OpenAI Is Going Public. But Here’s the Complicated Part

On June 9th, OpenAI did something quietly dramatic. It submitted a confidential S-1 to the US Securities and Exchange Commission — and then immediately told the world it had done so. “We expect it to leak,” Sam Altman said, “so we’re just announcing it.”

That line tells you almost everything about OpenAI’s relationship with its own story. The company was born as a nonprofit, pivoted into one of the most valuable private companies in history, and is now preparing for what could be the largest technology IPO since the dot-com era. And Sam Altman knows the narrative is going to be picked apart the moment those financials go public.

So what do we actually know? And what should you, as an investor, be paying attention to?

Let’s start with the headline number

OpenAI’s last private funding round, closed in March 2026, valued the company at $852 billion. That’s the post-money valuation after raising $122 billion — the largest single private funding round ever recorded. Analysts covering the listing expect the IPO itself to target a valuation north of $1 trillion.

For context, that $1 trillion figure would make OpenAI’s debut roughly four times larger than Alibaba’s landmark 2014 listing. The underwriters are Goldman Sachs and Morgan Stanley — two names that signal this isn’t a drill.

The revenue story sounds incredible

Revenue at OpenAI has grown at a pace that would make most Silicon Valley founders weep with envy. In 2023, the company was doing roughly $2 billion in annualized revenue. By February 2026, it was generating approximately $2 billion per month. That’s a 12x increase in a little over two years — faster than Google, faster than Meta, faster than almost any software company that’s ever existed.

By 2030, the company has told investors it expects to reach $280 billion in annual revenue, split roughly evenly between consumer and enterprise. ChatGPT now counts over 900 million weekly active users globally. The product penetration, on paper, is extraordinary.

That’s the bull case in one sentence. Growth is real. Scale is unprecedented. And the vision — transforming ChatGPT from a chatbot into a productivity layer across everything you do — is coherent.

Now for the part that keeps CFOs up at night

OpenAI has never turned a profit. And based on what’s been shared with investors, it doesn’t expect to be cash-flow positive until 2030. In 2026 alone, the company is projected to lose approximately $14 billion. Here’s why the math looks the way it does.

The single biggest cost is compute — the GPUs, data centres, and electricity required to run ChatGPT at global scale and train the next generation of models. OpenAI has committed to spending approximately $600 billion on compute through 2030. Late last year, Sam Altman had been quoting $1.4 trillion. The revised number reflects a strategic pivot: instead of owning all that infrastructure through the Stargate joint venture, OpenAI is now renting significant compute capacity from Microsoft Azure, Amazon Web Services, and others. The debt, essentially, sits with Oracle and others who built it.

There’s also a more structural problem that every investor should understand: inference costs. Running an AI model for a user — every chat, every query, every image generation — is called inference. And inference costs for OpenAI grew fourfold in 2025. That’s what compressed gross margins from 40% to 33% in a single year. The more people use ChatGPT, the more it costs to serve them. That is the opposite of how traditional software economics works.

Compare this to how great software companies have scaled

The dream of every software investor is a business where adding the next customer is almost free. WhatsApp added 100 million users on a skeleton team. Google’s 10th advertiser costs nearly the same to serve as the 10-millionth. This compounding of margin with scale is what creates the legendary returns in software.

OpenAI’s 33% gross margin looks more like a semiconductor company or a cloud infrastructure provider than a software platform. The question for every investor — and Sam Altman knows this — is whether that margin can expand meaningfully as the business scales. The company is betting yes: that future model efficiencies, inference optimisation, and pricing power will flip the unit economics. That’s a bet, not a certainty.

The structure of this company is unlike anything public markets have seen

Here’s the part that rarely gets explained clearly. OpenAI is not an ordinary corporation. It was founded as a nonprofit in 2015. In October 2025, after almost a year of negotiations, it completed a restructuring that turned its commercial arm into a “public benefit corporation” — OpenAI Group PBC — while keeping a nonprofit entity called the OpenAI Foundation in control.

What does that mean in practice? The Foundation holds a 26% equity stake in the commercial entity, valued at approximately $130 billion. Microsoft holds roughly 27%. Current and former employees and investors hold the remaining 47%. And crucially, the Foundation continues to control the board — meaning it can appoint all directors of OpenAI Group. The mission of beneficial AI can, in theory, override decisions that would otherwise maximise shareholder returns. This has never been tested at a publicly listed company of this scale, and the S-1 will need to explain to public investors exactly what rights they are purchasing.

The competitive picture is getting tighter

One of the most revealing data points of 2026 has nothing to do with financials. It’s about market share. ChatGPT’s share of the AI chatbot app market fell from 69% in January 2025 to 45% by early 2026. Over the same period, Google’s Gemini climbed from 15% to 25%. Elon Musk’s Grok went from near-zero to 15%. These are consumer metrics, not enterprise metrics, but they matter for the IPO story because OpenAI has always been defined by ChatGPT’s dominance.

Sam Altman internally declared a “code red” in December 2025, ordering staff to reprioritise core ChatGPT quality improvements over newer projects like an AI wearable and an advertising platform. The framing from OpenAI’s own leadership is that the company is at a “precarious moment” — a term used openly by market analysts covering the listing. The competitive gap between OpenAI, Google DeepMind, Anthropic, and Meta is narrowing faster than anyone expected two years ago.

Three things to watch when the S-1 goes public

1. The Microsoft revenue-share terms- OpenAI and Microsoft renegotiated their deal in May 2026, capping revenue-share payments to Microsoft at $38 billion through 2030 — down from an earlier projected $135 billion. The exact economics of this relationship (cloud spend, IP licensing, compute costs) will be one of the most scrutinised sections of the S-1. How dependent is OpenAI on a single partner that also competes with its products?
2. The gross margin trajectory – Will inference costs continue to rise with usage, or does OpenAI have a credible roadmap to expand margins toward 50–60%? Investors will want to see model efficiency improvements, proprietary chip development, and evidence that the cost-per-query is falling as models improve. Without a convincing answer here, a $1 trillion valuation is very difficult to defend with discounted cash flows.
3. The governance section- The S-1 must clearly disclose what shareholder rights actually look like under PBC and Foundation control. Does the Foundation have veto rights over major commercial decisions? What happens if the board decides a product line conflicts with the AI safety mission? These aren’t hypothetical. They’re the legal architecture of the company you’d be investing in.

So why go public now?

Interestingly, Sam Altman has been somewhat reluctant about the IPO process. “We have not decided on timing yet,” OpenAI said in its own announcement. “It may be a while because there are things we want to do that are likely easier as a private company.” That’s a remarkable admission for a company simultaneously filing to go public.

The honest answer is capital. Frontier AI is not cheap. The cumulative cash burn through 2029 is projected at $112 billion. OpenAI doesn’t have a lot of other places to find that kind of money, and public markets — if the reception is warm — represent a theoretically unlimited pool. The timing also reflects a broader AI IPO moment: Anthropic filed its own S-1 the week before OpenAI, and SpaceX is about to begin trading. There’s a window right now, and Sam Altman is choosing to keep the option open.

The real test isn’t the filing. It’s whether public markets, which price things quarterly, can reconcile with a company that openly admits it won’t be profitable until 2030 and is betting on infrastructure commitments that dwarf any technology company in history.

That reconciliation — between the hype and the math — is what this IPO is really about.