Microsoft recorded $24.1 billion in sales attributable to OpenAI during the fiscal year ended in June 2026, according to disclosures reported by Bloomberg — a figure that amounts to roughly 70% of Microsoft’s total AI revenue for the year. The disclosure has reframed how analysts view Microsoft’s AI business, revealing that the vast majority of what the company books as AI revenue traces back to a single customer: its own strategic partner and largest external AI supplier.
How the Money Actually Flows
Under the terms of Microsoft and OpenAI’s long-running partnership, OpenAI pays Microsoft for Azure computing power, costs tied to training and running its models, and a share of its own revenue. Much of that $24.1 billion, according to the Bloomberg reporting and follow-on analysis from Wheresyoured.at, is effectively OpenAI’s own cloud compute bill being routed through Azure and then booked by Microsoft as AI income. In other words, a large chunk of Microsoft’s marquee AI revenue figure represents money moving from one side of a joint venture-like arrangement to the other, rather than new AI sales to unrelated third parties.
The Scale of Microsoft’s AI Business
Bloomberg estimated Microsoft’s overall AI business generated approximately $34 billion in the twelve months through June 2026. Microsoft also disclosed it closed the fiscal year with 30 million Copilot seats sold across its enterprise customer base and that Azure revenue grew 43% year-over-year, according to the company’s own reporting cited in coverage of its fiscal fourth quarter. Set against Microsoft’s total company revenue, which runs well over $280 billion annually, the OpenAI-linked $24.1 billion represents less than 10% of the overall business — a reminder that even blockbuster AI figures are still a modest slice of Microsoft’s broader commercial empire.
Why the Concentration Matters
Investors and analysts have flagged the concentration as a genuine risk factor. If OpenAI’s own revenue growth stalls, faces a funding disruption ahead of its planned IPO, or renegotiates its compute arrangement with Microsoft on more favorable terms, a large share of what Microsoft currently classifies as AI revenue growth could evaporate or flatten. The dependency also raises questions about how genuine the AI revenue narrative is across the hyperscaler sector broadly, since Amazon, Google, and Oracle have each disclosed similarly outsized commitments tied to a concentrated handful of frontier AI lab customers rather than a broad base of paying enterprise clients.
Microsoft’s Framing: This Is the Point of the Partnership
Microsoft executives have consistently defended the arrangement as evidence the OpenAI partnership, first struck in 2019 and expanded multiple times since, is working exactly as designed. The company has pointed to its equity stake in OpenAI’s restructured for-profit arm and its exclusive Azure hosting rights as sources of long-term value that extend well beyond the compute-for-cash mechanics visible in quarterly disclosures. Microsoft CEO Satya Nadella has repeatedly argued that owning the infrastructure layer beneath the world’s most prominent AI lab is a durable strategic asset regardless of how the revenue is categorized in any single quarter.
Skeptics See a More Circular Story
Critics, including several technology finance writers, argue the disclosure supports a broader critique that AI infrastructure spending across the industry has become significantly circular — compute providers investing in AI labs, which then spend that capital back on the providers’ own cloud services, inflating apparent revenue growth on both sides of the relationship. That dynamic has fueled comparisons to vendor financing arrangements seen in previous tech cycles, where headline growth figures obscured how much revenue was generated from related-party transactions rather than organic third-party demand.
What’s Next
The disclosure arrives just as OpenAI prepares its own IPO paperwork, meaning far more detailed financials about the Microsoft-OpenAI compute relationship should become public in the coming weeks when OpenAI’s S-1 is filed publicly. That filing will let analysts cross-reference OpenAI’s reported compute expenses against Microsoft’s Azure revenue figures for the first time, offering the clearest picture yet of how much of the AI boom’s headline numbers reflect genuine external demand versus a tightly coupled financial relationship between two companies that increasingly need each other to hit their respective growth targets. Investors preparing to evaluate OpenAI’s IPO will likely press for clarity on how sustainable the Azure compute arrangement is if OpenAI eventually diversifies further toward Oracle and its own Stargate infrastructure, a shift that could, over time, reduce the very revenue concentration Microsoft now relies on to tell its AI growth story.