The mainframe bill moves on capacity discipline, not discount asks
Most IBM mainframe software prices on Million Service Units of capacity, with Monthly License Charges following the rolling four hour average peak: a short spike can set the bill for the whole month, which is why workload timing matters more than total volume. Across the estates we advised, the bill almost always moved more from peak control than from the discount conversation everyone wanted to start with.
Prepared by Redress Compliance · August 7, 2026 · IBM advisory. Based on 15 to 25 mainframe estates advised 2024 to 2025.
Executive summary
MSU is a capacity meter, and the peak is the primary lever. You pay for the capacity the workload peaks into, not the volume it processes, and Monthly License Charges track the rolling four hour average high water mark across the measured period.
In our estates, an uncontrolled peak inflated monthly charges 15 to 30 percent, and capping defined capacity plus moving batch windows off the online peak cut MLC by the same band before any discount conversation started. Rate is the last lever, not the first.
Sub capacity reporting is the single biggest avoidable gap. The Sub Capacity Reporting Tool lets you license below the full box capacity by reporting the actual measured peak; without it, the default is full capacity, the largest avoidable cost in the mainframe budget on a large box.
The discipline is monthly submissions, deliberate defined capacity caps that limit the chargeable peak, and reconciliation of reported peaks against entitlements before every renewal.
Tailored Fit Pricing is a baseline year decision, not a model upgrade.
TFP replaces the monthly peak model with an annual consumption baseline plus a growth allowance, which helps estates with steady growth and uncontrolled peaks and hurts flat or shrinking estates that would do better capping the peak model. Signed without a baseline year analysis.
It locked in 10 to 20 percent of avoidable cost: model a low year and a high year, compare the included growth to your real trajectory, and understand the exit before committing.
The Broadcom CA stack renews on Broadcom terms, and optionality resets it.
The former CA mainframe tools arrived at renewal with 20 to 40 percent uplift asks that softened sharply once third party options were on the table, and the counterweight is credible alternatives plus selective retirement of low value tools.
The same MSU peak drives both the IBM and the Broadcom bill, so the capacity discipline pays twice, once on each vendor's meter.
The four pricing models, and the buyer lever on each
| Model | Basis | The buyer lever |
|---|---|---|
| MLC full capacity | The box MSU rating | Reduce the box, or escape to sub capacity |
| MLC sub capacity | The reported peak via SCRT | Cap the peak with defined capacity |
| Tailored Fit Pricing | An annual consumption baseline plus growth allowance | Pick the baseline year deliberately |
| One Time Charge | Perpetual license plus support | Negotiate at purchase, when leverage peaks |
The rolling four hour average is the fact the whole model turns on.
Monthly charges follow the R4HA peak, so a short spike sets the bill for the entire month regardless of what the other 719 hours did: batch workloads stacked onto the online peak are the classic self inflicted charge, and moving two batch windows off peak dropped the average, and the bill.
Before any negotiation opened.
Peak capacity, defined capacity caps, and workload placement are the three dials.
The Tailored Fit Pricing decision, three points
- The baseline year: model a low year and a high year before signing, because a baseline set in an unusually high year prices the whole term against it.
- The growth allowance: compare the included growth to your real trajectory, since growth running below the allowance is consumption paid for and never used.
- The exit terms: understand how you leave if consumption falls, before the model makes falling consumption expensive.
- The fit: TFP helps steady growth estates with uncontrolled peaks, hurts flat or shrinking estates that should cap a peak model instead, and the only way to know is running both scenarios.
The IBM mainframe MLC and IPLA negotiation brief
The pricing models, the SCRT discipline, the TFP decision, and the renewal sequence worked on a representative estate.
Get the white paper →The Broadcom CA stack, renewing on new terms
The former CA Technologies mainframe tools, systems management, security, storage, and development, now renew under Broadcom, and the asks have been steeper since the acquisition: 20 to 40 percent uplift arrived as the opening position across our renewals.
Softening sharply once third party support options and selective retirement of low value tools entered the conversation, because the wider Broadcom playbook rewards customers who arrive with alternatives.
The portfolio mechanics, the MSPS bundle, the swap rights, and the eight renewal levers, are worked in the Broadcom CA mainframe pricing analysis; the operational reduction levers on the IBM side, workload shift, zIIP offload, and the ILMT discipline, in the MSU reduction guide.
And the agreement structure above both in the IBM ELA analysis.
- Percentile standing for your exact deal size and industry, from real closed transactions
- Scenario simulation before the call: test alternative terms and see the financial impact of each
- A negotiation playbook, talking points, and a two page executive brief on day one
What we saw across mainframe estates, 2024 to 2025
Across roughly 15 to 25 mainframe estates Morten Andersen advised between 2024 and 2025, the bill almost always moved more from capacity discipline than from discount negotiation:
Capping defined capacity and retiming batch, before the discount conversation opened.
Avoidable cost locked in by consumption models signed without a baseline year analysis.
The standard advice, chase a bigger percentage discount and treat the MSU bill as fixed, is wrong in a specific and expensive way: the discount ask moved the bill far less than capacity discipline did, because a discount prices the waste while a capped peak removes it.
The sequence that worked was control the peak first, model Tailored Fit Pricing against a real baseline second, and negotiate rate last, and treating rate as the first lever left the largest savings untouched in every estate where we found it done that way.
Your first five moves
- Confirm sub capacity reporting on every box, because the default without SCRT is full capacity, the biggest avoidable cost in the budget.
- Cap defined capacity deliberately and retime batch off the online peak, the 15 to 30 percent that precedes any negotiation.
- Model TFP against a low year and a high year before signing, the analysis whose absence locked in 10 to 20 percent.
- Reconcile reported peaks to entitlements before renewal, so the baseline you negotiate from is yours, not the default.
- Arrive at Broadcom CA renewals with alternatives priced, where 20 to 40 percent asks softened. The IBM practice runs the estate with you.
Frequently asked questions
How is IBM mainframe software licensed?
Mostly on Million Service Units of capacity: MSU is a capacity meter, not a usage counter, and Monthly License Charge software bills each month against the rolling four hour average peak the workload reaches.
You pay for the capacity peaked into, which makes peak control, defined capacity caps and workload timing, the primary lever on the bill.
What is the rolling four hour average?
The measurement basis for monthly mainframe charges: the peak four hour average across the period sets the bill, so a short spike can price the entire month regardless of average utilization.
Batch workloads stacked onto the online peak are the classic self inflicted charge, and retiming them dropped bills 15 to 30 percent in our estates before any negotiation.
What is sub capacity pricing on the mainframe?
Licensing below the full box capacity by reporting the actual measured peak through the Sub Capacity Reporting Tool: without SCRT reporting the default is full capacity, the single biggest avoidable cost on a large box.
The discipline is timely monthly submissions, deliberate capacity caps, and reconciling reported peaks against entitlements before each renewal.
Should a CIO move to IBM Tailored Fit Pricing?
Only after modeling both scenarios: TFP swaps the monthly peak model for an annual consumption baseline plus growth allowance, which helps steady growth estates with uncontrolled peaks and hurts flat or shrinking ones.
Signed without a baseline year analysis it locked in 10 to 20 percent of avoidable cost, so model a low year and a high year and read the exit terms first.
How do you handle Broadcom CA mainframe renewals?
With credible optionality: the former CA tools renew on Broadcom terms, opening asks ran 20 to 40 percent above prior spend in our reviews, and they softened sharply once third party support options and selective retirement of low value tools were on the table.
The same MSU peak drives both the IBM and Broadcom bills, so capacity discipline pays on both meters.
What moves the mainframe bill most?
Capacity discipline, not discount asks: capping the rolling four hour peak and retiming batch cut MLC 15 to 30 percent before any discount conversation, because a discount prices the waste while a capped peak removes it.
The working sequence is peak control first, TFP modeled against a real baseline second, rate negotiation last.
