What Is OpenAI Worth?
So you want to profit from the AI revolution?
Sam Altman wants a trillion dollars. Wall Street is polishing its pitch books. The press is already writing the superlatives. Before the bankers start printing prospectuses, let’s answer the question every retail investor should be asking: should I buy this, and if so, at what price?
The safest answer is also the simplest one. Buy an S&P index fund. You don’t pick the winner. You own collectively the AI revolution. You capture Nvidia’s chip dominance, Microsoft’s enterprise distribution, Google’s search transition, Amazon’s cloud infrastructure, and whatever emerges from the open source world. You also capture the efficiency gains rippling through every industry that adopts AI — healthcare, finance, manufacturing, logistics — as those savings flow straight to corporate earnings and stock prices. You do not bet on which consumer AI subscription survives the next five years of a brutal, capital-intensive war of attrition. The AI revolution happens. You make money. Nobody gets hurt.
If that is too boring, read on. Because the math on OpenAI specifically is more interesting than the headlines suggest.
The Right Question
Every valuation story you will read about the OpenAI IPO uses revenue multiples. Price-to-sales. Forty times revenue. A hundred times revenue. The argument is always the same: the growth rate justifies the multiple.
It doesn’t. And here’s why.
OpenAI is not a growth company in any meaningful sense. The consumer AI layer it is building behaves like a mature subscription market at scale. The market it is selling into — human beings who want AI access — is fixed. It grows only as fast as the global population grows, plus inflation. There is no new market to create. There is no untapped geography to conquer. Everyone on earth who can afford a phone is the market.
You do not value a fixed market with a growth multiple. You value it the same way you value AT&T or American Express. With earnings. With a price-to-earnings ratio appropriate for a mature, stable, recurring revenue business.
That is what we are going to do here. With numbers that exist today. Nothing else.
The Market
OpenAI is, at its core, a consumer AI company. ChatGPT has 910 million weekly active users. People don’t say they are going to use an AI assistant. They say they are going to ask ChatGPT. That is what brand dominance looks like. Google took twenty years to become a verb. ChatGPT did it in two.
The right comparable for this business is the mobile phone.
Today there are 9.2 billion mobile phone subscriptions worldwide. Not unique users — subscriptions. Lines being paid for every month. That number grows only with population and economic development. It is a mature, universal, utility market.
Consumer AI will get there. The question is not whether people will pay for AI access — they already are. The question is what they will pay and how many of them there are.
Ten dollars a month is a reasonable estimate. It is roughly what a basic phone plan costs in most of the world. It is cheap enough to be a rounding error in a household budget. Expensive enough to generate real revenue at scale. Call it a guess — but it is an informed one, anchored to what people already pay for utility access to a network.
The math is simple.
9.2 billion subscriptions × $10/month × 12 months = $1.1 trillion in annual revenue.
That is the consumer AI market. Every person on earth with a phone, paying ten dollars a month. One number. No speculation about displacement rates or labor market disruption or how many lawyers AI will replace. Just: this is the market if AI access becomes as universal as the phone.
The Earnings
Revenue is not the number that matters. Earnings are.
OpenAI’s gross margin today is 33%. Inference costs — the computing power required to answer every query — eat the rest. That is the number we have. That is the number we use.
$1.1 trillion × 33% = $363 billion in gross profit.
That is the earnings number. The number a rational investor should be applying a multiple to.
The Multiple
This is a consumer subscription utility. It grows with population. It grows with inflation. It does not grow because AI gets smarter — that is already priced into the subscription. It does not grow because white collar workers get displaced — that is the corporate market, a separate business with a separate customer and a separate conversation.
The right comparable is AT&T or American Express.
AT&T trades at 13.6 times earnings. Pure utility infrastructure. Essential, universal, nobody cancels, grows at roughly the rate of the economy.
American Express trades at 21.9 times earnings. A premium consumer franchise with a loyalty moat. Slightly more dynamic, but still mature and bounded.
Consumer AI sits between them. Call it 15 to 18 times earnings.
$363 billion × 13.6x = $4.9 trillion $363 billion × 18x = $6.5 trillion $363 billion × 21.9x = $7.9 trillion
The range for the entire consumer AI market, priced like what it actually is, is $5 to $8 trillion.
That is the whole market. Every human on earth. One company. No competition assumed.
What You Are Paying
One trillion dollars.
Against a market of $5 to $8 trillion, $1 trillion is 12 to 20 cents on the dollar.
For a company that already has 910 million weekly active users — roughly 10% of the total addressable market already in hand — that is not obviously crazy. It is a reasonable entry point if you believe consumer AI becomes as universal as the phone and OpenAI is still standing when it does.
At $1 trillion, OpenAI is not wildly overpriced. It is priced at a fraction of what the market is worth if AI access becomes infrastructure — which, based on everything we can observe today, it will.
The Conclusion
Based on what we know today, $1 trillion is a defensible price for OpenAI.
It is not cheap. It assumes the consumer AI market reaches full penetration, that $10 a month becomes the global price of AI access, and that OpenAI captures and holds that market the way AT&T holds its phone subscribers. Those are large assumptions. But they are not fantastical ones.
One number worth keeping in mind: of OpenAI’s 910 million weekly active users, only 50 million are paying subscribers today — 0.54% of the total addressable market. That looks like a problem. It isn’t. It is the pipeline. The free tier exists to get people hooked and to achieve market dominance. People become addicted to it, or more precisely, it becomes an essential part of their lives — the same way a camera became essential not because anyone decided to buy a camera, but because it came with the phone. Once ChatGPT is that embedded, the free tier shrinks and the price goes up slowly, the boiling frog model, the way a frog does not notice the water heating. The 860 million convert — not because they want to pay, but because they cannot imagine not having it. The market the math describes is real. The bridge from here to there is human nature.
Right now, OpenAI has no meaningful competition for world dominance in the consumer AI market. That will change. Competition will come. But there is no way to estimate when, from whom, or at what cost — and anyone who tells you otherwise is guessing. So we leave it out. The model is built on what exists, not on what might.
The IPO is now. The unknowns — competition, margin trajectory, inference cost curves, regulatory risk — are real but unquantifiable today. If you need those numbers before you can get comfortable, the OpenAI IPO is not for you. You cannot value what cannot yet be valued. Buy the S&P and wait. That is not a bad outcome.
If you can live with the math we have — known market, known margin, known comparables — then $1 trillion is not a bubble. It is a bet on a market that demonstrably exists, priced at a fraction of what that market is worth. As the industry matures, the model can be refined. Right now, today, at IPO time, the honest math produces a range of $5 to $8 trillion for the whole market and a $1 trillion ask for a company already holding 10% of it.
Whether it pays off is a question for another day.
If you want to subtract China, where local competitors dominate and OpenAI has no real path to market, knock off roughly 1.7 billion subscriptions — the market drops to about $4 to $6.5 trillion and $1 trillion is still 15 to 25 cents on the dollar.