OpenAI's IPO pitch is brutally simple: public investors are being asked to buy extraordinary growth before the company has proved it can make that growth pay.
There's a number in OpenAI's pre-IPO financials that you should sit with for a moment. The company spent about $34 billion in 2025 while bringing in about $13 billion of revenue. The Financial Times reported those figures today, citing audited financials, and they make the coming listing much harder to treat like a normal software IPO.
OpenAI confidentially filed its S-1 on June 8, according to the company's own announcement and reporting from The Guardian. The company didn't give investors a date. Its post said the timing hadn't been decided and that "it may be a while," which is a useful line to remember whenever bankers or rivals start talking as if September is already locked in. The filing gives OpenAI the option to move. It doesn't make the business any less expensive to run.
The headline valuation is the easy part to understand. OpenAI was last valued at $852 billion after a huge private funding round, and the public market chatter now runs up to $1 trillion. At 2025 revenue of $13 billion, you're looking at a multiple that only works if growth remains violent for years and if the cost structure eventually bends. Investors aren't buying a calm, cash-generating company. They're being asked to fund a race.
To be fair to the bull case, the growth is real. The FT reported that OpenAI moved from about $1 billion a quarter in revenue at the end of 2024 to $2 billion a month by the end of 2025. ChatGPT is no side project sitting inside a research lab anymore. It is a consumer product, an enterprise tool, a developer platform and, increasingly, a public-market story that will pull in anyone who wants direct exposure to frontier AI rather than buying Microsoft, Nvidia or Alphabet as a proxy.
But the loss figure needs careful handling. The same FT report said OpenAI's net loss attributable to the company rose to roughly $39 billion in 2025, and that the figure was heavily affected by a roughly $30 billion non-cash charge tied to its former investor structure before its public benefit corporation shift. Strip out that charge and other non-cash items, and the operating loss was closer to $8 billion. That's still an enormous amount of money. It just isn't the same as saying the core business burned nearly $39 billion in cash.
The cost problem is the story
Here's the thing about OpenAI's spending: some of it builds a moat, and some of it is simply the admission fee for staying near the front. The FT said about $19 billion went to research and development in 2025, with nearly $6 billion more spent on sales and marketing. You can't cut that R&D line casually when Google DeepMind, Anthropic and Meta are training models of their own. In frontier AI, thrift can quickly become surrender.
That doesn't make the spending harmless. According to Reuters and CNBC reporting earlier this year, OpenAI told investors it was targeting around $600 billion in compute spending through 2030, down from an earlier figure of about $1.4 trillion. Read that again. The more reassuring number is still $600 billion. If you run a startup, you know what this means in plain terms: the company has to keep raising money long after the IPO bell rings.
The Microsoft relationship cuts both ways too. Microsoft gives OpenAI distribution, cloud capacity and a financial backer with the balance sheet to keep the machine running. It also reminds you that OpenAI doesn't own the whole stack in the way Google, Amazon or Meta do. When compute is the product, paying someone else for a large share of it is not a footnote. It's a structural fact.
OpenAI's defenders will point to the revenue projections, and they have something to work with. CNBC reported that OpenAI has shown investors a path to about $280 billion in revenue by 2030, split across consumer and enterprise businesses. Axios has also reported that OpenAI expects advertising to become a major revenue line, with projections of $2.5 billion in ad revenue this year and $100 billion by 2030. Frankly, that is where the business starts to look less like a pure AI subscription story and more like the next consumer internet platform trying to make the math close.
Public investors should be clear about what they're being offered. This isn't a bet on whether ChatGPT is important. It plainly is. It's a bet on whether OpenAI can turn scale into margins before the infrastructure bill forces another round, another partner concession, or another rethink of the product. The company may get there. But at anything close to $1 trillion, you aren't being paid for certainty. You're being asked to pay for it.
Also read: The Hormuz deal Trump announced may not hold and the world economy is not finished with this crisis, STMicroelectronics raises $1.5 billion in convertible bonds as its AI photonics bet reshapes how the market sees the company, Singapore bets on AI to crack the materials science bottleneck holding back semiconductors and clean hydrogen
