Billion-Dollar AI Startups With Zero Revenue: How It Works

👁 114 reads

Imagine a company that has existed for eight months. It has forty employees, a sleek off-white website, and a demo that impresses at conferences. It has no product in customers’ hands. No contracts. Not a single invoice. Yet investors just valued it at a billion dollars and lined up to add more. This isn’t an anecdote about one specific company - it’s a template that has repeated so many times in the last two years that it no longer surprises anyone.

And that’s the most interesting part. Not that such valuations happen - because every bull market has its crazy moments - but that we’ve stopped considering them crazy. A billion without revenue has become the new normal, a reference point discussed in meetings with the same calm as office rent costs. I want to break down this mechanism into its component parts, because underneath lies something more than greed. There’s a quite rational, cold logic - which doesn’t necessarily mean everything will end well.

Where the Billion That Doesn’t Exist Comes From

Let’s start with something seemingly obvious that most people misunderstand. A startup’s valuation in a funding round is not the company’s value. It’s the price someone paid for a small piece of the company, multiplied by one hundred. If a fund puts in twenty million for two percent, on paper a “billion-dollar valuation” appears. But nobody bought the entire company for a billion. Nobody would pay that much for it. That number is an extrapolation from a single transaction at the margin, not the result of any market.

This distinction sounds like accounting pedantry, but it’s the crux of the matter. Journalists write “startup worth a billion,” investors happily echo that headline, and the public remembers that a company worth as much as a medium-sized airport has been created. In reality, an option has been created - a bet that in a few years this company will be worth many times more, or that you can resell it to someone before anyone asks uncomfortable questions about revenue.

In the AI world, this mechanism runs on steroids because it adds an element that wasn’t present in previous bubbles with such intensity: talent. A handful of people who worked on the most important models of the last decade can start a company and raise hundreds of millions before they write the first line of production code. An investor isn’t buying revenue then. They’re buying people who can’t be bought any other way, and doing it before the competition does.

Who’s Really Playing and What They’re After

To understand why this game continues despite the lack of revenue, you need to see that each participant is playing for something different than we think.

A venture capital fund isn’t playing for a given company to succeed. It’s playing for a portfolio. This industry’s model is based on nine out of ten investments being allowed to fail if the tenth returns a hundred-fold. In that math, paying a billion for a company without revenue isn’t madness - it’s buying a lottery ticket where the jackpot is a trillion. And since AI looks today like the most likely place where such a trillion will emerge, every fund without a prominent AI name in its portfolio must explain to its investors why it missed the revolution. Fear of missing out is as strong on the fund side as it is for a child scrolling social media.

Founders are playing for something different - for position in a race where the stakes are existence itself. Training a large model costs absurd amounts of money, mainly for computing power. If you don’t raise hundreds of millions now, you simply drop out of the game because you can’t afford the graphics cards that run this entire industry. A high valuation without revenue isn’t vanity for them. It’s a ticket to the table where you can play at all.

In this game, nobody pays for what the company does today. Everyone pays for the right to be nearby if it turns out the company does something that changes the world. It’s a bet on presence, not on product.

And the third player, easy to overlook, is the big tech companies. They often stand behind these rounds - not always directly with capital, but with credits for computing power in their clouds. It looks like an investment, but in practice it can be a clever way for a giant to finance a startup so that startup spends the money back with them, buying server access. Money circulates in a closed loop, and each circulation inflates the official numbers. This is where I start to feel uneasy - because part of the AI ecosystem’s “revenue” is essentially the same money we’re counting several times under different names.

Why Lack of Revenue Stopped Worrying Anyone

Here we come to the most fascinating part: the narrative that allows rational people to ignore something taught in first-year economics. Lack of revenue is explained today in three ways, and each contains a grain of truth, which makes them so effective.

First: “revenue will come later, because first you need to build the technology.” This is an argument from the history of many successful companies that burned cash for years before they started earning. The problem is that for every such company, there’s a graveyard of dozens of others that burned cash just as convincingly and simply disappeared. We remember the survivors, not the burned. This is classic survivorship bias, on which the entire startup mythology is based.

Second: “the market size is so enormous that even a small share will make a fortune.” Also true, and also a trap. The market can be gigantic, yet every company in it can lose if the cost of acquiring a customer is higher than what the customer will ever pay. In AI, this is a real risk because the cost of serving a single query - unlike classic software - doesn’t drop to zero at scale. Each use costs electricity and computing power. This means the more customers, the more the company loses if the price is too low. And the price is too low because everyone is fighting for users.

The third argument is the most insidious: “this is too important to miss.” And this one isn’t even economic - it’s existential. It assumes AI will change everything so deeply that normal measures of value stop applying. When I hear this argument, I’m reminded that exactly the same tone appeared with every major technological wave. “This time is different” is the most expensive sentence in investment history.

What Happens When the Music Stops

I’m not claiming AI is a fraud. I think the opposite - that it’s one of the most important technologies of my lifetime and that some companies getting crazy valuations today will actually build something lasting. That’s precisely why the distinction matters. A bubble and a breakthrough don’t exclude each other. Railroads were a breakthrough and a bubble simultaneously. The internet was a breakthrough and a bubble simultaneously. The tracks remained, the companies that laid them - mostly didn’t.

When I ask myself what will happen when some of these valuations turn out to be air, I see several layers of consequences. The shallowest is emotional: headlines about “the end of the hype,” disappointment, memes. That will pass quickly. Deeper is structural. The capital that flowed into the industry is largely other people’s money - pension funds, university endowments, insurers. When dozens of AI startups turn out to be worth a fraction of their valuations, losses will trickle down to people who never heard of any of these startups but had exposure to the “tech market” in their portfolios.

There’s another layer that worries me most, because it concerns not money but trust. Every bubble that bursts leaves cynicism behind. After it, good ideas are harder to finance because investors remember how they got burned on bad ones. Real, useful AI applications - those in medicine, education, tedious office work that excites nobody - may be punished for the sins of those who sold dreams without plans. That would be the worst possible irony: for technology that can truly help to lose funding because of those who treated it like a casino chip.

How to Tell a Company From Fireworks

I don’t have a crystal ball and I’m not going to point fingers at specific companies. But I have a few questions I ask myself when I read about another billion without revenue - and I think they’re worth knowing for anyone watching this.

Question one: does this company have something the competition can’t simply copy in six months? In a world where foundation models are increasingly accessible, the mere ability to use them isn’t a defensive moat. Question two: is anyone paying for what the company does, even a dollar - because the first dollar from a real customer says more than a hundred million from an investor. Question three, the hardest: does the economics of a single transaction have any chance of ever closing, or will the company lose more with each new customer.

These questions aren’t deep. They’re downright boring. And that’s precisely why in euphoria nobody asks them. Boredom is the last thing you want to hear at a table where everyone is playing for a trillion.

The story of a billion-dollar startup without revenue isn’t a story about AI. AI is just decoration, a costume this story put on this season. Earlier it wore the internet costume, then mobile apps, then cryptocurrencies. This is a story about how people react to something they don’t fully understand but that looks like the future - and how fear of missing that future can shut down common sense in people who professionally assess risk.

A billion dollars for a promise is neither smart nor stupid in itself. It’s a bet. And the thing about bets is that some of them win. The trouble is that in a casino where everyone simultaneously bets on the same color, convinced that this time really is different, someone at the end has to pay for the chips. And I’m afraid that, as usual, it won’t be those who dealt the cards.

// read moreA Billion Dollars for a Promise: AI Startup That Never SoldMagazineTwenty Dollars and the Year I Regret WastingMagazineHow AI Is Quietly Eliminating Jobs Without LayoffsMagazine

Kinetyra