Internal OpenAI documents forecast $14 billion loss in 2026

Pcgamer reports that internal OpenAI documents seen by The Information forecast a $14 billion loss for 2026 as part of a set of financial projections for the company.

The documents say OpenAI could continue to record losses totalling about $44 billion through 2029 before turning a profit and generating "Nvidia-style" revenues. The projections include a $100 billion revenue scenario split just over 50% to ChatGPT, roughly 20% to sales of AI models to developers via APIs, and about 20% to "other products" such as video generation, search and proposed AI research assistants. The $14 billion 2026 loss is said to be roughly three times worse than early estimates for 2025.

It is also reported that inference costs are falling fast and that spending on acquiring training data is forecast to drop from $500 million this year to about $200 million annually toward the end of the decade. The documents do not make clear exactly what this means for training methods or data sources; the report suggests it could indicate more recursive training on AI-generated data, but details remain unclear and the financial outcomes are uncertain.

Key Topics

Business, Openai, Chatgpt, Nvidia, Ai Models, Inference Costs