Leaked documents shed light into how much OpenAI pays Microsoft

After a year of frenzied dealmaking and rumours of an upcoming IPO, the financial scrutiny of OpenAI is intensifying. Leaked documents obtained by tech blogger Ed Zitron provide a deeper look into the company’s financials — specifically its revenue and computing costs over the past couple of years.

Zitron reported this week that in 2024, Microsoft received $493.8 million in revenue share payments from OpenAI. In the first three quarters of 2025, that number jumped to $865.8 million, according to documents he reviewed.

OpenAI reportedly shares 20% of its revenue with Microsoft as part of a previous deal in which the software giant invested over $13 billion in the AI startup. (Neither OpenAI nor Microsoft has publicly confirmed this percentage.)

However, the situation is more complex. Microsoft also shares revenue with OpenAI, reportedly returning about 20% of the revenues from Bing and the Azure OpenAI Service, a source familiar with the matter told TechCrunch. OpenAI powers Bing, while the Azure OpenAI Service provides developers and enterprises with access to OpenAI’s models via the cloud.

The source also noted that the leaked payments refer to Microsoft’s net revenue share, not its gross share. In other words, the figures exclude the amounts Microsoft pays back to OpenAI from Bing and Azure OpenAI royalties. Microsoft deducts those amounts from its internal reporting, according to the source.

Microsoft does not break out the revenue generated by Bing and Azure OpenAI in its official financial reports, making it difficult to estimate the full extent of the payments involved.

Nevertheless, the leaked documents offer a rare window into one of the hottest companies in the private market — revealing not only how much OpenAI earns, but also how much it spends to operate.

Based on the widely reported 20% revenue-share figure, OpenAI’s estimated revenue was at least $2.5 billion in 2024 and $4.33 billion in the first three quarters of 2025 — though it’s likely higher. Previous reporting from The Information pegged OpenAI’s 2024 revenue at around $4 billion, with $4.3 billion in revenue during the first half of 2025.

OpenAI CEO Sam Altman recently said the company’s revenue is “well more” than reports of $13 billion per year, adding that it expects to end the year with an annualised revenue run rate above $20 billion — and could reach $100 billion by 2027.

According to Zitron’s analysis, OpenAI may have spent roughly $3.8 billion on inference in 2024. That figure reportedly climbed to $8.65 billion in the first nine months of 2025. Inference refers to the compute resources used to run trained AI models to generate responses.

OpenAI has historically relied almost entirely on Microsoft Azure for compute access, but has since expanded to work with CoreWeave, Oracle, and, more recently, AWS and Google Cloud.

Previous reports estimated OpenAI’s total compute expenses at around $5.6 billion for 2024, with its cost of revenue reaching $2.5 billion in the first half of 2025.

A source familiar with the company told TechCrunch that while OpenAI’s training costs are mostly non-cash (paid via credits from Microsoft’s investment), its inference spending is primarily cash-based.

While the leaked numbers don’t present a complete financial picture, they suggest that OpenAI may be spending more on inference than it earns in revenue.

That revelation could further fuel ongoing debates about an AI bubble — raising questions about whether OpenAI’s massive model operations, and by extension those of its competitors, are sustainable amid skyrocketing infrastructure costs and high valuations.