In September, OpenAI’s reported annualized revenue of $70 billion raced through the market as evidence that the company was closing in on Anthropic’s scale. But the number was not OpenAI’s standard top line: it was an investor-driven attempt to make the two rivals look comparable, using Anthropic’s treatment of cloud-partner sales.¹

The distinction is technical, but consequential. Under the approach described by accounting professor Francine McKenna, Anthropic can record the full value of a sale made through a cloud provider and then book the provider’s share as an expense. OpenAI records only its portion of certain partner transactions. Both methods can comply with GAAP; the disagreement turns on who controls the customer relationship and delivers the service.¹

By late last month, OpenAI’s annualized revenue was instead “about $50 billion,” up from roughly $28 billion at the end of June, according to an OpenAI spokesperson. The figure still signals striking growth, but it is $20 billion below the headline number that had helped fuel comparisons with Anthropic.²

That clarification landed badly in a market already demanding proof that enormous AI investment can translate into durable returns. Technology shares exposed to the sector fell after the revenue update, as investors confronted the possibility that the industry’s most eye-catching metrics may depend as much on presentation as on underlying sales.²

The pressure is particularly acute for OpenAI, which is trying to justify huge capital needs after raising $122 billion in a March funding round. Its leaked 2025 financials reportedly showed about $13 billion in revenue alongside far heavier spending, while a once-anticipated IPO has reportedly slipped to early 2027.³

For now, the $50 billion run rate remains formidable. Yet the episode underscores the harder question investors cannot annualize away: how much of AI’s dazzling revenue narrative will withstand the cleaner, less forgiving disclosure standards of public markets?