Leaked audited financials reveal a company racing to scale revenue while spending still dwarfs every dollar coming in.

OpenAI is generating revenue at a pace that would make most startups envious. According to newly leaked financial documents reviewed by the Financial Times and independently obtained by journalist Ed Zitron, the company brought in $13.07 billion in revenue in 2025, up from $3.7 billion the year before. Monthly run-rate approached $2 billion by the end of last year.

But the spending is even more staggering. Research and development costs alone hit $19.18 billion in 2025, with $10.59 billion of that flowing directly to Microsoft for compute and infrastructure. The figures reflect the enormous cost of training next-generation models and running inference at scale for millions of users.

On top of R&D, OpenAI spent $7.5 billion on cost of revenue, up from $2.65 billion in 2024, and $5.73 billion on sales and marketing, nearly five times the prior year. The result: an operating loss of $20.92 billion, widening from $8.78 billion in 2024.

The numbers land at a critical moment. OpenAI has filed its S-1 with the SEC ahead of a widely anticipated IPO, and the company has told investors it aims to reach profitability by 2030. As a percentage of revenue, losses did improve slightly, dropping from 237 percent in 2024 to 160 percent in 2025, suggesting the business model may eventually converge, but the path remains long.

The financials underscore a central tension in the AI industry: the technology works, the demand is real, but the economics of building and running frontier models at global scale remain punishingly expensive. For OpenAI, the upcoming IPO will test whether public market investors are willing to fund a decade-long runway to profitability.

Image source: freeimage.host