CNBC reports that OpenAI told investors it had approximately $50 billion in annualized revenue at the end of September, below the widely reported $68 billion figure. A person familiar with the matter told CNBC that the larger figure included gross revenue from OpenAI’s partners. The investor presentation also cited 77% total run-rate growth in the third quarter and 107% growth for the enterprise business. CNBC reports same-day declines of 3% for Nvidia, nearly 6% for Oracle, and nearly 8% for CoreWeave.
Mechanism: Investors comparing AI companies need to distinguish revenue booked by a company from gross revenue that includes partners. A change in that boundary can alter perceived scale and comparability without, by itself, demonstrating a change in underlying usage, infrastructure commitments, or future growth. The market reaction indicates sensitivity to the disclosed figures, not proof of a causal change in fundamentals.
Why now: OpenAI’s reported $50 billion run rate and the explanation for the previously circulated $68 billion figure make the accounting boundary newly salient as investors assess the company’s scale.
First implication: For infrastructure suppliers, exposure to a customer’s reported growth is not a substitute for evidence about that customer’s actual compute purchases or commitments.
The CNBC account attributes the explanation of the $68 billion figure to an unnamed person familiar with the matter and says the $50 billion figure was confirmed by CNBC. The article records share-price moves and investor presentation metrics; it does not provide the presentation itself or establish how much of the market decline was caused by this revenue clarification. Evidence boundary: The report supports describing the disclosure, attributed explanation, and contemporaneous share moves, but not a conclusion about future revenue or compute demand.
First watch condition: Watch for later company filings or investor disclosures that define whether reported revenue includes partner activity and clarify actual infrastructure commitments.
With no prior New Runtime record supplied, this begins a distinct analytical record about how revenue attribution shapes market interpretation of AI infrastructure exposure.