An Asia Pacific bank renegotiated $18M of mainframe MIPS capacity.

The bank runs a mainframe estate that sits at the centre of its retail and payments processing. The capacity is approximately fourteen thousand MIPS across two production sysplexes, with peak utilisation driven by a quarterly batch window that lasts for between two and three hours every ninety days. The rest of the quarter, the workload utilisation runs at roughly forty percent of peak. The Broadcom mainframe software contract that was up for renewal billed against peak MIPS, which meant the bank was paying for the quarterly peak as if it were the steady state. The renewal trajectory put the cost across the new term at fifty four million dollars. The bank's head of mainframe engineering had been raising the metric issue for two renewal cycles. This was the cycle the procurement function brought the metric issue into the negotiation as a structural item.

The engagement ran six months. The signed contract closed at thirty six million dollars across the new term, with the billing metric converted from peak MIPS to monthly licensed capacity using the MSU based model that the contract had permitted as an option for years but had never been switched onto. The eighteen million dollar saving against the trajectory was the visible outcome. The structural outcome was that the contract was now denominated in the metric the bank actually wanted to manage to, which made the next renewal a different conversation.

The Quote

The trajectory cost of fifty four million dollars was built on three assumptions. First, the billing metric would continue to be peak MIPS as it had been across the prior contract. Second, the capacity tier would be set against the prior peak observation, which was the quarterly batch window. Third, the contract would carry forward the existing mainframe software product mix, including two products the bank had reduced its dependence on over the prior two years but had not formally retired.

None of the three assumptions was wrong on the seller's terms. All three were the position the seller would default to if the buyer side did not propose alternatives. The mainframe procurement lead at the bank had been told for two cycles that the metric was structural and could not be changed. That was not true. The contract permitted an MSU based metric as an alternative. The option had never been formally invoked.

The Find

The work to surface the MSU option was contractual archaeology. The original mainframe software agreement, signed many years before, contained an optional metric clause that permitted the buyer to elect MSU based licensing at any renewal. The clause had been preserved across two renewals since the Broadcom acquisition. The clause had not been exercised because neither the prior procurement leads nor the prior account teams had raised it. The clause was on the buyer side. The buyer side had not used it.

"The mainframe contract had the right answer inside it the entire time. The buyer side had to read the contract carefully enough to find it. Account teams will not raise an option that costs the seller money."Lead mainframe advisor

The second finding was on the capacity tier. The MSU based metric the contract permitted was tier banded. The bands were structured such that a small capacity reduction at the right tier boundary produced a disproportionate cost reduction. The bank's quarterly batch peak sat just above a tier boundary. The engineering team confirmed that the batch window could be reshaped using existing scheduling tooling to bring the peak inside the lower tier. The reshape required a six week implementation but was not technically demanding.

The third finding was on the product mix. Two mainframe software products that had been in the contract for more than a decade were no longer in active use. The bank had migrated their function onto alternative tooling over the prior eighteen months. The maintenance was still being paid. The product retirement could be incorporated into the renewal as a formal scope reduction.

The Restructure

The restructure proposal was tabled in writing. It made four asks. First, the billing metric converted from peak MIPS to monthly licensed capacity on the MSU based model, invoking the optional clause in the master agreement. Second, the capacity tier set at the band immediately below the prior peak, contingent on the bank completing the batch reshape inside the first quarter of the new term. Third, the two retired products formally removed from the contract with a documented scope reduction. Fourth, the contract term extended by twelve months in exchange for a price hold on the converted metric.

The account team's response on the metric conversion was that the conversion was permitted but the resulting tier and pricing needed to be modelled. The modelling took six weeks and was shared between Broadcom and the bank with the Desk reconciling the two models. The two models converged at a number close to the bank's working budget. The capacity tier was agreed in the same conversation, conditional on the batch reshape. The product retirement was agreed inside two weeks. The term extension and price hold took the final round.

The signed contract closed at thirty six million dollars across the new term, with the conversion executed and the batch reshape completed inside the first quarter as committed. The savings were realised across each year of the term and compounded forward into the next renewal cycle, where the buyer side started from the converted metric rather than from peak MIPS.

The Outcome

The eighteen million dollar saving was the headline number. The more durable outcome was the structural shift in how the contract was denominated. The bank now manages mainframe capacity to a metric that aligns with how the workloads actually run, which means the engineering decisions and the financial outcomes are now in the same units. The retired products are gone from the contract. The batch reshape is in production. The next renewal will be structured around capacity choices the bank controls, not around a peak observation that happened to occur during a single quarterly window.

The pattern repeats across the mainframe engagements we have worked on. The Broadcom mainframe software contracts often contain optional clauses that benefit the buyer if invoked. The clauses are not raised by the account team. The buyer side has to find them, read them carefully, and elect them at renewal. The clauses do not stay in scope indefinitely. They tend to disappear from successor contracts when not exercised. The renewal cycle is the moment to exercise them.

The takeaway

• Mainframe software contracts often contain optional billing metrics the buyer side has never elected. Reading the master agreement for those options is the highest value early work on any renewal.
• Capacity tier banding makes small workload reshapes disproportionately valuable. A batch window moved across a band boundary can produce a multi million dollar saving compounding annually.
• Product retirement should be incorporated into the renewal, not handled as a separate maintenance cancellation. The renewal is the moment when scope reductions translate into pricing concessions.