OpenAI Financial Leak: Billions in Annual Losses Amid Rapid Revenue Growth
OpenAI reports multi-billion dollar annual losses despite surging revenue
Leaked financial documents indicate that OpenAI is currently losing billions of dollars per year, a deficit driven largely by aggressive research and development (R&D) spending. While the company is operating at a significant loss, its gross revenue is growing rapidly, reaching $13 billion in 2025.
Revenue and Cost Analysis for 2025
OpenAI generated $13 billion in gross revenue in 2025, with the cost of that revenue totaling $7.5 billion. This suggests that while the core operation of providing AI services is generating substantial income, the overhead and infrastructure costs remain high.
Key Financial Metrics
- Gross Revenue (2025): $13 billion
- Cost of Revenue: $7.5 billion
- User Base: Over 900 million weekly active users of ChatGPT
- Paid Conversion: Approximately 50 million paid subscribers
The R&D Burden and Path to Profitability
The majority of OpenAI's financial losses are attributed to R&D costs, which constitute the "lion's share" of expenditures. This indicates that the current losses are a strategic choice to prioritize the development of future models over immediate profitability.
Analysis of Operational Costs
Industry observers note that if R&D spending were reduced, the company would be closer to profitability. Some analysts argue that the current financial structure is typical for companies attempting to achieve market dominance at a breakneck pace, where losses are sustained until the company "wins" the market.
The Challenge of User Conversion
With only 50 million paid subscribers out of 900 million weekly active users, OpenAI faces a significant challenge in converting free users to paid tiers. The proliferation of free alternative models makes this conversion increasingly difficult.
Technical and Economic Perspectives from the Community
Technical analysts and users have raised several points regarding the sustainability and efficiency of OpenAI's current trajectory:
Inference Costs and Diminishing Returns
There is an ongoing debate regarding whether the focus should shift from raw R&D to improving the cost of inference. As models become more capable, the question arises as to whether the marginal productivity gain of a more expensive model justifies the cost increase for the end user.
"I wonder if we are at a point where the focus can shift to improving the cost of inference. Unless we are genuinely pushing to find AGI... LLMs in their current form don't replace knowledge workers but are an effective force multiplier."
Infrastructure and Asset Depreciation
Questions remain regarding how OpenAI calculates the cost of revenue, specifically concerning rapidly depreciating hardware assets and the cost per arithmetic operation for inference. The financial sustainability of the model may depend on whether the cost of inference continues to fall or if the company must rely on expensive GPU leases.
Market Reach and Customer Acquisition
Some observers suggest that the unit economics are better than expected, noting that the company spends roughly $100 on sales for each paying customer. However, reaching profitability may require a massive expansion of the paying user base, potentially requiring billions of additional users willing to pay US-market prices.