OpenAI Internal Memo Accuses Anthropic of Overstating Revenue by $8 Billion

OpenAI memo accusing Anthropic of inflating AI revenue by 8 billion dollars through cloud provider grossing up

OpenAI's Chief Revenue Officer sent an internal memo accusing Anthropic of inflating its revenue run rate by approximately $8 billion by "grossing up" revenue share arrangements with Amazon and Google rather than reporting net revenue, Bloomberg reported. The memo, sent by OpenAI CRO Dane Mitchell, frames Anthropic's widely cited revenue figures as misleading competitive intelligence and attempts to reframe the narrative around which AI company is actually winning enterprise customers. The dispute cuts to the heart of how frontier AI companies account for revenue in platform partnerships — a question with significant implications for how investors and customers assess competitive positioning in the AI market.

What "Grossing Up" Revenue Share Means

When a company like Anthropic runs its models through Amazon Bedrock or Google Cloud's Vertex AI platform, revenue flows through the cloud provider's billing system. The question of whether to report gross revenue (the total amount customers pay, including the platform's cut) or net revenue (what Anthropic actually receives after the platform takes its share) can produce dramatically different headline numbers. Mitchell's memo alleges Anthropic is reporting the gross figure — including Amazon's and Google's platform margin — which would inflate Anthropic's apparent revenue run rate by billions of dollars. Anthropic has previously reported its revenue tripling to $3 billion, with significant contributions from its cloud partnerships.

The accusation, if accurate, would mean Anthropic's actual net revenue is substantially below the figures in circulation — making OpenAI's own revenue position look stronger by comparison. The timing of the memo is notable: it comes as both companies are in active fundraising and enterprise sales cycles where relative revenue figures are used as competitive signals to potential customers and investors.

What This Says About the AI Revenue Race

The OpenAI-Anthropic revenue dispute reflects a broader dynamic in frontier AI: the gap between "AI revenue" as companies define it internally and what actually flows to their bottom lines is large and poorly understood by the market. Both companies rely heavily on cloud distribution partners who take meaningful revenue shares. Both report "run rates" — annualized projections from recent periods — rather than trailing revenue, introducing additional room for interpretation. The memo's existence suggests OpenAI views Anthropic's reported revenue figures as a competitive threat serious enough to require internal counter-messaging, which is itself a signal of how closely the two companies are tracking each other in enterprise sales.

Frequently Asked Questions

What did OpenAI's CRO say about Anthropic's revenue?

OpenAI CRO Dane Mitchell sent an internal memo alleging Anthropic overstates its revenue run rate by approximately $8 billion by grossing up revenue share with Amazon and Google rather than reporting net revenue received.

What is Anthropic's actual revenue run rate?

Anthropic has publicly cited revenue figures including a $3 billion annual run rate. OpenAI's memo alleges the true net figure is approximately $8 billion lower than Anthropic's reported gross run rate.

Why does this matter for AI customers and investors?

Revenue run rates are used by enterprise customers to assess which AI vendor has achieved scale and staying power. Inflated figures could mislead customers about Anthropic's competitive position relative to OpenAI and other vendors.

The Bottom Line

An OpenAI internal memo accusing a competitor of inflating its revenue figures is unusual public-relations territory for a company that typically focuses its messaging on technical capability. The fact that it leaked suggests the document was circulated widely enough internally that containment was difficult — and the substance of the allegation, if accurate, would meaningfully change how the AI revenue race is understood. Whether Anthropic's accounting reflects legitimate gross revenue reporting or misleading inflation of headline figures, the dispute underscores that "AI revenue" is not a standardized metric and that the competitive narrative around it is actively being shaped by the companies themselves.