OpenAI Prepares for IPO with 810 Million Users and Massive Losses
Sam Altman has publicly stated that he is "zero percent excited" about taking OpenAI public. Yet, the company hired Sarah Friar, the executive who led Square and Nextdoor to their respective public market debuts, as CFO. Additionally, former DocuSign CFO Cynthia Gaylor has been brought on board to lead investor relations. OpenAI is also in talks with several Wall Street banks to structure an IPO projected for the fourth quarter of 2026. When a CEO's words point in one direction while hiring decisions point in the opposite, actions speak louder.
What’s happening with OpenAI is one of the most intriguing cases I’ve seen in years: a company boasting over $20 billion in annualized revenue, 810 million monthly active users, and one million business clients, yet still burning cash at a rate its founders describe as structurally unsustainable without ongoing access to external capital. The question is not whether the IPO will happen; it’s already in motion. The question that no headline seems to answer is: what exact business model are they selling to future shareholders?
The Front Stage Business vs. The Back Stage Operations
OpenAI’s numbers are, at first glance, the kind any investment banker would dream of projecting on a slide: revenue growth at triple-digit rates, valuations between $730 billion and $850 billion in recent private rounds, and backing from Amazon, SoftBank, Nvidia, and Microsoft. The narrative is that of a dominant tech platform that arrived first, embedded itself in popular culture, and is turning that scale into a recurring revenue machine.
However, the problem lies a layer deeper. OpenAI operates on a computing infrastructure whose costs exceed what any software company faced prior to the era of large-scale language models. Training and operating these models requires clusters of cutting-edge chips, long-term power contracts, and cloud agreements worth tens of billions. The company has projected an $1.4 trillion infrastructure plan over eight years, including data centers, proprietary chip design, and energy procurements. This is not a software company with 70% margins; this is a heavy infrastructure company that will list with software multiples.
The difference matters. An institutional investor purchasing OpenAI stock at the IPO is not buying a business with variable costs and a lean structure. They are betting that the demand for artificial intelligence will indefinitely justify a massive fixed-cost structure. That bet may pay off. But the market needs to clearly understand what it is purchasing, and the narrative of a "productivity platform" that the company is building internally, with a mandate to position ChatGPT as a work tool, does not resolve the underlying arithmetic.
Productivity Mandate as a Signal, Not a Promise
The internal directive that ChatGPT must evolve into a "productivity tool" isn’t a product vision statement. It’s a signal to the capital markets. Institutional investors evaluating a company for an IPO can distinguish between speculative revenue and recurring business contracts. Positioning ChatGPT as work infrastructure rather than a viral consumer product shifts the valuation narrative: from social network with uncertain scale to enterprise software platform with predictable contracts.
This reframing has direct financial logic. OpenAI already has a million business clients, a figure that justifies talking about recurring revenues with a degree of certainty. However, there’s a tension that roadshows will need to transparently resolve: the same company that seeks to sell business stability also plans to integrate advertising into ChatGPT to diversify revenue. These two models don’t necessarily contradict, but they do create user experience frictions that product teams will need to manage with surgical precision. A corporate client paying for a work tool doesn’t expect the same flow of interruptions they accept in a free social network. Finding that balance is not a mere communication exercise; it’s a retention experiment that lacks validated results.
What is validated is the scale of demand. 810 million monthly active users are not a hypothesis. They are a base on which monetization models can be built, as long as the company identifies precisely which segment of that base is willing to pay prices that support the cost structure. So far, most of the growth has been based on free or low-cost access. The transition to revenues that cover a $1 trillion infrastructure requires a leap in the value proposition that the current numbers do not fully certify.
Power Dynamics Behind the IPO
There’s something that financial analyses of OpenAI’s IPO tend to undervalue: the power dynamics among its main shareholders and what each needs from this operation. Microsoft has enjoyed a privileged position in OpenAI for years, trading access to technology for cloud infrastructure. Amazon is negotiating an investment of up to $50 billion. SoftBank and Nvidia also hold significant positions. Each of these players has different incentives regarding the timing and structure of the IPO, and their interests are not identical to those of future retail shareholders.
The most significant structural change has already quietly occurred: OpenAI shifted from its original structure as a nonprofit organization to become a public-benefit corporation. This move, which many read as a philosophical pivot, was primarily a financial plumbing operation. Without this transition, none of the recent funding rounds would have been possible, and the IPO would have been legally unfeasible. The narrative of "benefit to humanity" survives in its legal name, but the governance mechanism now operates under the logic of capital return demanded by any public market.
That’s not necessarily negative. Public markets have accountability mechanisms that private capital does not enforce with the same rigor. Once OpenAI begins presenting its quarterly results, scrutiny over the relationship between revenues, infrastructure costs, and progress toward profitability will be systematic and transparent. This forced level of transparency could be the most useful pressure the company faces since its founding.
The IPO Doesn’t Validate the Model, It Exposes It
OpenAI’s IPO in the fourth quarter of 2026, if it happens, will not be the moment when the business model is proven. It will be the moment when it is entirely exposed. So far, valuations between $730 billion and $850 billion have been constructed in private rounds where information is selective, and investors negotiate with asymmetrical access. The public market operates differently: it demands quarterly disclosures, subjects numbers to thousands of analysts simultaneously, and punishes without delay the gaps between expectation and reality.
The history of major tech IPOs shows a consistent pattern: investors purchasing on the first day of trading frequently see returns lower than those who held positions in earlier rounds. Not because the business is bad, but because the exit price already discounts the optimism. With a target valuation that could exceed $1 trillion, OpenAI will need to demonstrate, quarter by quarter, that its infrastructure costs are turning into competitive advantages and not structural burdens.
Hiring Sarah Friar and Cynthia Gaylor suggests that the management team understands this. Friar has direct experience managing public market pressure in contexts of high growth and sustained losses. That’s not a decorative profile; it’s a sign the company is preparing for a radically different scrutiny regime than it knows.
The only business model that survives that scrutiny is one built with decisions made in constant contact with market reality, frequently adjusted, and unanchored from Excel spreadsheets projecting the future from a boardroom. Companies that go public with untouched financial hypotheses, never corrected by the friction of the paying customer, are those that suffer the worst post-IPO quarters in history. Leadership that understands this does not celebrate IPO day; it treats it as the beginning of the most demanding validation it will ever face.
