OpenAI is facing heightened scrutiny as newly leaked financial documents shed light on the company’s rapidly expanding revenue — and its equally fast-rising compute costs. Documents seen by blogger Ed Zitron indicate that Microsoftreceived $493.8 million in revenue-share payments from OpenAI in 2024, surging to $865.8 million in the first three quarters of 2025. These figures reflect a complex two-way revenue arrangement stemming from Microsoft’s more than $13 billion investment in the AI leader.
The leaks suggest that OpenAI generated at least $2.5 billion in revenue in 2024 and more than $4.33 billion in the first nine months of 2025, aligning with external reports estimating revenue at roughly $4 billion in 2024 and $4.3 billion in the first half of 2025. CEO Sam Altman has recently told investors the company is operating at a run rate above $20 billion, with long-term expectations that revenue could reach $100 billion by 2027.
Yet the financials also highlight escalating costs. Zitron’s analysis indicates OpenAI spent approximately $3.8 billion on inference compute in 2024, rising to $8.65 billion in the first three quarters of 2025. While training compute is largely offset by credits from Microsoft, inference — the compute required to run models — is predominantly a cash expense. Deals with CoreWeave, Oracle, AWS, and Google Cloud have diversified its infrastructure beyond Azure, but the numbers imply that inference spending may be outpacing revenue.
Amid this financial pressure, OpenAI has also rolled out a smaller but widely noticed product improvement. After months of user frustration, Altman announced that ChatGPT will now comply when instructed not to use em dashes, following updates to personalization controls. The company said users can better restrict the punctuation through custom instructions across platforms.
OpenAI declined to comment on the leaked financial data, while Microsoft did not respond to requests for comment.
