Every sub-capacity rule, every ILMT requirement, and every quarterly report exists to keep you on the right side of one number. Sub-capacity lets you license only the virtual cores allocated to an IBM product. Full-capacity charges as though the product runs on every physical core in the host. The discipline of sub-capacity compliance feels like overhead until you see the cost delta it protects, at which point it becomes obvious why audits target exactly this.
The worked example.
Take a single WebSphere instance using 4 virtual cores on a 32 core physical host, on processors valued at roughly 70 PVU per core.
- Sub-capacity: 4 cores times 70 PVU is 480 PVU. You license the capacity the software actually uses.
- Full-capacity: 32 cores times 70 PVU is 3,840 PVU. You license the whole host as if WebSphere ran on all of it.
That is an eight to one difference on one workload, on one host. The same software, the same deployment, priced eight times higher because the evidence to claim sub-capacity was missing. Multiply that across an estate, and the audit math that produces multi-million claims stops being mysterious.
What flips you to full-capacity.
You do not have to over-deploy to get a full-capacity bill. You only have to lose the evidence. IBM defaults the affected period to full-capacity charging when any of the sub-capacity conditions fail:
- No approved tracking tool within 90 days of the first eligible deployment.
- The tool stops running, or an agent breaks and a host goes uncovered.
- Quarterly reports are missing, not generated, or not retained for the two-year window.
- Ineligible technology, such as an unsupported older platform, sits under the workload and disqualifies sub-capacity entirely.
Each of these is an evidence failure, not a deployment failure, which is what makes the full-capacity exposure so common and so avoidable.
The lookback multiplier.
The cost delta does not apply to one moment. An audit lookback can run two to five years, and where the sub-capacity evidence is missing for that window, the full-capacity rate applies across every period in it. The eight to one delta on a single workload becomes eight to one across years of back-charges. This is why a quiet gap in one quarter of ILMT reporting, left undiscovered, can carry a finding far larger than the workload itself would ever justify.
Defending the delta.
The good news inside the bad math is that the full-capacity default is contestable. Where a workload was genuinely sub-capacity and the deployment can be reconstructed from clean evidence, the sub-capacity position can be reinstated rather than conceded. A buyer side defense recalculates the PVU correctly, restores the agent and report coverage that proves the virtual core count, and challenges every period IBM tried to price at full-capacity. The delta that works against you when the evidence is missing works for you once it is rebuilt.