San Francisco financial institution resets Azure OpenAI agreement.

The client is a San Francisco financial institution with a multi billion dollar enterprise IT estate spanning consumer banking, wealth management, and capital markets. The institution had run an active Microsoft Enterprise Agreement for the prior nine years. The Azure consumption baseline carried a Microsoft Azure Consumption Commitment that anchored the cloud spend through the multi year term. The 2025 cycle introduced an Azure OpenAI agreement on top of the MACC framework that bundled OpenAI workload commitments with the broader Azure commitment.

This case study describes how the workload audit, the MACC reset, and the contract carve out framework reset the Azure OpenAI agreement on a footing that preserved strategic flexibility and cut the projected spend across the term. The engagement followed the buyer side framework Redress applies on every Azure agreement. For the broader picture, read the Microsoft services overview and the Azure MACC negotiation guide.

Client background

The institution operates across consumer banking, wealth management, and capital markets. The IT estate spans a significant Azure footprint with a strong Power Platform deployment, a Microsoft 365 enterprise base, and an Azure Active Directory Premium tenancy at the parent corporate level. The Azure consumption baseline ran across a mature data platform, a regulated workload set, and a series of pilot AI workloads. The institution had not yet committed to a multi year Azure OpenAI position before the publisher's commercial outreach opened.

The Microsoft account team had run a strong relationship with the institution through the prior MACC cycle. The 2025 outreach moved the conversation from infrastructure consumption to AI commitment, and from a measured cloud spend curve to a multi year AI dollar commitment.

The trigger event

The trigger event was the publisher's 2025 Azure OpenAI agreement proposal. The proposal carried a multi year dollar commitment for OpenAI workloads, a series of regional capacity reservations, and an integration line for the Microsoft 365 Copilot deployment. The Azure OpenAI commitment was framed inside the MACC envelope. The bundled framing meant the institution would absorb the AI commitment through the same renewal cycle as the broader Azure consumption, with the same multi year stacking clause.

The chief technology officer flagged the proposal for advisory review before the procurement cycle could close. The defense engagement opened inside the proposal window, before the publisher could move to a signed commitment.

The commercial challenge

The commercial challenge had three layers.

1. The AI consumption baseline did not yet exist. The institution had run pilot OpenAI workloads but had not yet established a production deployment that could anchor a multi year commitment.
2. Cross border data residency limited the region selection. The regulated workload set across capital markets carried a series of cross border data residency requirements that limited the Azure region selection for the OpenAI workloads.
3. The broader AI strategy had not aligned on Microsoft as the only platform. The pilot evaluation cycle had also engaged Anthropic, Google, and a series of smaller model providers.

The combination meant the Microsoft proposal anchored a multi year commitment to a strategy the institution had not yet confirmed. The publisher's framing assumed the strategy was confirmed and the only open question was the dollar commitment. The buyer needed to reset the framing without disrupting the broader Microsoft relationship.

The Redress approach

The Redress approach followed the buyer side AI platform framework. The framework runs three workstreams in parallel: the workload audit that maps the active and planned AI deployments, the MACC reset that decouples the AI commitment from the broader Azure consumption, and the contract carve out framework that preserves the buyer's strategic flexibility across vendors. The advisory team holds the proposal hold against the publisher through the engagement window. The publisher receives one structured response, not a series of point answers that build the publisher's commitment position. Read the AI contract renewal strategy page for the cross vendor framework.

Workload audit

The workload audit ran across the institution's active and planned AI deployments. The audit identified three workload categories.

1. Production candidate workloads. Azure OpenAI was the most appropriate choice given the existing Azure footprint and the regulated workload constraints.
2. Multi vendor workloads. The Anthropic Claude or Google Gemini family carried capability advantages that justified a parallel deployment.
3. Pilot workloads. Model selection was still open and a multi year commitment would prematurely close the strategic optionality.

The combined audit produced a workload map that anchored the Azure OpenAI commitment to the first category alone. The second and third categories sat outside the Microsoft commitment, preserving the institution's optionality across the broader AI platform landscape.

MACC reset

The MACC reset moved the Azure OpenAI commitment outside the broader MACC envelope. The reset framing preserved the existing MACC for the infrastructure and data platform consumption. The OpenAI commitment moved to a separate line item with a smaller dollar value and a shorter term commitment. The reset preserved the publisher's commercial recognition of the broader Microsoft relationship while limiting the buyer's exposure on the AI commitment to the production candidate workloads identified in the audit.

Read the Azure MACC negotiation guide for the broader MACC framework.

Contract carve outs

The contract carve out framework added three protections to the agreement.

1. Multi vendor carve out. Preserved the institution's right to deploy Anthropic, Google, and other AI platforms outside the Microsoft commitment.
2. Regional capacity carve out. Aligned the OpenAI capacity reservations to the regulated workload boundaries.
3. Renewal optionality clause. Preserved the institution's right to reset the commitment at the end of year one or year two of the term.

The combined carve outs gave the institution the strategic flexibility the original proposal had compressed.

Outcome

The outcome of the engagement was an Azure OpenAI agreement that anchored a meaningful Microsoft AI commitment without compromising the institution's broader AI platform strategy. The commitment dollar value sat well below the publisher's first proposal. The carve outs preserved the optionality for Anthropic, Google, and other AI platforms. The MACC reset held the broader cloud consumption on the existing trajectory without the AI uplift. The publisher closed the deal cycle on a footing that recognized the institution's strategic flexibility, and the broader Microsoft relationship continued on the same trajectory through the next MACC renewal cycle.

Takeaway for the buyer

The San Francisco engagement is a representative buyer side defense on the Azure OpenAI agreement framing. The workload audit, the MACC reset, and the contract carve out framework are the three workstreams every buyer should run on receipt of a Microsoft AI agreement proposal. Buyers who run the framework before the formal commitment cycle closes preserve the strategic flexibility across the broader AI platform landscape. Buyers who accept the publisher's bundled framing on the first proposal lock in a multi year commitment to a strategy that the institution may not yet have confirmed.

To open an Azure OpenAI engagement, book a scoping call. To read the broader framework, download the AI Platform Contract Negotiation Playbook or read the Microsoft knowledge hub.