Will AI replace mainframe programmers?
While legacy COBOL systems are increasingly being managed by AI for maintenance, the scarcity of human experts keeps the role relevant for critical financial infrastructure. However, the long-term risk is high as AI facilitates full migration to modern languages.
Why AI struggles to replace this job
- Legacy systems often lack documentation, requiring 'detective work' AI cannot yet perform perfectly.
- Modifying live banking systems requires a level of risk-aversion and accountability AI cannot provide.
- Understanding the business logic embedded in 40-year-old code requires historical context.
- AI struggles with proprietary or highly customized mainframe environments that lack training data.
Tasks AI could automate
- Translating legacy COBOL code into modern Java or Python equivalents.
- Identifying syntax errors and deprecated functions in JCL scripts.
- Generating test cases for batch processing routines.
- Monitoring system logs for hardware interrupts or storage overflows.
The 10-year outlook
The job market will shrink as institutions finally retire legacy hardware. However, remaining experts will command high 'emergency' rates for maintaining critical systems that refuse to die.
Common questions
Will AI replace mainframe programmers?
While legacy COBOL systems are increasingly being managed by AI for maintenance, the scarcity of human experts keeps the role relevant for critical financial infrastructure. However, the long-term risk is high as AI facilitates full migration to modern languages.
What is the AI replacement risk for mainframe programmers?
Mainframe Programmer scores 55/100 — Parts of this job will change — adaptation matters. Roughly 75% of the tasks in this role could be automated with current and near-future AI.
How much do mainframe programmers earn?
The US median salary for a mainframe programmer is about $95,000 per year, with projected employment growth of -5% over the next decade (declining).