OpenAI’s 2025 financial picture is a sharp reminder that the AI boom is not just a product race. It is also a spending race, and the bills are getting very large.
Audited financial figures put OpenAI’s 2025 revenue at about $13.07 billion, with total costs and expenses of roughly $34 billion. The company’s loss from operations was listed at $20.92 billion, while the net loss attributable to OpenAI landed at about $38.53 billion after a complicated set of accounting adjustments tied to ownership interests and liabilities.
That is a major jump from 2024, when OpenAI had $3.7 billion in revenue, $12.48 billion in costs and expenses, and a net loss attributable to the company of $5.09 billion. In simple terms: revenue grew quickly, but spending grew from an already enormous base, and the final loss figure increased far faster than sales.
The headline numbers
The clearest way to understand OpenAI’s reported 2025 results is to separate operating performance from accounting items that changed the final net loss figure. On an operating basis, the company spent far more than it brought in. Then additional fair-value and ownership-related items pushed the final loss attributable to OpenAI higher.
| Metric | 2024 | 2025 |
|---|---|---|
| Revenue | $3.7 billion | $13.07 billion |
| Total costs and expenses | $12.48 billion | $34 billion |
| Loss from operations | $8.78 billion | $20.92 billion |
| Net loss attributable to OpenAI | $5.09 billion | $38.53 billion |
The revenue growth is substantial. A company moving from $3.7 billion in annual revenue to more than $13 billion in one year is growing at a pace most software businesses never see. But OpenAI is not a typical software company. Its products depend on expensive model training, inference capacity, data-center access, talent, and distribution. Those costs show up across research and development, cost of revenue, sales and marketing, and general administration.
Where the spending went
OpenAI’s largest reported 2025 expense category was research and development, at $19.18 billion. Sales and marketing reached $5.73 billion, while general and administrative expenses were $1.57 billion. The full costs and expenses line was $34 billion.
For 2024, the same broad expense categories were smaller but still large: $2.65 billion in cost of revenue, $7.81 billion in research and development, $1.11 billion in sales and marketing, and $907 million in general and administrative expenses.
| Expense category | 2024 | 2025 |
|---|---|---|
| Research and development | $7.81 billion | $19.18 billion |
| Sales and marketing | $1.11 billion | $5.73 billion |
| General and administrative | $907 million | $1.57 billion |
| Total costs and expenses | $12.48 billion | $34 billion |
The practical read is straightforward: OpenAI is scaling revenue and spending at the same time, but not at the same economic margin profile that investors usually expect from mature software. The company is still paying heavily to build and run the systems behind its products.
Why the net loss is more complicated than the operating loss
The difference between the $20.92 billion operating loss and the $38.53 billion net loss attributable to OpenAI matters. The 2025 figures include a $41.55 billion loss tied to changes in the fair value of convertible interests and warrant liability, connected to OpenAI’s corporate restructuring. After other items, including interest income and interest expense, the overall net loss was listed at $60.35 billion.
That number was then reduced through amounts attributed to other interests: $17.87 billion in net loss attributable to noncontrolling members’ capital and $3.95 billion in net loss attributable to redeemable noncontrolling interests. The final loss attributable to OpenAI was $38.53 billion.
This is why the 2025 result should be read carefully. The operating loss shows the gap between the company’s core revenue and expenses. The larger net loss attributable to OpenAI reflects both that operating gap and a major restructuring-related accounting impact. Neither is small, but they answer different questions.
Microsoft and SoftBank show up throughout the figures
The 2025 figures also show significant payments involving Microsoft and SoftBank. SoftBank paid OpenAI $867 million in 2025, while Microsoft paid OpenAI $303 million.
OpenAI’s payments to Microsoft were much larger. The company paid Microsoft $10.59 billion for research and development expenses in 2025. The figures also include a $6.047 billion charge related to cost of revenue, $527 million tied to sales and marketing, and $42 million in general and administrative expenses. Altogether, OpenAI’s expenses to Microsoft came to about $17.2 billion.
At the end of 2025, OpenAI also had $3.64 billion in liabilities to Microsoft, plus $21 million in accrued expenses and other current liabilities, and $58 million in non-current liabilities.
Those numbers underline how central cloud and compute relationships are to OpenAI’s economics. Microsoft is not just a strategic partner in the background; it appears deeply embedded in the cost structure behind OpenAI’s model development and product delivery.
What the 2025 loss says about AI economics
OpenAI ended 2025 with just over $50 billion in assets, with nearly half of that in cash. That cushion matters, because the business is absorbing losses while expanding a product portfolio that serves consumers, developers, and enterprise customers.
The bullish case is easy to understand: revenue is growing quickly, demand for AI tools remains high, and OpenAI has built one of the most recognized brands in technology. The harder question is whether that growth can eventually outrun the costs of training models, running inference, selling into enterprises, and keeping pace with competitors that are also spending heavily.
The reported 2025 numbers do not answer that question by themselves. They do, however, make the scale of the challenge much clearer. OpenAI is not merely funding a startup growth curve. It is funding an infrastructure-heavy AI platform where the cost of staying near the front of the market can reach tens of billions of dollars in a single year.
For readers trying to make sense of the AI business model, the key takeaway is not just that OpenAI lost a large amount of money. It is that the company’s rapid revenue growth and its rising losses are happening at the same time. That tension is likely to define the next phase of the AI market: impressive adoption on one side, and a very expensive race to support it on the other.