What It Means for Anthropic to Overtake OpenAI with $65 Billion in Annualized Revenue — August 18, 2026
· CompaniesAutomation
Radar Flash Edition: Anthropic reported to investors an annualized revenue run rate exceeding $65 billion at the end of July, compared to OpenAI's $40 billion, with nearly 80% coming from the API and token-billed enterprise contracts. What to measure in your AI bill today.
Flash edition. Yesterday, the Radar mentioned that Anthropic's IPO was being valued based on a 2028 forecast. Today, the data that supports it has arrived: $65 billion annualized as of the end of July and, for the first time, ahead of OpenAI. Complement to today's Radar.
What happened
Bloomberg reported on August 17 that Anthropic informed its investors of an annualized revenue run rate exceeding $65 billion at the end of July, compared to $47 billion at the end of May and approximately $9 billion at the end of 2025: $18 billion added in two months. The preliminary second quarter closed at $11.5 billion, compared to $787 million the previous year (Source). Reuters confirmed this and recalled the confidential draft filed in June with the US regulator, with a possible IPO in the fall (Source). The real news is the comparison: OpenAI is around $40 billion annualized, so Anthropic has just overtaken it (Source).
Why it matters
Look at where that money is coming from: nearly 80% comes from the API and enterprise contracts billed per token, according to data from the May round. These aren't individual subscriptions; they are invoices from companies like yours. And adding $18 billion in two months isn't explained solely by new customers, but by existing customers consuming more: this is exactly what agents do—consume in a loop without anyone signing anything. None of this is investment advice; it's the signal that the game is being played in enterprise consumption.
For your company
Three things, and none of them involve negotiating the contract. One: stop looking at the monthly bill and calculate the cost per completed task—per invoice processed, per email classified, per report generated; this is the number that tells you if the automation holds up when volume multiplies. Two: set hard limits today—maximum budget per project, alerts at 50% and 80%, and a different key for each process, so that a looping agent is a mid-morning notification and not a surprise at the end of the month. Three: route by tiers—a cheap model for classification and extraction, and the powerful one only where it changes the outcome; this is usually the fastest saving without touching quality. If you want this instrumentation set up and measured, an AI consultancy can solve it in weeks.
Frequently Asked Questions
Does an annualized run rate of $65 billion mean Anthropic has actually billed that much?
No: it is the revenue from the last month multiplied by twelve. It measures the slope, not the year, and in a token-based consumption business, it can correct as quickly as it rose. Audited figures will arrive with the IPO prospectus.
Does Anthropic overtaking OpenAI mean I should switch providers?
No. A provider's market share doesn't tell you if they are right for your use case. Your evaluation with your data and your cost per task should decide; the industry ranking is context, not criteria.
I use AI within a third-party tool, I don't contract the API. Does this affect me?
Yes, with a delay: that provider passes their token cost on to you in the rate or in the plan limits. Ask them which model is underneath and what happens to your price if theirs goes up.