OpenAI losses reached $38.5 billion in 2025, an eightfold increase from the previous year for the artificial intelligence startup OpenAI. The company’s revenue for the same period reached $13.05 billion, according to official financial documents cited by The Financial Times.
Financial Performance and OpenAI Losses
In 2024, the company’s financial results were significantly different. OpenAI reported $3.7 billion in revenue and a net loss of $5.09 billion. Operating losses in 2024 stood at $8.78 billion, but a portion of these losses was attributed to non-controlling interests, keeping the net loss at $5.09 billion.
In 2025, the initial loss reached $60.35 billion, but only a portion was recognized on the company’s balance sheet. This resulted in a final net loss of $38.53 billion, with operating losses totaling $20.92 billion.
Breakdown of Major Expenses
The rising losses in 2025 were driven by a significant increase across all categories of operating expenses:
- Cost of revenue: increased from $2.65 billion in 2024 to $7.5 billion.
- Research and development (R&D): spending rose from $7.81 billion to $19.18 billion.
- Marketing: expenses increased more than fivefold to $5.73 billion.
- Total expenses: amounted to $34 billion.
Financial Ties with Microsoft and SoftBank
The financial statements also detail transactions with key partners. SoftBank paid OpenAI $867 million for services, while Microsoft paid $303 million.
In turn, OpenAI paid Microsoft $10.59 billion for research and development services. These costs were primarily related to renting Microsoft’s hardware infrastructure to train artificial intelligence models. In total, Microsoft received $17.2 billion from OpenAI in 2025. By the end of the year, OpenAI’s outstanding debt to Microsoft stood at $3.64 billion.
Asset Valuation and Market Outlook
By the end of 2025, OpenAI’s total assets were valued at $50 billion, with approximately half held in cash and cash equivalents.
Operating losses of this scale could complicate negotiations with investors ahead of a potential initial public offering (IPO). The disclosed figures also fuel debate in the financial community regarding the risks of an “AI bubble” in the tech startup market.
OpenAI Losses Analysis
The financial statements illustrate the challenges faced by companies lacking their own hardware infrastructure. In an artificial intelligence market currently experiencing a gold rush, the reported figures suggest several conclusions:
- The “picks and shovels” strategy remains highly profitable, as Microsoft demonstrates. The tech giant invests in OpenAI, and a significant portion of those funds returns to Microsoft to pay for cloud computing and server rentals ($10.59 billion). This arrangement makes OpenAI a major, highly dependent client, allowing the infrastructure provider to capture much of the financial benefit.
- The advantage of a full-cycle business model. OpenAI’s financial situation shows the vulnerability of relying on rented computing power. When a company owns its entire technology stack—from designing its own processors and building data centers to integrating AI into end-user services, as Google does—the business economics are different. Owning the hardware infrastructure eliminates the high margins charged by third-party cloud providers, improving financial viability even during periods of heavy research investment.
- Marketing vs. technology. The fivefold increase in marketing expenses to $5.73 billion indicates that the product is becoming harder to sell organically, despite the company being at the forefront of innovation. Retaining users is proving more expensive than anticipated.
- The AI bubble: real technology, questionable business models. Neural networks and generative AI themselves are not a bubble, as they already provide measurable value in programming, data analysis, and science. Instead, the bubble risk lies in the valuation of startups that compete in the foundation model market without owning physical assets. The costs of training systems are growing exponentially, while subscription pricing for users faces natural limits.
NEWSROOM IN notes that on June 8, 2026, OpenAI filed a registration statement with U.S. regulators for an initial public offering. According to Reuters, the company could go public in September 2026 and aims to raise up to $60 billion.
