OpenAI is preparing to confidentially file a draft of its IPO prospectus as soon as this Friday, CNBC confirmed on Wednesday. Goldman Sachs and Morgan Stanley are leading the effort. If the timeline holds, this will be one of the fastest pivots from existential legal threat to public market debut in tech history.

Forty-eight hours ago, OpenAI was sitting in an Oakland courtroom waiting for a jury to decide whether Sam Altman had violated fiduciary duties, whether the company's entire $500 billion corporate restructuring should be unwound, and whether Elon Musk was owed $150 billion in damages. The jury needed less than two hours to say no to all of it.

That verdict didn't just clear a legal cloud. It was the starting gun.

The Speed Says Everything

OpenAI CFO Sarah Friar told CNBC last month that it's "good hygiene" for a company of OpenAI's size to "look and feel and act" like a public company. She wouldn't commit to a timeline. Now we know the timeline was: the moment the jury came back.

The confidential filing, known as a DRS-1, lets OpenAI submit its prospectus to the SEC without making it public. The company gets to negotiate with regulators behind closed doors, make revisions, and only go public with the document when it's ready for the roadshow. For a company whose financial details are already the subject of intense speculation, this is the smart play: control the narrative before Wall Street sees the numbers.

IPOX Vice President Kat Liu put it plainly: "Resolving that legal overhang removed a major obstacle to an IPO and likely gave OpenAI the confidence to accelerate its timeline."

The Numbers OpenAI Has to Sell

Private investors currently value OpenAI at more than $850 billion. The company is targeting a trillion-dollar public valuation. To justify that, it needs to show institutional investors a revenue growth story that can withstand scrutiny, and the story is complicated.

OpenAI gives 20% of its revenue to Microsoft through 2030. Its partnership with Apple, which was supposed to drive billions in ChatGPT subscriptions, has reportedly fallen apart so badly that OpenAI is preparing possible legal action against Cupertino. And the company burns cash at a rate that would make a defense contractor blush: model training, inference infrastructure, and a workforce that now includes some of the highest-paid AI researchers on the planet.

The bull case is straightforward: OpenAI has $55 billion in annualized revenue, dominates enterprise AI adoption, and just won the right to exist as a for-profit company in court. The bear case is equally straightforward, and it came out the same day from the same network.

The Timing Problem Nobody Wants to Talk About

Hours after CNBC broke the IPO filing news, the same outlet published a detailed analysis titled "Cheap AI could derail OpenAI and Anthropic's IPOs." The thesis: Chinese AI labs are matching frontier capability at a fraction of the cost. Claude costs nearly nine times more than the cheapest Chinese alternative for the same workload. On OpenRouter, Chinese models went from 1% of usage in 2024 to more than 60% in May 2026.

Even Google's Sundar Pichai is making the argument for cheaper AI, telling developers at I/O this week that his largest Cloud customers could save more than $1 billion a year by switching to the cheaper Gemini Flash model. When your competitors are actively selling the idea that frontier AI is overpriced, your IPO prospectus needs a very convincing answer.

OpenAI's answer, presumably, will be that it's not selling a commodity. It's selling the platform, the brand, the enterprise relationships, the distribution. That might work. But the S-1, whenever it becomes public, will need to address the margin compression that every enterprise customer is already experiencing. Meta, Shopify, Spotify, and Pinterest all flagged rising AI inference costs as a drag on margins this earnings season. Those are OpenAI's customers.

OpenAI is racing to go public while the valuation premium still exists. The question for investors isn't whether the company is impressive. It is. The question is whether $850 billion is the right price when the moat is being undercut from Beijing and Mountain View simultaneously.