OpenAI in 2025: Losses Soar to $38.5 Billion — What Lies Behind the Numbers?
OpenAI's financial results for 2025 show a sharp deterioration: the company's net loss reached $38.5 billion, nearly eight times the previous year's figure. These data, obtained from audited reports, shed light on the enormous costs associated with scaling and commercializing artificial intelligence technologies.
Trend Analysis: 2024 vs 2025
To understand the scale of what is happening, it is worth looking at the comparative dynamics. In 2024, OpenAI's revenue was $3.7 billion, with total expenses of $12.48 billion. The operating loss then reached $8.78 billion, and the net loss, including interest income and expenses, was $8.84 billion. After write-downs for non-controlling interests, the company's own loss was recorded at $5.09 billion.
In 2025, the picture changed dramatically. Revenue grew to $13.07 billion, but expenses surged to $34 billion, and the operating loss reached $20.92 billion. The main burden fell on research and development — $19.18 billion. Another $7.5 billion was the cost of revenue, and $5.73 billion was sales and marketing. A key factor was OpenAI's transition from a non-profit to a commercial structure, resulting in a one-time loss of $41.55 billion due to the revaluation of convertible instruments and warrants. After adjustments, the net loss was reduced to $38.53 billion.
Settlements with Microsoft: The Main Beneficiary
Transactions between OpenAI and Microsoft deserve special attention. In the 2025 calendar year, OpenAI paid Microsoft $10.59 billion under the "research and development" line — apparently the cost of training models. Another $6.047 billion went to the cost of revenue, $527 million to sales and marketing, and $42 million to administrative expenses. In total, OpenAI's expenses in favor of Microsoft amounted to $17.2 billion.
By the end of the year, OpenAI's liabilities to Microsoft reached $3.64 billion, plus $21 million in accrued and other current liabilities, and another $58 million in long-term liabilities. The company has already contracted about $600 billion in future data center expenses, indicating the long-term nature of the cost model.
Expert Comment: These figures expose a fundamental problem for the entire AI industry: the cost of servicing billions of daily queries on expensive chips is many times higher than current revenue. OpenAI is essentially operating like a giant startup, where operating expenses for infrastructure and model training cannot yet be offset without constant external infusions. This calls into question the sustainability of the business model in the long term, especially amid growing competition.