agencies, and financial institutions are scrambling to recruit engineers fluent in Common Business-Oriented Language (COBOL), a 67-year-old programming language that quietly underpins trillions of dollars in daily global transactions while its last generation of expert practitioners edges toward retirement.
COBOL processes approximately USD 3 trillion in daily financial transactions and powers 95 percent of ATM transactions in the United States, yet the average developer who maintains these systems is now 55 years old, with around 10 percent retiring every year. That combination of systemic dependence and accelerating workforce exit has created one of the most unusual labour shortages in modern finance.
The problem is structural. More than 85 percent of universities dropped COBOL from their curriculum since the 1990s, leaving a generation of developers trained almost entirely in Python, Java, and JavaScript, none of which interface naturally with the legacy mainframe architectures still running at the core of global banking.
Demand has pushed compensation to levels that reflect the skill’s scarcity rather than its age. Mainframe COBOL developers now earn an average of USD 125,525 per year, with demand projected to grow 15 percent over the next decade as institutions race to maintain systems they cannot simply switch off. Hourly contracting rates on platforms such as ZipRecruiter currently average USD 55.52, with experienced specialists commanding significantly more in regulated sectors.
The stakes extend well beyond private banking. COBOL remains embedded in the US Treasury, the Internal Revenue Service, the Social Security Administration, and the Department of Veterans Affairs, where even minor errors during any migration could cascade into service disruptions affecting millions of people.
The retirement of the last generation of mainframe experts exposes mission-critical systems in global banking, insurance, and government to unacceptable levels of operational risk. Some institutions have already begun rehiring retired engineers on consulting contracts, specifically because those individuals carry institutional knowledge of systems built before modern documentation standards existed.
Artificial intelligence has entered the conversation as a partial fix. Banks are testing AI tools capable of reading, annotating, and translating COBOL codebases into modern languages. But because these systems process legally sensitive financial transactions in real time, businesses prefer to modernize rather than completely replace, and full automation without human oversight remains too risky for most institutions to accept.
The result is a paradox that few technology strategists anticipated: in an industry consumed by cloud migration and generative AI, one of the most valuable technical skills a developer can hold today was designed the same year NASA was founded.
