Start here
A short overview of the whole landscape this site covers: the languages, the people who still run them, the vendors who modernize them, and the terms you'll run into along the way. Everything here links to somewhere deeper.
All eight are still in production somewhere. What differs is the buyer: mainframe banking and government for most of them, mid-market IT for a couple, and scientific computing for one that doesn't belong in the same modernization conversation at all.
| Language | Runs on | Reference entries |
|---|---|---|
| COBOL | The batch-processing backbone of global banking and insurance. Still compiling, still running the overnight run. | 73 |
| RPG (IBM i / AS/400) | IBM i shops running RPG II through RPG IV and ILE, with a distinct dialect history under one green-screen roof. | 72 |
| PL/I | IBM's systems-and-business hybrid language, still quietly load-bearing today. | 70 |
| Assembly / HLASM | The layer everything else on the mainframe ultimately compiles down to. | 47 |
| PowerBuilder | Client/server-era 4GL still running mid-market line-of-business apps. | 37 |
| Delphi / Object Pascal | Object Pascal's most durable home, with a small and loyal audience still shipping products on it. | 75 |
| ColdFusion | A platform-decline story rather than a workforce-shortage one. Worth evaluating on its own terms. | 38 |
| Fortran | Still the workhorse of HPC and scientific codebases, serving a different buyer entirely. | 54 |
The global mainframe modernization market alone is estimated at roughly 9 billion dollars in 2026, projected to reach 13.34 billion by 2030. That number exists because two things are true at once: the systems still run the transactions (COBOL alone handles a large share of banking and card processing worldwide), and the people who can safely maintain them are retiring faster than anyone is replacing them. Neither problem is solved by pretending the other one doesn't exist, which is why “just rewrite it” and “just hire more people” are both, individually, incomplete answers.
Rewrite / convert
Translate or rebuild the application in a new language, automated tooling plus review, or a full manual rewrite. Highest cost and risk, but ends the dependency on the old language entirely.
Wrap / expose
Leave the existing code alone; build APIs or interfaces around it so new applications can call into it. Fastest, lowest-risk, but the old code (and its risks) stays exactly where it was.
Managed support
Keep the current platform current: stay on the latest compiler or IDE release, patch security issues, don't otherwise touch it. The right call when the system isn't the problem, the surrounding tooling is.
Staffing
The system and the plan for it are both fine; you just don't have enough people who can safely touch it. Solved with people, not a project.
We track 26 vendors across these four approaches. Find which ones cover your language →