How an Asia Pacific bank renegotiated $18M of mainframe capacity.

This case is one of three mainframe restructures the Desk worked through in the second half of 2025. The bank is a top five domestic institution in an Asia Pacific market. The mainframe estate had been on a CA branded portfolio for fourteen years, transitioned into the Broadcom mainframe portfolio after the acquisition, and renewed twice on continuation terms. The third continuation was the one that did not happen. The renewal opened at a capacity uplift the bank could not absorb, the systems team's read of the actual workload showed something different than the seller's read of the contract, and the resulting eighteen week renegotiation produced eighteen million dollars of capacity restructure across the new term. The piece below is the arc of how that happened.

Names and identifying details are stripped throughout. The figures and the structural moves are real. This is published with the bank's permission and with a deliberate scrubbing of the operational specifics that would identify either the institution or its account team. What is left is the negotiation arc, which is the part that travels.

The opening position

The renewal opened in March 2025 with a seller quote that proposed a thirty one percent uplift on the prior contract value. The justification on the cover sheet referenced compounded escalator growth from the prior three years, a workload growth assumption built off a 2023 capacity projection, and a SKU mix that included two product additions the seller positioned as essential to keeping the portfolio current. The total annual contract value moved from forty one million to fifty four million. The five year contract value moved from two hundred and five million to two hundred and seventy million.

The bank's procurement team had three weeks of internal questions about the number before bringing it to the systems group. The systems group's first read was that the workload growth assumption did not match what they were seeing on the floor. The bank's mainframe estate had been on a declining workload trajectory for eighteen months, driven by a digital channel migration that had moved transaction volume off the mainframe and onto a hybrid cloud stack. The 2023 capacity projection the seller was anchored on did not include that migration. The contract was being priced for a workload that no longer existed.

The find

The Desk was engaged in week four of the renewal cycle. The first piece of work was a 36 month capacity profile, built from the bank's own monitoring data rather than from any seller produced number. The profile showed peak capacity utilisation declining from 89 percent of contracted capacity in 2022 to 67 percent in 2025, with the decline accelerating in the most recent six months as the digital migration completed. The contracted capacity ceiling was 38,400 MSU. The actual peak in the last six months was 25,700 MSU. The bank was paying for roughly thirteen thousand MSU of capacity that had not been used at peak in eighteen months and was not forecast to be used in the next thirty six.

The second piece of work was a structural read of the contract. The contract was on an IPLA equivalent structure with capacity bands that only moved up. There was no contractual mechanism for capacity true down at renewal, which was the seller's argument for why the renewal had to price against the existing ceiling. That argument was contractually correct on the existing contract. It was not relevant to the renewal, which by definition was a new contract and could be structured against the actual workload rather than against the historical ceiling.

"The seller was pricing thirty eight thousand MSU. The bank was running twenty five thousand at peak. The other thirteen thousand had been gone for eighteen months. Nobody had asked the question because the contract did not require the question to be asked."Mainframe Lead, The Desk

The restructure

The negotiation reframed the renewal around three structural moves. First, the contracted capacity ceiling was rebuilt against the actual workload profile, with the new ceiling at 28,500 MSU plus a 12 percent headroom band that could expand into a defined upper ceiling at 32,000 MSU if growth returned. Second, the SKU mix was rebuilt against the active module set on the mainframe, removing two SKUs that had been carried forward from prior renewals but were no longer in operational use, and rationalising three more that had been duplicated through portfolio reorganisations. Third, the escalator structure was rewritten with year over year caps tied to a published index rather than to a seller defined escalator.

The seller's first response to the capacity restructure was to refuse it. The argument was that the seller's deal desk could not approve a capacity reduction of this magnitude inside an active renewal cycle. The bank's response was to produce the workload profile, the monitoring data, the digital migration roadmap, and a documented willingness to take the renewal off the seller's table if the structure could not be addressed. The seller's second response, six weeks later, was a counter proposal at 31,000 MSU with a stepped reduction to 28,500 over the first eighteen months of the new contract. The bank accepted the stepped structure as a workable compromise.

The outcome

The renewal closed in week eighteen at an annual contract value of thirty six million, a five million dollar reduction against the prior contract and an eighteen million dollar reduction against the seller's opening quote. Across the five year term the capacity restructure produced ninety million in headline reduction against the seller's opening, of which roughly eighteen million was the immediate capacity rationalisation and the balance was the compounding effect of the rebuilt escalator structure and the SKU mix correction. The bank also gained an annual capacity review clause that allowed for further true down at year two and year four against a documented workload profile.

What the case shows about the seller's position

The seller's deal desk could approve the capacity reduction. The first refusal was a negotiation position, not a contractual or procedural ceiling. The buyer's willingness to walk, combined with the evidentiary weight of the workload profile and the digital migration roadmap, produced an internal approval process inside Broadcom that the seller's account team had said was not available. This pattern is consistent across the three mainframe restructures we worked in the second half of 2025. The seller's first answer is that the structural move is not approvable. The seller's second answer, after evidence and a credible alternative, is meaningfully different. The buyers who get the second answer are the ones who organise the evidence and the credible alternative before opening the conversation.

What the case shows about the buyer's position

The bank's win was not negotiation skill in the conventional sense. The win was preparation. The systems team produced a workload profile the seller could not contest. The procurement team protected an eighteen week negotiation window rather than collapsing into a four week procurement sprint. The executive sponsor was briefed at week four and signed off on the credible alternative scenario as a real possibility, not as a tactic. None of those moves are negotiation theatre. They are the structural preconditions for a negotiation that actually moves the contract. Banks that try to produce the same outcome in four weeks, against a seller calendar, almost always settle for continuation.

The takeaway

• The bank's renewal moved by eighteen million annual and ninety million across the term because the workload profile and the contract no longer matched. The structural mismatch was the entire negotiation. Commercial concession came later, on top of structure, not instead of it.
• The seller's first answer was that the capacity restructure could not be approved inside an active renewal cycle. That answer was not a ceiling. It was a position. Evidence and a credible alternative produced a different answer six weeks later.
• Preparation produced the outcome. A workload profile the seller could not contest. A protected negotiation window. An executive sponsor briefed on the credible alternative as a real possibility. These are the preconditions, not the negotiation itself.