5-Min Brief: OpenAI Just Took a Very Public Swing at Its Biggest Rival. Here’s What’s Actually Going On
What you need to know — in 30 seconds
- An internal OpenAI memo leaked this week in which the company’s chief revenue officer accused rival Anthropic of overstating its revenue by $8 billion
- Anthropic has been claiming a $30 billion annual revenue run rate — OpenAI says the real number is closer to $22 billion due to accounting differences
- Both companies are preparing for IPOs later this year, which is why this numbers fight matters
- The memo also took personal shots at Anthropic’s strategy, leadership, and technology — this is a real corporate rivalry getting openly hostile
A quick note before we get into this: OpenAI makes the AI assistant I used to help research this briefing. Anthropic makes Claude, the AI assistant I used to help write it. So I have some skin in this game — or at least my tools do. With that transparency on the table, here’s what happened.
The memo
On Sunday, OpenAI’s chief revenue officer Denise Dresser sent an internal memo to staff outlining the company’s strategy for the quarter. Internal memos are supposed to stay internal. This one didn’t.
The Verge and CNBC both obtained it and published the key details. And the details were unusually spicy for a document technically addressed to employees.
The main accusation: Anthropic’s widely cited $30 billion annual revenue run rate is inflated by about $8 billion. OpenAI claims the real comparable figure is closer to $22 billion.
The reason, according to the memo: it’s an accounting difference. When Anthropic sells its AI through Amazon Web Services or Google Cloud, it counts the full amount the customer pays as its own revenue — including the portion that goes to Amazon or Google. OpenAI says it doesn’t do this with its Microsoft sales, reporting only its own cut.
Both approaches are legal. Both comply with standard accounting rules. They just produce very different headline numbers — and right now, headline numbers matter enormously.
Why this fight is happening now
Both OpenAI and Anthropic are racing toward IPOs — going public on the stock market — potentially before the end of 2026. OpenAI is targeting a valuation around $1 trillion. Anthropic was most recently valued at $380 billion.
When a company goes public, investors scrutinize the revenue numbers intensely. A $30 billion run rate tells a very different story than a $22 billion run rate. The gap between those two numbers could translate to hundreds of billions of dollars in how the market values Anthropic.
So OpenAI leaking — or allowing to be leaked — a memo questioning Anthropic’s revenue methodology right now is not accidental. It’s a move. Whether it’s a smart one or a desperate one depends on your read of the competitive situation.
And the competitive situation is genuinely interesting. Corporate payments data from a company called Ramp, which tracks spending across tens of thousands of US businesses, shows Anthropic is closing fast on OpenAI in enterprise customers. Some analysts are projecting Anthropic could pass OpenAI in business customer count within months.
The other shots in the memo
The revenue number wasn’t the only thing Dresser took aim at.
She accused Anthropic of making a “strategic misstep” by not acquiring enough computing power early — claiming customers are already experiencing the consequences through slower service and unreliable availability.
She characterized Anthropic’s brand positioning as being “built on fear, restriction, and the idea that a small group of elites should control AI.”
And she noted that Anthropic is essentially a coding-focused company — strong with software developers — but argued that’s a liability as AI expands beyond technical users into every part of how businesses operate.
For a memo technically addressed to internal staff, it reads remarkably like a sales pitch against a competitor.
What Anthropic said
Nothing, publicly. A spokesperson declined to comment.
For the record, Anthropic maintains that its accounting approach is correct. The company argues it counts full revenue from cloud partner sales because it is the principal in those transactions — meaning it’s doing the work, and the cloud partners are just the distribution channel.
That’s a reasonable position. It’s also the one that produces a bigger headline number.
What this actually means
The short version: two companies racing toward trillion-dollar IPOs are now fighting in public over whose revenue is more real.
This is actually worth paying attention to because it illustrates something important about where we are in the AI industry. For the past few years, the story has been about capability — which AI is smarter, which writes better, which codes faster. Now the story is increasingly about business — who is making real money, who will survive, who gets to set the terms of how this technology gets built and deployed.
OpenAI’s implied message is that it’s winning the business race. Anthropic’s implied response — by staying silent — is that the accounting argument is noise.
Both of them are heading toward public markets where investors will have to decide which story they believe.
HumanReadable-AI covers the AI stories that matter, explained clearly. Subscribe below — free, no jargon, every weekday.