Sam Altman has wanted to take OpenAI public for a long time. This week, he is finally doing it, over the objections of the person whose job is to make the numbers add up.
OpenAI is preparing to confidentially file a draft IPO prospectus with the Securities and Exchange Commission as soon as Friday, May 22, with Goldman Sachs and Morgan Stanley running the deal, according to a report from CNBC. The company is targeting a public debut as early as September at a valuation that could top $1 trillion. If it lands, it would rank among the largest IPOs ever attempted.
A confidential filing means the public does not see the paperwork yet. Under SEC rules, a company going public can submit a draft registration statement privately, work through regulators' comments away from the spotlight, and only reveal the full financials weeks before trading begins. So the document itself stays sealed. The decision to file does not.
For anyone who builds on OpenAI's models, this is the moment the company stops being a private research lab answerable to a handful of investors and starts becoming a public company answerable to the quarter. That shift will shape model pricing, release cadence, and the cost of the API long after the confetti settles. And inside the company, not everyone agreed it should happen now.
The Filing Is the Easy Part
The mechanics are straightforward, and OpenAI spent the past year clearing the way for them.
The decisive move came in October 2025, when OpenAI converted from a nonprofit-controlled capped-profit structure into a Public Benefit Corporation. That change is what made a conventional stock listing legally possible. Without it, there were no normal shares to sell.
The capital was already in place. In February, OpenAI closed the largest private funding round in history, raising $110 billion at a 730 billion-dollar pre-money valuation, then expanding it to a post-money mark of 852 billion dollars.
Microsoft, which ended its exclusivity arrangement with OpenAI in April, holds a 26.79 percent fully diluted stake, worth roughly $228 billion at that valuation. Altman himself holds no equity in the restructured cap table.
A jump from an 852 billion-dollar private mark to a public valuation above $1 trillion is aggressive but not absurd, given the trajectory. The harder question is whether the business underneath can survive the disclosure that comes with being public.
The CFO Counseled Patience
Sarah Friar, OpenAI's chief financial officer, has reportedly pushed back on Altman's timeline. Her concern is not the demand for ChatGPT. It is the gap between what OpenAI has promised to spend and what it currently earns.
The revenue is real and growing fast. OpenAI's annualized revenue rate climbed from roughly 20 billion dollars at the end of 2025 to about $25 billion by February 2026. But annualized run rate flatters a company growing this quickly. Full-year 2025 actual revenue came in near 13.1 billion dollars, a figure that looks small next to the commitments stacked against it.
Those commitments are staggering. OpenAI is on the hook for roughly $600 billion in compute over five years, spanning semiconductors and data centers. The contrast is the entire problem a public-market investor will fixate on:
| Metric | Figure | Period |
|---|---|---|
| Full-year revenue (actual) | ~$13.1 billion | 2025 |
| Annualized revenue run rate | ~$25 billion | February 2026 |
| Compute commitments | ~$600 billion | Five years |
| Last private valuation | $852 billion | March 2026 |
Friar has reportedly warned that sluggish revenue could limit the data center buildout, and that the company is behind on the financial reporting readiness a public company needs. Going public on an aggressive schedule means subjecting that mismatch to quarterly earnings calls, analyst models, and short sellers. Altman wants the listing now. His CFO wanted more time to get the house in order.
Going Public Changes the Product, Not Just the Cap Table
For practitioners, the abstract becomes concrete fast once a company trades publicly.
A public OpenAI faces pressure to show a path to profitability on a clock it does not fully control. The most direct levers are the ones developers feel: API pricing, rate limits, the speed at which older models are deprecated, and how aggressively free usage gets pushed toward paid tiers. None of that is hypothetical. It is the standard playbook for a high-growth company that has to satisfy public shareholders while burning cash on infrastructure.
The IPO also formalizes the race that has defined the past year. Anthropic is reportedly preparing its own public listing and seeking fresh capital at a valuation near $900 billion, potentially topping OpenAI's own private mark. OpenAI filing first puts pressure on its closest rival to follow, and gives enterprise buyers a public scoreboard to judge both labs against. When the company you depend on for inference has to publish its numbers every 90 days, those numbers become part of your own planning.
The Other Side Sees a Company Ready to Run
The bull case is not hard to make, and OpenAI's bankers will make it loudly.
Revenue is compounding at a rate almost no company its size has matched. ChatGPT remains the most-used AI product in the world, the enterprise business is growing, and the demand problem most companies face simply does not apply here. Supporters argue that locking in public capital now is the prudent move precisely because the compute bills are so large: an IPO gives OpenAI a currency, its own stock, to fund the buildout without endlessly returning to private investors.
The legal cloud has also lifted. A jury recently ended Elon Musk's long-running lawsuit against OpenAI, removing an overhang that had dogged the company's restructuring for more than a year. With the courtroom risk gone and the PBC conversion done, the bankers' view is simple: the window is open, demand is peaking, and you list into strength, not weakness.
The skeptics' reply is the one Friar reportedly made internally. A company that earns tens of billions and has promised to spend hundreds of billions is making a bet on a future that has to arrive on schedule. Public markets are unforgiving when it does not.
The Bottom Line
OpenAI is about to do the thing it was structurally built, over the past year, to be able to do. The nonprofit became a Public Benefit Corporation, the largest private round in history filled the war chest, and the Musk lawsuit cleared. Friday's filing is the logical last step.
What it cannot resolve is the gap at the center of the story. OpenAI has the revenue of a very large software company and the spending plans of a national infrastructure program. The IPO does not close that gap. It simply moves the question of whether OpenAI can close it from a boardroom in San Francisco to a ticker that anyone can short.
Altman bet that the moment to go public is now. His own CFO bet that it wasn't. One of them is about to be proven right in front of everyone.
Sources
- OpenAI to confidentially file for IPO as soon as Friday: Source — CNBC, May 20, 2026
- OpenAI preparing confidential US IPO filing, aiming for September listing — WION, May 21, 2026
- OpenAI prepares confidential IPO filing with Goldman Sachs and Morgan Stanley — Crypto Briefing, May 20, 2026
- The Trillion-Dollar IPO Test: SpaceX and OpenAI Face Public Markets — Investing.com, May 2026
- OpenAI IPO Q4 2026, $852B valuation, Musk lawsuit, governance — TradingKey, May 2026
- Sam Altman to Receive No Stake in OpenAI's New Structure — PYMNTS, 2026
- OpenAI revenue, valuation and funding — Sacra, 2026