The rules of the IPO game have changed, and the change starts with where the money now lives. A broad shift toward passive investing has concentrated capital in a handful of giant index vehicles: the S&P 500, the NASDAQ 100, the triple Qs. People who want exposure to the market without thinking about it have poured money into these funds. On average markets rise, businesses improve, people fix problems, and that arrangement has worked well for ordinary investors. It has also produced strange effects. Index funds do things that are sometimes smart and sometimes dumb, simply because they buy mechanically. Those peculiarities first showed up in individual stocks. Now they are showing up in IPOs.
Manufactured Demand and the Death of Price Discovery
The core mechanic is straightforward. Large index funds rush to buy new listings before the deals are seasoned and before the companies have proven themselves as good companies. They buy because the company is big, not because it is good. That buying creates demand almost the moment a stock lists.
When index funds become major buyers immediately, price discovery disappears. The funds are simply buying. That does not guarantee these stocks go up, but it does mean they are manufactured to go up. Put a limited float into the deal, drum up interest among both professional and retail investors, and then layer on the indices that are required to buy the name in size. Even with a tiny float, there are waiting buyers whether the company deserves them or not. Real price discovery never happens, because there are buyers standing there when there are not yet any sellers.
SpaceX is the clearest recent example. Only about 4.3% of the shares are actually out there, or roughly 4.9% if you include the green shoe. Demand has been overwhelming from every direction at once: people who believe it is a good company, and people who think it is a lousy company but expect the stock to rise anyway. Why people buy these things is genuinely puzzling, but between that appetite and the indices forced to buy, there are buyers with no matching sellers. That imbalance will not last. It changes fast, and for SpaceX it changes soon, because lockups and share unlocks are arriving very quickly.
What SpaceX Actually Engineered
Whether this approach counts as effective depends entirely on where you sit. What is good for SpaceX, or for a trillionaire like Elon Musk, is not automatically good for everyone else. It might be, but that judgment turns on what you think an IPO is for. If the purpose is to bring a company public, let ordinary and professional investors buy in, and compel a large company to actually sell shares, then the SpaceX structure looks like something else. There used to be rules forcing companies past a certain number of shareholders to either sell to the public or cap their shareholder count.
The SpaceX deal was built to go up, and it worked. Only a sliver of shares went out, precisely to create heavy demand against thin supply. The promotion around it was unlike anything seen before in decades of studying these deals: executive interviews, road shows, tweets from the CEO during what would normally be the quiet period. When Google went public, none of that would have happened. Here it was everywhere. Limited float, a concerted push to build demand, and the indices obligated to buy. The thin float was designed to lift the price and hold it there.
Then came the financing on top. The company borrowed $2 billion. That raise had little to do with the underlying health of the business. This is a money-losing operation growing at 15% year over year. There are hopeful pieces inside it, and there are ugly ones, including the collapse of what used to be Twitter now sitting inside the SpaceX structure. The bond raise rode on the back of a successful equity raise, which itself rode on the limited float that drove the stock up. Put out a little stock, manufacture the demand that lifts the price, then borrow against that success. The result is roughly $25 billion in the bank or the capacity to pay down debt, most of it acquired from buying Twitter. As a piece of financial engineering combining the IPO with structured financing, it was creative. What it says about whether anyone should actually own the stock, or about the underlying quality of SpaceX, is a separate question each investor has to answer alone.
Will Anthropic Copy the Playbook?
Anthropic is the closest comparison for scale, so the natural question is whether it looks at that low-float, demand-manufacturing approach and adopts it. Predicting Anthropic is hard. The company does things differently and sees the world in its own way, so it is an open question whether it reaches for these tools at all.
The tools clearly work for raising money and lifting the price. The long-term cost is a different matter. Take any stock trading at 100 times revenues. If it comes back down to earth, there is going to be a lot of pain. A mispriced stock does not serve a company's long-term interest. My bet is that Anthropic is weighing exactly this, and that genuine price discovery may be something it actually wants. Of course it wants a share price that rises and rewards insiders and investors. A short-term overpriced stock helps no one.
Anthropic could also turn out to be a far larger deal than SpaceX. The company may be on pace toward $100 billion in revenues by the end of this year. These are projections, and the real numbers are unknown. But if it is doing $100 billion a year and priced at 10 times revenues at year-end, that is a trillion-dollar IPO on its own. It almost certainly will not be priced at 100 times revenues the way SpaceX was, and it is growing faster than SpaceX. There are reports that it is profitable, which SpaceX is not.
The size of the market for satellite phones you have to use outdoors and for rocket ships can be estimated. The AI market invites far more guessing. Right now Claude, Mythos, and Fable stand as the leading models across AI, the arena where tens of billions and hundreds of billions of dollars are being spent. These are the leaders, growing the fastest, possibly profitable, and potentially staring at $100 billion in revenues. That makes it a very big deal regardless of how much stock actually gets floated. And the tools remain on the table to release a very small amount of stock purely to push the price up.
Backers, Agents, and an Unknowable Market
Anthropic already carries strategic backing from two mega-cap names, Amazon and Alphabet. That does not change the investment story as much as it might seem. SpaceX also had backing from companies like Google, along with many of the world's biggest venture firms and private investors. Anthropic has the biggest backers in the world right now, and it is going through spectacular growth, but the backing itself is not the differentiator.
The pace of change in AI is relentless, shifting week to week and even day to day. Anthropic's best product ever was illegal on Friday and legal today. The company is heading toward an IPO while its latest large language model, built with billions of dollars of investment, does incredible things, based on the brief window it was legally available to those who could use it. Having emerged as a genuine leader over just the last couple of years, right down the street in San Francisco, the company sits at the center of that churn.
The most interesting structural feature is who the customer is. This business is driven not merely by human users but by AI agents, robots that are only starting to appear. Nobody knows how big that market is. That uncertainty sounds like the pitch of an oil and gas promoter asking who knows how much oil sits under his foot, so buy these shares. This is not a recommendation to invest in Anthropic or anything else. But a business where robots build robots that then use these tools, across an unseen number of use cases and an unknown count of users, is a genuinely fascinating thing to watch. It will be a compelling IPO to study, and the S1 document from a company this interesting is worth waiting for.
