OpenAI Reportedly Earns $20 Billion Less Than Previously Thought

OpenAI's annual revenue may be $20 billion lower than previously thought. Earlier reports put the lab's annualized revenue at around $70 billion. A new report covered by TechCrunch gives a substantially lower figure. Subtracting the stated gap gives roughly $50 billion. That is still an enormous sum, but the difference is almost 29% of the earlier estimate.

A caveat: the available coverage contains neither the methodology behind the new calculation nor any confirmation or denial from OpenAI itself. It is more accurate to describe this as a dispute between two estimates than as a proven error.

Where the gap could come from

Annualized revenue (run-rate) almost never equals the money actually received over a year. The usual method takes revenue from a short period, such as a month or a quarter, and multiplies it by the number of such periods in a year. The method is sensitive to spikes. A single strong month, large one-off contracts or seasonal peaks can easily inflate the result.

Estimates can also diverge in other places. Revenue can be counted gross or net of partner shares, including the agreement with Microsoft. Contracts that have been signed but are not yet generating money may or may not be included. The available data does not say which of these factors played a role here. These remain hypotheses, not conclusions of the report.

Key figures

• Previous estimate: about $70 billion on an annualized basis.
• Discrepancy with the new report: roughly $20 billion, about 29% of the earlier figure.
• Implied level: around $50 billion.
• Past benchmark: at the end of 2025, Sam Altman spoke of a run-rate above $20 billion.
• Infrastructure commitments: according to Altman, about $1.4 trillion over several years and around 30 GW of capacity.

Why the number matters for the whole market

OpenAI's revenue underpins a whole chain of deals. Oracle, Nvidia, AMD, Broadcom, cloud contractors and data center operators are all planning on the assumption that the lab can pay for the compute it has contracted. The higher the revenue, the safer those commitments look. If the base is $20 billion smaller, the gap between income and costs widens at the same level of spending. Closing it would take new funding rounds, debt or partner financing.

It is also worth noting that even $50 billion, if confirmed, would mean very fast growth compared with the end of 2025. The problem is not weak demand for the models. The problem is that the market may have priced in a higher bar than the real one.

How to read numbers like these

For developers and companies building products on top of the OpenAI API, this news alone is no reason to panic. But it is a good reason to check your dependence on a single vendor. Build in the ability to switch to another model, keep an abstraction layer over the provider, and avoid signing long contracts without clear pricing terms.

Investors and analysts would do well to ask one question about any headline on AI company revenue: what exactly was counted? Run-rate, invoiced revenue and reported revenue produce different numbers. The next things to watch are an official comment from OpenAI, disclosures from partners such as Microsoft and Oracle, and the terms of the next funding round. If the company confirms the new figure, expectations for the payback of AI infrastructure will have to be revised along with it.