Leaked Documents Reveal OpenAI's Massive Compute Costs May Exceed Revenue

Financial Transparency Under Scrutiny

OpenAI's financial operations have come under intense scrutiny following the leak of internal documents that reveal the company's substantial payments to Microsoft and soaring compute costs. Tech blogger Ed Zitron obtained documents showing that Microsoft received $493.8 million in revenue share payments from OpenAI in 2024, with that figure jumping dramatically to $865.8 million in just the first three quarters of 2025 1

1

.

These payments stem from OpenAI's agreement to share 20% of its revenue with Microsoft as part of the software giant's investment of over $13 billion in the AI startup 1

1

. However, the financial relationship between the two companies is more complex than initially apparent, as Microsoft also kicks back approximately 20% of revenues from Bing and Azure OpenAI Service to OpenAI 1

1

.

Complex Revenue-Sharing Arrangements

The leaked payment figures represent Microsoft's net revenue share rather than gross amounts, meaning they exclude whatever Microsoft paid back to OpenAI from Bing and Azure OpenAI royalties 1

1

. This accounting method makes it challenging to determine the true financial flow between the companies, as Microsoft deducts its payments to OpenAI from its internally reported revenue share numbers.

Based on the widely reported 20% revenue-sharing arrangement, the leaked documents suggest OpenAI's revenue was at least $2.5 billion in 2024 and $4.33 billion in the first three quarters of 2025, though the actual figures are likely higher 1

1

. Previous reports from The Information estimated OpenAI's 2024 revenue at around $4 billion, with first-half 2025 revenue reaching $4.3 billion.

Soaring Compute Costs Raise Profitability Concerns

Perhaps more concerning for OpenAI's financial outlook are the company's inference costs, which have grown exponentially. According to Zitron's analysis, OpenAI spent approximately $3.8 billion on inference in 2024, with costs ballooning to roughly $8.65 billion in the first nine months of 2025 2

2

. Inference refers to the computing power required to run trained AI models and generate responses for users.

The company has historically relied almost exclusively on Microsoft Azure for compute access, though it has recently diversified with deals involving CoreWeave, Oracle, AWS, and Google Cloud 1

1

. A source familiar with the matter revealed that while OpenAI's training costs are largely covered by Microsoft credits awarded as part of the investment agreement, inference costs are predominantly paid in cash 1

1

.

Industry Implications and Future Projections

The financial data suggests OpenAI may be spending more on inference costs than it generates in revenue, raising significant questions about the company's path to profitability 2

2

. This revelation comes at a time when CEO Sam Altman has made bold revenue projections, claiming the company's annual revenue is "well more" than the frequently cited $13 billion figure and projecting an annualized revenue run rate above $20 billion by year-end, with potential to reach $100 billion by 2027 1

1

.

These financial revelations have intensified discussions about the sustainability of current AI industry valuations and business models. If OpenAI, widely considered the leading AI company, is struggling with profitability despite massive revenue growth, it raises broader questions about the financial viability of the entire AI ecosystem and the jaw-dropping valuations that have become commonplace in the sector.