Microsoft disclosed $3.2B tied to its Anthropic stake while characterizing OpenAI’s relationship as more uneven, according to its fiscal 2026 reporting

Microsoft’s latest earnings communications offered a fresh window into how it is managing its two biggest, competing bets in generative AI, Anthropic and OpenAI. Reporting for its fiscal 2026 year, which ended June 30, Microsoft said it recorded $3.2 billion tied to its investment in Anthropic, a detail that gives investors another concrete data point on the economics of its model partnerships.

In the same set of disclosures, Microsoft’s accounting and commentary also suggested a more mixed outcome from its OpenAI exposure, contrasting with the straightforward magnitude it attached to the Anthropic investment. The company did not frame the takeaway as a simple win-lose story, instead implying that returns from different AI collaborations have not lined up evenly.

The juxtaposition matters because Microsoft’s cloud business and its Azure platform increasingly rely on generative AI workloads that are built on underlying foundation models. When customers buy AI capabilities, they are buying access to model performance, reliability, and cost structure, not just a software interface. That makes Microsoft’s investment and partnership economics a direct driver of margin and roadmap decisions.

Microsoft’s disclosures come as AI labs and model providers compete to lock in enterprise and cloud distribution. Anthropic and OpenAI are both in the center of that race, but they operate with different commercial terms, licensing structures, and technology roadmaps. Microsoft’s reported numbers and its “mixed bag” framing indicate that the financial mechanics vary across those arrangements.

For Microsoft, the practical question is how investments translate into usable capacity and product value across its stack, from copilots and developer tooling to cloud-hosted AI services. Microsoft has repeatedly positioned Azure as the principal route by which enterprises access AI models, which means partner economics can influence pricing, scaling, and product timetables even when the technical model is largely supplied by others.

What Microsoft did not provide in the posted report is additional breakdowns that would let outsiders fully reconcile the $3.2 billion figure to specific components like timing, valuation methods, or how much of the amount reflects gains, impairments, or other accounting categories. Without those details in the publicly shared excerpt, readers are left with the headline magnitude and the directional message that Anthropic investment economics have been more favorable than its OpenAI experience.

Going forward, investors will likely watch for continued disclosure in subsequent quarters: whether Microsoft reiterates the Anthropic investment figure, whether it quantifies how OpenAI-related accounting trends are evolving, and how those narratives map to Azure consumption trends tied to generative AI. Any additional clarity could also announcement how Microsoft is prioritizing spend and partnership structure as model providers continue to differentiate on performance and deployment options.