① Journal · Workforce
Industry surveys put roughly 60 percent of mainframe professionals over 50, with a large share expected to retire within five years. The headlines are about who keeps the code running. The quieter cost is what happens to the licensing knowledge when those people leave, because that is exactly the knowledge a vendor relies on you not having.
48 hour mobilization Audit notice or renewal under 18 months out? We mobilize within 48 hours.
Get expert help →The mainframe workforce is aging in a way few other parts of enterprise IT are. Commonly cited figures put a majority of practitioners over 50, with a meaningful share planning to retire inside five years, and most universities long since dropped COBOL and z/OS from the curriculum. The operational risk is well understood: fewer people who can keep a forty year old estate running. The licensing risk is less discussed and just as expensive. The person who remembered that a product was only ever installed for a project that ended in 2014, that a particular LPAR is capped for a contractual reason, or that the last renewal carried a hand negotiated price hold across refreshes, is often the same person walking out the door. None of that lives in the contract. It lives in their head, and when they go, the estate loses its memory.
Vendors do not need to do anything aggressive to benefit. They simply price the next renewal against an estate that can no longer explain itself. Shelfware that nobody can confidently identify as unused gets renewed because cancelling it feels risky. A baseline that crept upward goes unchallenged because no one remembers what normal looked like. An uplift framed as the market rate goes unquestioned because the institutional counter argument retired last quarter.
The skills gap does not appear as a line item called skills gap. It appears as four recurring patterns, each of which quietly favors the publisher:
| What is lost | How it shows up | Who benefits |
|---|---|---|
| Who installed what, and why | Shelfware renews because cancelling feels too risky to attempt | The publisher, on support it no longer earns |
| What normal consumption looked like | A crept baseline is accepted as the new floor at renewal | The publisher, on capacity nobody planned |
| What the last deal actually won | Price holds and caps go unenforced because no one can find them | The publisher, on a reset metric |
| How the contract is structured | Audit requests are answered too broadly, expanding exposure | The publisher, on a wider finding |
Each row is a place where a few percent leaks every year, and a few percent compounding on a multi million dollar estate is not a rounding error. The estate that cannot explain itself pays a premium for that ignorance, renewal after renewal.
The defense is to convert tacit knowledge into documented fact while the people who hold it are still in the building. That means building a real inventory now, while someone can still tell you why a product is there, not after. It means snapshotting the baseline and the deal terms so the estate has a memory that survives turnover. And it means treating the licensing knowledge transfer as deliberately as the technical one, because the cost of losing it is just as real, it simply arrives on a different invoice. The practical starting point is an inventory built while the institutional memory is still available: see building your mainframe software inventory and the discipline of quarterly license governance that keeps it current. When a renewal or an audit lands and the people who knew the estate are gone, our mainframe license negotiation work rebuilds the buyer side picture from the contracts and the consumption data, so the gap in your bench does not become a gap in your leverage. For the related modernization angle, see the mainframe exit studies that never ship.
Capture what is installed and why, while someone still knows.
The estate that cannot explain itself is misread at the top too.