OpenAI Doubled Its Revenue to $40 Billion — and Everyone Is Running for the Exit

OpenAI Doubled Its Revenue to $40 Billion — and Everyone Is Running for the Exit

Let me tell you something that's been sitting with me since the Bloomberg report dropped Thursday. OpenAI — the company that started this whole AI gold rush — just doubled its annualized revenue to over $40 billion. In eight months. That is the loudest demand signal this industry has produced all year. And my first reaction wasn't excitement. It was: why is everybody leaving?

Because the same week that number leaked, the chief revenue officer quit. The chief operating officer quit two days before that. The company is still burning through cash at a rate that would bankrupt a normal business, and it's about to hand its books to the public market for the first time. Ent? Something doesn't add up.

This is not a doom piece. I'm a founder who runs real infrastructure, and I've been loud about the AI buildout being real. But when the biggest customer in the AI economy doubles its revenue and the executives sprint for the exits in the same week, you stop and look at both numbers. Because one of them is lying to you.

The Number That Changes the Conversation

Let's get the headline right first. Bloomberg reported Aug. 13, citing people familiar with the company's performance, that OpenAI is on track to generate annualized revenue of more than $40 billion — roughly double its run rate at the end of 2025. The drivers are the ones you'd expect: AI coding tools, enterprise services, subscriptions, even advertising. July alone saw monthly revenue climb more than 20 percent.

Put that in context. OpenAI was doing around $2 billion a month by spring, about $25 billion annualized in April and May. Now it's $40 billion plus. That is not a rounding error. That's a company whose products people are actually paying for, at a scale that would make most public software companies weep.

And it isn't just OpenAI. Anthropic — the rival that passed OpenAI in revenue earlier this year — is reportedly running near $47 billion annualized on a $965 billion valuation. Both companies have filed confidential IPO paperwork. Both are targeting autumn listings. The two most important private companies in AI are sprinting to Wall Street at the same time, both armed with revenue numbers that would have sounded like science fiction two years ago.

Reading One — The Demand Story Is Real

Let me be fair to the bull case, because it isn't stupid. For a decade I've watched cloud providers and hosting companies sell "AI-ready" capacity on promises. The $40 billion number is not a promise. It's money hitting the bank. Coding tools, enterprise contracts, subscription tiers — that's recurring revenue from companies that vote with budgets, not with tweets.

For the infrastructure economy, this matters more than any earnings call. OpenAI doesn't just sell AI; it consumes compute at a scale that moves markets. Every dollar of that $40 billion has a data center attached somewhere. The demand that justifies the buildout — the capex programs, the GPU orders, the power contracts, the colo leases — is being fed by real revenue now. That's the strongest answer yet to the people who've been calling the whole thing a bubble.

I said it before and I'll say it again: this buildout is not fake. The question was never whether people would use AI. It was whether anyone would pay for it. Answer: they're paying $40 billion a year and climbing.

Reading Two — Revenue Isn't Profit, and the Exit Door Is Wide Open

Now here's the number the headlines buried. OpenAI is still losing money — something like $14 billion projected for 2026, and as of late spring it was losing more than a dollar for every dollar it earned. Revenue doubled, sure. But the losses didn't halve. The gap between the top line and the bottom line is still a canyon, and the only bridge across it is the public market.

That's why the executive departures matter. This week Brad Lightcap — the chief operating officer, eight years at the company, one of Sam Altman's closest people — announced he's leaving to "start something new." Two days later, revenue chief Denise Dresser, an eight-month veteran recruited from Salesforce to win enterprise deals, said she's out too. Her replacement, Dali Rajic, comes from cybersecurity firm Wiz. Dresser called the work "nothing short of incredible" in her farewell post. Sure.

I've been around long enough to know what executive churn at IPO time looks like. Sometimes it's innocent — people cash out, people start new things. But when the CFO, the COO, and the CRO all shuffle in the same quarter the S-1 goes public, the message the market hears is: the smartest people in the room are selling their future in the company, not buying more. And the only ones left to buy are you.

The Secondary Bottleneck — The IPO Window Is the New Governor

Here's the part nobody's pricing in, and it's the part that keeps me up at night. The AI capex cycle doesn't run on revenue. It runs on financing. OpenAI raised $122 billion in March at an $852 billion valuation. It needs the public markets to rotate that private money — to give the early investors, the employees, the banks, and the debt holders an exit. Anthropic needs the same thing. So does SpaceX, which is about to debut in the largest IPO in history with the same two lead banks.

Now count the listings stacking into the same window: Labor Day through Thanksgiving. OpenAI at a trillion-dollar target. Anthropic at $965 billion. SpaceX on top. That's not a window, that's a traffic jam. If even one of these IPOs prints soft — if the public market looks at a $1 trillion company losing $14 billion a year and says "no thanks" — the signal ripples through the entire financing stack. Junk bonds, vendor financing, construction loans, colo leases. Everything built on the assumption of endless private capital suddenly has to answer to public-market math.

That's the secondary bottleneck nobody's talking about. Not GPUs. Not power. Not water. The IPO window itself. The moment the fantasy meets the P&L.

What This Means for Independent Hosting Providers

So what do you do with this if you're running an independent hosting business, a colo, a cloud shop? Same thing I'm doing.

First — treat OpenAI's revenue as the demand signal it is, but don't extrapolate the profit. Real revenue means real compute demand. If you've got capacity, the AI economy is still buying. But price your contracts against a world where the biggest customers can still blow up, because revenue without profit is a house of cards with a nice front door.

Second — watch the IPO calendar like a hawk. The September-through-Thanksgiving window is now the single biggest risk factor in AI infrastructure financing. If OpenAI or Anthropic slips or prints soft, hyperscaler credit tightens within weeks. That's your early-warning system for capacity prices, hardware lead times, and your own borrowing costs.

Third — diversify your customer base while the window is open. Right now everybody wants to be the hosting provider for an AI lab. Fine. But the labs are burning cash and their investors need exits. When the music stops, the providers with 20 percent AI concentration will feel it a lot less than the ones with 80 percent. I lived through 2008. Concentration is how businesses die.

Fourth — get your own house public-market ready. If you ever want to raise, sell, or borrow at good rates, the bar just got higher. The same investors who scrutinize OpenAI's losses will scrutinize your books. Clean them up now.

The Bottom Line

OpenAI doubling its revenue to $40 billion is the best news the AI buildout has had all year. It's proof the demand is real, the customers are paying, and the infrastructure economy has a foundation. But it's also the moment the industry stopped being a private-market fantasy and became a public-market responsibility. Revenue doubled. Losses didn't. Executives left. The IPO is the exit.

That's not a reason to panic. It's a reason to be honest. The AI economy is real, and it's also fragile, and both of those things are true at the same time. The founders who win the next five years are the ones who can hold both readings in their head without flinching.

I know which one I'm planning for.

— Allan Ali, Founder

This article was produced with AI-assisted research and editorial support. Sources: Bloomberg, Quartz, Yahoo Finance, Financial Post, NY Post, Briefs, BeInCrypto, Blockonomi.