Leaked Documents Reveal OpenAI's Revenue Share and Rising Inference Costs with Microsoft
Key Points
- Leaked documents show Microsoft received $493.8 million from OpenAI in 2024 and $865.8 million for the first three quarters of the next year.
- The revenue‑share agreement between the two companies is set at 20 percent.
- OpenAI’s estimated revenue exceeds $2.5 billion for 2024 and $4.33 billion for the first three quarters of the following year.
- Inference (run‑time) costs rose to $3.8 billion in 2024 and $8.65 billion in the first nine months of the next year.
- Inference spending, which is largely cash‑based, may now outpace the company’s revenue.
- Training costs are largely covered by non‑cash Microsoft credits, while inference costs are cash‑intensive.
- OpenAI primarily uses Microsoft Azure but also works with CoreWeave, Oracle, AWS, and Google Cloud.
- Both OpenAI and Microsoft declined to comment on the leaked financial details.
Newly obtained documents show that OpenAI paid Microsoft $493.8 million in revenue share for 2024 and $865.8 million for the first three quarters of the following year, reflecting a 20 percent share of OpenAI’s earnings. The data suggests OpenAI’s revenue may have topped $2.5 billion in 2024 and $4.33 billion in the first three quarters of the next year, while its cash‑based inference spend rose to $3.8 billion in 2024 and $8.65 billion in the first nine months of the subsequent period. The disparity between revenue and inference costs raises questions about the startup’s profitability, prompting both firms to decline comment.
Background and Revenue Share Details
Leaked internal documents obtained by a tech blogger provide a rare glimpse into the financial relationship between OpenAI and Microsoft. According to the files, Microsoft received $493.8 million in revenue‑share payments from OpenAI for the year 2024 and $865.8 million for the first three quarters of the following year. The documents confirm a 20 percent revenue‑share arrangement that was established after Microsoft invested heavily in the AI startup.
Revenue Estimates
Based on the disclosed 20 percent share, analysts infer that OpenAI’s total revenue exceeded $2.5 billion in 2024 and reached $4.33 billion in the first three quarters of the next year. These figures align with earlier reports that placed the company’s 2024 revenue around $4 billion and its first‑half 2025 revenue near $4.3 billion.
Inference Spending Surge
The documents also detail OpenAI’s cash‑based inference costs – the compute required to run trained models for user queries. In 2024, inference spending was roughly $3.8 billion, and it climbed to about $8.65 billion in the first nine months of the subsequent year. This rapid increase suggests that OpenAI may be spending more on inference than it is earning in revenue.
Funding Structure and Cloud Partnerships
OpenAI has historically relied on Microsoft Azure for the bulk of its compute needs, though it has also pursued agreements with CoreWeave, Oracle, AWS, and Google Cloud. Training expenses, which involve the initial development of models, are largely covered by non‑cash credits awarded by Microsoft as part of its investment, whereas inference expenses are predominantly cash outlays.
Implications and Company Responses
The disparity between revenue and inference costs fuels ongoing debate about the sustainability of OpenAI’s business model and the broader AI investment climate. Both OpenAI and Microsoft declined to comment on the leaked information.