Microsoft Corp. ended fiscal 2026 with a record $678 billion in commercial remaining performance obligation (RPO), an 84% jump from the previous year, signaling robust demand for its cloud and AI services. However, comments from CFO Amy Hood during the earnings call suggest that a significant portion of this growth is concentrated with a single customer, OpenAI. This concentration raises questions about dependency risks as Microsoft seeks to broaden its AI ecosystem beyond its primary partner.

Hood disclosed that commercial RPO increased 25% when excluding OpenAI. Based on Microsoft’s prior-year disclosure of $368 billion in commercial RPO for fiscal 2025, this implies that non-OpenAI backlog grew to approximately $460 billion. The difference between the total reported $678 billion and the estimated $460 billion suggests that OpenAI-related commitments could account for roughly $218 billion, or about 32% of the total commercial backlog.

The Math Behind Microsoft's OpenAI Exposure

The estimation relies on the assumption that OpenAI was a negligible part of the backlog a year ago. If OpenAI had a material presence in the prior year’s backlog, the non-OpenAI base would be smaller, potentially making OpenAI’s share larger than 32%. Microsoft did not explicitly disclose OpenAI’s specific backlog figure, but the derived estimate highlights the extraordinary scale of the partnership.

Diversification Beyond OpenAI

Despite the heavy reliance on OpenAI, CEO Satya Nadella emphasized that Microsoft’s AI strategy extends well beyond this single partnership. Nadella stated that Azure now offers more than 11,000 AI models, including offerings from OpenAI, Anthropic, Mistral, xAI, and Microsoft’s own MAI family. He noted that the number of customers building applications with models from multiple providers has increased fivefold since the start of the year, underscoring a push toward a model-agnostic AI platform.

What the Numbers Show

The data reveals a divergence between Microsoft’s strategic messaging and its financial reality. While management highlights a diversified portfolio of AI partners and a fivefold increase in multi-model customers, the backlog composition suggests that OpenAI remains the dominant growth engine. With commercial bookings increasing 18% year over year excluding OpenAI, yet RPO growing 25% on the same basis, the outsized contribution of the AI startup continues to drive the top-line pipeline. Investors should monitor whether the diversification narrative translates into reduced concentration risk in future quarters.

This implied $218 billion figure is based on investor calculations rather than a direct company disclosure. Nevertheless, it offers a fresh perspective on the scale of Microsoft’s AI partnership and why investors continue to watch the relationship between Microsoft and OpenAI as closely as Azure’s headline growth metrics.

Microsoft Corp. delivered robust fiscal fourth-quarter results, reporting a 43% year-over-year surge in Azure revenue as enterprises increasingly adopt its multi-model artificial intelligence architecture. CEO Satya Nadella signaled a strategic pivot away from exclusive dependence on OpenAI, declaring that "every model is substitutable" within Microsoft’s cloud infrastructure. This approach aims to enhance business continuity and resilience by allowing customers to mix and match AI providers based on cost, performance, and specific use cases. The company’s broader Microsoft Cloud segment generated $214.4 billion in revenue for the fiscal year, while commercial remaining performance obligations rose to $678 billion, underscoring sustained enterprise demand.

The earnings call highlighted Microsoft’s effort to decouple its software layer from underlying AI models, enabling seamless switching between providers. CFO Amy Hood explained that keeping the "harness separate from the model" ensures any given model is swappable at any time. This structural change reduces vendor lock-in risks for clients and positions Microsoft as an agnostic infrastructure layer rather than a partner tied to a single AI developer. Nadella noted that customer adoption of this multi-provider strategy is accelerating, with a fivefold increase in the number of customers building with models from multiple providers.

Azure now hosts over 11,000 models, expanding beyond OpenAI to include offerings from Anthropic, Mistral, xAI, and Microsoft’s own MAI family. This diverse catalog allows enterprises to optimize their AI stacks dynamically. Levi Strauss & Co. serves as a key example of this trend, utilizing both OpenAI and Anthropic models through Microsoft’s Foundry platform to deploy more than 1,000 domain-specific AI agents. The ability to integrate multiple models into a single workflow represents a significant evolution in how businesses implement generative AI solutions.

Financial Performance Highlights

Strategic Implications for Investors

The emphasis on model substitutability marks a departure from Microsoft’s earlier narrative, which was closely tied to its partnership with OpenAI. By positioning itself as the infrastructure host for competing AI models, Microsoft mitigates the risk associated with any single provider’s technological or market setbacks. This strategy supports long-term revenue stability by embedding deeper into enterprise workflows regardless of which AI model leads in performance. The strong growth in Azure and rising remaining performance obligations suggest that this diversified approach is resonating with corporate buyers seeking flexibility and resilience in their AI investments.