A complex financial picture is emerging for OpenAI, the prominent artificial intelligence research lab, as newly leaked internal documents suggest the immense computational cost of running its powerful AI models may be consuming a substantial portion, if not all, of its rapidly growing revenue. The documents, brought to light by sources including Techcrunch and tech blogger Ed Zitron, detail a sophisticated and massive flow of money between OpenAI and its primary partner, Microsoft. While OpenAI's revenue is soaring to unprecedented heights for a company of its age, the staggering expenses associated with "inference"—the process of generating responses from its models—are raising fundamental questions about the long-term economic viability of large-scale AI and the company's path to profitability.

At the heart of the revelations is the sheer scale of OpenAI's operational spending. According to analysis of the leaked documents, the company's inference costs, which represent the computing power needed to run its models for users, reached an estimated $3.8 billion in 2024 and ballooned to approximately $8.65 billion in just the first nine months of 2025.[1][2][3][4][5] This dramatic escalation in expenditure appears to be outpacing the company's impressive revenue growth, creating a scenario where the cost of delivering its AI services could be higher than the income they generate.[6][7][5] This financial pressure is compounded by the nature of these payments; while the initial, and also considerable, cost of training new models is often covered by cloud computing credits from Microsoft as part of its investment, the ongoing inference costs are reportedly paid for largely in cash, placing a direct and heavy burden on OpenAI's finances.[1][2][3]

The leaked financials also provide a rare window into the intricate, multi-billion dollar relationship between OpenAI and Microsoft. The partnership involves a reciprocal revenue-sharing agreement. Documents show Microsoft received significant payments from OpenAI, reportedly totaling $493.8 million in 2024 and rising to $865.8 million through the first three quarters of 2025.[1][2][8][6][4] This is believed to be part of an arrangement where OpenAI shares roughly 20% of its revenue with its largest backer.[1][2][8] However, the financial traffic is not one-way. Microsoft, in turn, pays OpenAI a share of the revenue it generates from products that incorporate OpenAI's technology, such as the Bing search engine and the Azure OpenAI Service.[1][2][3][8] The leaked figures allegedly represent the net amount Microsoft keeps after these deductions, making the gross financial exchange between the two tech giants even larger and more complex than the numbers initially suggest.[1][8][6]

These revelations have significant implications for the broader artificial intelligence industry, which has been buoyed by massive valuations and investments predicated on future profitability. OpenAI is widely seen as the market leader, and if its financial footing is precarious due to operational costs, it suggests that smaller competitors and the sector as a whole face a daunting and capital-intensive path forward.[1] The high cost of inference is a fundamental challenge for all large language model developers, threatening the scalability and sustainability of their business models.[9][10] While OpenAI's CEO Sam Altman has projected an annualized revenue run rate that could exceed $20 billion by the end of 2025, the leaked expenditure figures cast a shadow on these optimistic forecasts, highlighting the immense gap that must be bridged between revenue generation and the cost of service delivery.[11][1][12] The data suggests that even at the industry's cutting edge, the technology's operational demands are immense, making profitability an elusive target.[13]

In conclusion, the leaked financial documents paint a sobering picture of the economic realities facing the generative AI boom's leading company. Despite phenomenal revenue growth driven by the widespread adoption of products like ChatGPT, OpenAI appears to be locked in a costly battle with its own operational expenses. The enormous and rapidly increasing cost of inference presents a formidable obstacle not just for OpenAI but for the entire AI industry, suggesting that the road to a profitable and sustainable business model may be longer and more challenging than previously understood. The intricate financial dance with Microsoft underscores the deep capital requirements needed to compete at the highest level, leaving the question of whether revenue can ultimately outpace the voracious appetite of the underlying technology open for the foreseeable future.