On October 9, 2026, conflicting reports put OpenAI’s annualised revenue near $50 billion by one measure and on course for $70 billion by the end of 2026 by another — a $20 billion fog sitting underneath the valuations of the entire AI industry.
The discrepancy, as the reporting explains, stems in part from how sales through cloud providers are counted. When a customer buys OpenAI’s models through Microsoft’s Azure, whose revenue is it — the model maker’s, the platform’s, or both booked gross? Annualised figures carry their own sleight: take the best recent month, multiply by twelve, and a growth spurt becomes a run-rate. Add the difference between contracted revenue, recognised revenue and cash collected, and two honest accountants can produce numbers $20 billion apart for the same company in the same quarter. Neither figure is necessarily wrong; they are answers to different questions wearing the same headline.
That fog matters because AI valuations rest on growth rates that may not be measured on the same basis from company to company. Investors comparing OpenAI’s trajectory with Anthropic’s or Google’s are implicitly trusting that the numerators were built identically. In public markets they would have no choice: securities law imposes a common grammar on reported revenue. In private markets, the grammar is homemade, the auditors’ definitions vary, and the numbers circulate through funding rounds as marketing as much as measurement.
What would clear the air
Until private AI firms report on a common basis — recognised revenue, net of partner shares, over a stated period — revenue headlines will keep moving markets more than they inform them. The first of these companies to file for a public listing will have to adopt that grammar under oath, and the rest of the sector’s numbers will be repriced against it within the week. That prospectus, whenever it arrives, may be the most informative document the AI boom produces. GlobeNews9’s Business & Technology Desk will follow the filings and the figures as they are made public.
Related reading: Orkut Founder Wants to Bring the Social Network Back to Life · Oracle Slide Highlights Power Bottleneck Behind AI Data Centres · AI Marketplaces Gain Ground as Companies Try to Cut Token Costs
