① Licensing · Mainframe TCO
Most mainframe cost conversations start at the hardware and stop too soon. The real model is dominated by recurring software and support, and those are the most negotiable lines of all. Build the full model and you stop reacting to a single quoted number and start seeing where the cost is actually made, which is the same place it can be unmade.
Software runs the model, not iron.
Mainframe total cost of ownership is the full recurring and capital cost of running the platform: hardware, the software stack and its support, specialty engines, people, facilities, and disaster recovery. The single fact buyers most often miss is that across most large estates the recurring software bill is the largest part of the model, not the hardware. The machine is a periodic capital purchase; the software and its annual support are a continuous operating cost that compounds for the life of the estate.
That matters because the negotiable lines and the large lines are the same lines. Hardware is mostly fixed once chosen. The recurring Monthly License Charge software, the IPLA products, the third party ISV stack, and the annual support and subscription are larger and more movable, yet they are routinely treated as fixed. A real cost model puts every line on one page so the movable money is visible.
Every line, on one page.
A real mainframe TCO model has more lines than most budgets show, and the recurring ones dominate. Here is the full set, with an illustrative share of annual spend for a typical large estate. The shares vary widely by estate; what holds is the shape, software heavy and recurring.
| Cost line | Type | Illustrative share |
|---|---|---|
| IBM MLC software (z/OS, CICS, Db2, IMS, MQ) | Recurring | ~30% |
| Third party ISV software and support | Recurring | ~22% |
| IBM IPLA software and S&S | Recurring | ~12% |
| Hardware (amortized capital) | Periodic | ~14% |
| People (operations, systems programming) | Recurring | ~12% |
| Disaster recovery and continuity | Recurring | ~6% |
| Facilities, power, network | Recurring | ~4% |
| Total recurring software and support | Recurring | ~64% |
Illustrative shares for shape only, not a benchmark for your estate. The point is the concentration of cost in recurring software and support.
Read the bottom line. On this illustrative estate roughly two thirds of annual mainframe spend is recurring software and support, and the hardware everyone fixates on is a seventh of the total. The negotiation leverage follows the money: the lines you can actually move, the MLC peak, the ISV stack, the support renewals, are the same lines that dominate the model. A buyer who arrives at a renewal with only the hardware number in mind is arguing about the wrong seventh of the bill.
Where the model hides money.
Three lines are consistently underweighted, and each is both large and movable, which is the worst combination to leave unexamined.
The recurring support and subscription line is treated as a fixed tax on owning the software, when it is one of the more negotiable elements of a renewal and the thing you are paying for, continued use and updates, is worth re testing against what the estate actually consumes. The ISV stack priced on MSU or capacity grows quietly with the machine and accumulates products nobody has reviewed for years, so redundancy and overlap sit unbilled in plain sight. And the cost of carrying licensed but unused capacity, the gap between what is contracted and what is consumed, is invisible on a budget line yet real every month. None of these show up if the model stops at hardware. All of them show up the moment it does not.
How to use the model.
Turning a TCO model into saved cost
Across 500+ engagements and $180M+ of negotiated mainframe spend, the estates that reduced TCO most were not the ones that changed platform. They were the ones that built the full model, found that two thirds of the cost was recurring and movable, and worked the movable lines. The platform earned its place. The avoidable cost wrapped around it did not. For the engagement that does this end to end, see mainframe cost optimization.
Software, not hardware. Across most large estates the recurring software bill, IBM MLC software, IBM IPLA products, and third party ISV products, together with their annual support, is the largest part of TCO. The hardware is a periodic capital cost; the software is a continuous operating cost that compounds over the life of the estate.
Most commonly the recurring support and subscription line, the ISV stack priced on capacity, and the cost of carrying licensed but unused capacity. Hardware and headcount are visible and budgeted; the recurring license and support lines are larger and more negotiable, yet often treated as fixed when they are the most movable part of the model.
Because you cannot negotiate what you have not measured. A real cost model shows which lines are large, recurring, and tied to capacity you could optimize. The renewal moves when the buyer arrives with the full model and can show the vendor exactly where the number is built, rather than reacting to a single quoted figure.
Frequently and substantially. Optimizing the billable MSU peak, reclaiming unclaimed pricing treatments, retiring redundant ISV products, right sizing support, and removing licensed but unused capacity all cut TCO without touching the workload the business depends on. The platform stays; the avoidable cost around it does not.
Audit notice or renewal under 18 months out? We mobilize within 48 hours.
Get expert help →