OpenAI ended February at $25 billion in annualized revenue, up 17% from the $21.4 billion run rate the company exited 2025 with. Three years ago, that number was effectively zero. Two years ago, it was a few billion. The growth curve for what is now arguably the most consequential private company in the world has gotten so steep that the prior year's milestones look quaint by Q1.
This is the inflection point where the IPO conversation stops being theoretical and becomes a scheduling question.
The Numbers That Matter
OpenAI's internal targets, per The Information's reporting, contemplate a public-market filing in the second half of 2026 with a 2027 listing. Discussed valuations have reached up to $1 trillion — which would make it the largest IPO in U.S. history by a significant margin. The company has also reportedly committed to roughly $600 billion in cumulative compute spending through 2030, a number that essentially requires public-market access to fund without violating Microsoft's appetite for further preferred-equity injections.
Anthropic, meanwhile, hit $19 billion in annualized revenue, growing roughly 10x year-over-year against OpenAI's 3.4x. Epoch AI's projection — which I suspect undercounts seasonal enterprise contract noise — has Anthropic crossing OpenAI's run rate by mid-2026. Whether or not that exact crossover happens, the directional read is clear: the spread between the two companies is compressing faster than the market expected even six months ago.
Why This Is a Categorical Shift, Not a Continuation
For two years, the working assumption among institutional capital was that OpenAI was a generationally large private company with no clear public-market path — too unprofitable, too dependent on Microsoft, too structurally unusual (the capped-profit hybrid) to list. That assumption has now broken. $25 billion in annualized revenue is not a startup number. It is a Fortune 100 number — comparable in scale to companies like Lockheed Martin or Goldman Sachs's investment banking division. At that revenue level, the Microsoft dependency starts to look less like an existential constraint and more like a commercial arrangement.
The structural questions don't disappear — the gross margins on inference are still being figured out, the capital intensity of the compute roadmap is unprecedented, and the regulatory environment for frontier AI is genuinely undefined — but they become questions investors price into a stock rather than reasons a stock can't exist.
What an OpenAI IPO Actually Means
If OpenAI lists at anything close to $1 trillion, it doesn't just mark the company's transition from private to public. It marks the moment AI as a sector becomes legible to ordinary equity investors. Right now, the way most retail and institutional capital plays AI is by buying the picks-and-shovels: Nvidia, AMD, the hyperscalers, the data center REITs. None of those are pure-play exposure to the underlying technology — they're plays on the build-out.
OpenAI as a public company changes that calculus. It would be the first major listed company whose entire valuation thesis is "this technology is going to remake the economy." Whether that's true at the price the market pays is the real question, but the price discovery itself becomes a kind of cultural event. Every quarterly earnings call would be a referendum on the trajectory of artificial intelligence as a commercial reality.
The Scenario That's Actually Worth Watching
Here's the real tension: an OpenAI IPO at $1 trillion in 2027 prices in continued dominance. But Anthropic's growth curve says the dominance has a finite shelf life. If Anthropic surpasses OpenAI on annualized revenue by mid-2026 — even briefly, even on a contract-cycle technicality — the IPO narrative becomes more complicated. Public-market investors don't reward duopoly with monopoly multiples.
The more interesting outcome is one where OpenAI lists, Anthropic doesn't, and Anthropic continues to compound private without the public-market governance overhead. That would invert the conventional read: the company with the marquee IPO becomes the one with the structural disadvantage, and the company that stays private becomes the smarter long-term position. Apple's iOS 27 move to open Siri to multiple AI providers — ending ChatGPT's exclusive distribution — is already a small data point in that direction.
The Bottom Line
OpenAI at $25 billion in annualized revenue and Anthropic at $19 billion is the headline. The subtext is that AI as a category has now generated enough revenue to support at least one and possibly two public companies of generational scale. Whoever gets to public markets first writes the template; whoever gets there second has to differentiate against it. The next twelve months will decide who runs which playbook.
Whatever happens, 2027 is going to be the year an AI company tests the upper bound of what U.S. public markets are willing to value technology at. The price will tell us more about the market than about the company.
Sources: The Information, Reuters via Yahoo Finance, Benzinga.
Comments