OpenAI‘s Rising Costs: Is the AI Giant Spending More Than It Earns?
OpenAI, the driving force behind ChatGPT and other groundbreaking AI technologies, is experiencing rapid growth. Revenue is projected to exceed $20 billion this year, possibly reaching $100 billion by 2027. Though, leaked documents and analysis suggest a concerning trend: the company’s operational costs, particularly those related to running its AI models, might potentially be outpacing its revenue.
Here’s a breakdown of the situation:
* Revenue Growth: openai is currently on track to generate over $13 billion in annual revenue, with projections exceeding $20 billion.
* Future projections: Ambitious forecasts suggest OpenAI could hit $100 billion in revenue by 2027.
* Inference Costs Soaring: Inference – the computational power needed to use a trained AI model – is a major expense. It’s estimated OpenAI spent roughly $3.8 billion on inference in 2024, climbing to $8.65 billion in the first nine months of 2025 alone.
The Compute Conundrum
running these powerful AI models requires massive computing resources.Historically,OpenAI has relied heavily on Microsoft Azure for this infrastructure. However, the company is diversifying, forging deals with CoreWeave, Oracle, AWS, and Google Cloud to secure access to vital compute power.
A Closer Look at the Numbers
Previous reports indicate OpenAI’s total compute spend reached approximately $5.6 billion in 2024. moreover, the company’s “cost of revenue” reached $2.5 billion in the first half of 2025. These figures raise a critical question: is OpenAI spending more to run its models than it’s earning from them?
Training vs. Inference: Understanding the Costs
It’s important to distinguish between training and inference.
* Training involves the initial process of building and refining an AI model, often utilizing credits awarded by Microsoft as part of their investment. This is largely considered a non-cash expense.
* Inference,on the other hand,is the ongoing cost of using that trained model to generate responses. This is primarily a cash expense.
Sources indicate that OpenAI’s inference costs are considerably contributing to its overall financial strain.
Implications for the AI Industry
If OpenAI, a leader in the AI space, is struggling with profitability despite its impressive revenue growth, what does this mean for the broader industry? The situation fuels ongoing debate about the “AI bubble” and the sustainability of valuations for AI companies.
You might be wondering if this signals a broader issue.Are other AI companies facing similar challenges with escalating costs and uncertain profitability? The answer remains to be seen,but OpenAI’s situation serves as a crucial case study.
What’s Next?
OpenAI declined to comment on these findings, and Microsoft has not yet responded to inquiries. However, the leaked data provides a valuable glimpse into the financial realities of building and operating cutting-edge AI.
Have a tip? TechCrunch is actively investigating the inner workings of the AI industry. You can reach out to Rebecca Bellan at [email protected] or Russell Brandom at [email protected]. For secure interaction, contact them via Signal at @rebeccabellan.491 and russellbrandom.49.
Related reading