A global pharmaceutical company with 15,000+ employees was managing a sprawling IBM estate. At the center: a z/OS mainframe running GxP-compliant manufacturing batch processes (batch record validation, drug formulation traceability), IBM Cloud Pak for Data handling AI/ML workloads for drug discovery (protein folding simulations, clinical trial data analysis), and legacy WebSphere Application Server instances scattered across their technical estate.
When IBM's licensing team delivered the ELA renewal proposal for $22M over five years, it represented a 35% increase from the prior term. IBM cited two justifications: "expanded PVU consumption" on the z/OS mainframe and mandatory "watsonx migration rights" bundled into Cloud Pak licensing. The pharma's internal IT team had never conducted a formal ILMT (IBM License Metric Tool) sub-capacity audit. IBM knew this. They were counting on it.
The proposal sat on a director's desk with no serious pushback. Until the pharma reached out to us.
We started with data. We deployed ILMT collection scripts across their virtualized z/VM environment to establish sub-capacity eligibility and capture real peak PVU consumption over a 60-day measurement period. This is the work IBM's team should have done β but rarely does on the vendor side.
IBM's proposal was basing PVU charges on full-capacity billing: 4,200 PVUs. But ILMT data proved they qualified for sub-capacity reporting, which showed actual peak usage of just 2,600 PVUs. That's a 38% overcharge baked into the ELA.
On Cloud Pak for Data, IBM had positioned them for Watson Studio Enterprise (premium SKU at $2,400/month per resource). Our usage analysis confirmed their actual workloads fit within Watson Studio Essentials ($800/month per resource). They were paying 3x for capability they weren't using.
Then there was the watsonx "migration bundle." IBM had bundled watsonx rights into the Cloud Pak proposal without client request. This was a classic move: bundle future platform rights, justify a larger deal, lock in spending. We extracted it.
Finally, we contested the MLC (Monthly License Charge) true-up calculation. IBM's baseline had been inflated; their true-up methodology compounded the overcharge. We reset it to match audited consumption.
$1.9M over 5 years. Full-capacity to sub-capacity migration, reducing billed PVUs from 4,200 to 2,600. Sub-capacity PVU rates are ~20% lower than full-capacity. The combination of reduced PVU count plus rate reduction delivered 47% of total savings.
$900K over 5 years. Migration from Watson Studio Enterprise to Watson Studio Essentials, plus right-sizing the number of concurrent license seats. Usage audit showed they needed 12 seats of Essentials, not 24 seats of Enterprise.
$700K over 5 years. Reset MLC baseline to match sub-capacity PVU measurement, removed unsolicited watsonx bundling, and restructured true-up calculations to prevent future baseline inflation.
Total 5-Year Savings: $3.5M
Client Retention (75%): $2.625M
NoSaveNoPay Gainshare (25%): $875K
IBM will not volunteer sub-capacity eligibility. You must deploy ILMT measurement scripts and prove peak usage against your virtual infrastructure. This single lever delivered nearly 50% of savings.
IBM packages Cloud Pak with multiple SKUs (Studio, Essentials, Premium, Enterprise). Usage audits reveal that 70% of customers buy higher tiers than their workloads justify. Right-sizing alone saves 35-50% on Cloud Pak spend.
IBM's true-up methodology is complex. They use baseline inflation tactics, bundle unrelated products, and layer in penalty rates. Always require IBM to show the detailed MLC calculation; 40% of the time, errors favor IBM by 15-20%.
watsonx, z/Open Enterprise Edition, and other "future platform" bundles are negotiating leverage for IBM. Extract them. They add 8-12% to ELA cost and rarely map to actual technical roadmaps.
Pharmaceutical firms running GxP-regulated batch processing on z/OS sometimes qualify for specialized pricing. Document your validation requirements; IBM may allow lower-cost z/OS configurations for batch-only workloads.
If you're managing IBM mainframe, Cloud Pak, or any other enterprise software estate, an audit could reveal $500Kβ$5M in savings. We work on a 25% gainshare basis β you keep 75% of verified savings, or you pay nothing.
Get Free Estimate