OpenAI files to go public. The AI era is about to get a market price.
OpenAI filed confidentially for an IPO targeting a ~$1 trillion September listing. The real story isn't the first-day pop — it's that an S-1 finally puts numbers under oath on the most capital-intensive software ever built.
OpenAI has filed confidentially with the SEC — as early as today — for an IPO targeting a September listing, with Goldman Sachs and Morgan Stanley running the deal. At its $852 billion private valuation, and with chatter of a $1 trillion debut, it would be the largest technology IPO in history. The AI era is about to get something it has never had: a public-market price.
What's actually happening
A confidential filing isn't the IPO; it's the starting gun. The public S-1 is expected in late July or early August, with the listing penciled in for September. OpenAI runs at roughly $25 billion in annualized revenue.
It's not alone. SpaceX filed its public S-1 on May 20 and is targeting a June listing under the ticker SPCX. Anthropic is aiming for October at a reported $900 billion valuation. Prediction markets put the odds that OpenAI lists before Anthropic around 85%. After three years of private megarounds, 2026 is the year the AI giants meet the stock market.
Why now
Read the IPO as a financing necessity, not a victory lap.
Frontier AI is the most capital-intensive software ever built, and the cost center is compute. You can raise tens of billions privately for a while — SoftBank, sovereign funds, the Microsoft relationship — but eventually the cheapest, deepest pool of capital for building data centers at this scale is the public market. Going public is how OpenAI funds the next phase without running out of private whales.
The more interesting consequence is disclosure. For three years the AI boom has run on private valuations and conviction. An S-1 puts numbers under oath: the real revenue, the real losses, the real compute bill, the real growth curve of ChatGPT. Whatever you believe about OpenAI, you're about to find out if it's true.
My read
A ~$1 trillion price on ~$25 billion of revenue is a ~40x multiple on a business that loses a great deal of money turning electricity into tokens. That's the market pricing the future, not the present — fine, until rates move or a model generation underwhelms and "the future" gets re-rated.
The bull case writes itself: OpenAI is the consumer default for AI, and the S-1 will likely show ChatGPT revenue compounding at a rate that makes the multiple look cheap in hindsight. The bear case is in the same document: the cost of staying at the frontier tends to rise as fast as revenue, and a public market that has seen one capex supercycle after another will ask, pointedly, when the compute bill stops growing faster than the top line.
And there's a structural tension a roadshow can't paper over. OpenAI's capped-profit-under-a-nonprofit governance was explicitly designed to restrain the shareholder-maximization a public listing demands. You cannot serve a mission and an earnings call with the same hands indefinitely. The IPO is where that contradiction stops being a philosophical footnote and becomes a quarterly obligation.
For everyone building on top of OpenAI, that's the part to watch. The day it becomes a public company is the day it acquires a cost of capital and a fiduciary duty — which changes what it optimizes for: pricing, enterprise, margin, predictability. Watch the S-1, not the first-day pop. The numbers are the story, and we're finally going to see them.
