The IPO Is the Exit

The Setup

On June 12, 2026, SpaceX (SPCX) began trading at $135 per share, the culmination of the most hyped offering in a generation and a prospectus that marketed a $1.75 trillion target valuation at roughly 94 times revenue. Five weeks later, on July 15, the stock closed at $134.85, its first finish below the offer price. It has traded below that line ever since, at one point falling six consecutive sessions, with roughly a trillion dollars of market value gone from its peak. Short sellers have collected an estimated $8.7 billion since the debut.

The reflex is to treat this as a SpaceX story: the valuation, the governance, the Starship schedule. We covered all of that before a single share traded in The S-1: Who Is This IPO Actually For? This article argues something broader. SpaceX is not an outlier that went wrong. SpaceX is the modern IPO working exactly as designed.

The thesis, stated plainly: the initial public offering no longer marks the beginning of a company's value creation. It marks the end of it, the point where a decade or two of privately captured growth gets converted to cash at the best available price. Value flows from the buyers of newly public shares to the holders of formerly private ones. Three mechanisms do the work: duration, allocation, and aftermath. Each one is measurable, and this article measures them.

What this article is not: it is not a claim that every IPO loses money (CoreWeave exists, and we will get to it), and it is not an allegation that anyone broke a rule. Everything described here is legal, disclosed, and entirely rational for the sellers. That is precisely the point. A system does not need a villain to produce a transfer. It only needs incentives, and the incentives all point one direction.

When the IPO Was the Starting Line

In May 1997, Amazon (AMZN) went public at $18 per share, a valuation of roughly $438 million. The company was three years old. Nearly everything Amazon became, the marketplace, the logistics network, the cloud business that now powers half the internet, was built after the listing, in public view, with public shareholders along for every leg of the compounding. Anyone with a brokerage account could participate. Millions did.

That was the norm, not the exception. In 1980, the median company came to market with $16 million in revenue, about $64 million in inflation adjusted terms. In 1999, the median technology company went public at four years old. Google, now Alphabet (GOOGL), listed in 2004 at roughly $23 billion, a number considered enormous at the time, and still early enough in its story that the public captured the overwhelming majority of the value it went on to create. Between 1980 and 2000, more than 6,500 companies made that same trip to the exchange.

The bargain was simple and symmetrical. The company got growth capital it could not raise anywhere else. The public got the growth. Public markets were where American companies grew up, and public shareholders were paid, sometimes spectacularly, for showing up early. That symmetry is the thing that quietly disappeared.

The Two Decade Migration

Start with the pipe itself. From 2001 through 2022, fewer than 3,000 companies went public in the United States, less than half the pace of the prior two decades. The number of listed US companies has shrunk from roughly 7,500 in 1997 to under 4,000 today. Fewer companies come to market, and the ones that do arrive profoundly different.

Jay Ritter, the University of Florida economist known as Mr. IPO, has tracked the change for four decades. The median age of a company at its IPO was four years for technology issuers in 1999. By 2024 the median across all IPOs was 13.5 years. Median revenue at listing went from $16 million in 1980 to $218 million in 2024. Median market value at the IPO, adjusted for inflation, went from $105 million in 1980 to $1.33 billion in 2021. The company that rings the bell today is not a startup asking for fuel. It is a fully grown business arriving at the end of its growth curve.

The reason is capital. Private markets got deep enough to fund anything. Private equity funds managed $8.2 trillion in 2023, more than double the 2018 total. There are more than 1,200 private companies valued above a billion dollars. Tender offers and secondary markets now deliver liquidity to employees and early investors on a schedule, no listing required. OpenAI raised at a $500 billion valuation without filing so much as an S-1. SpaceX itself ran private tender offers for two decades and reached a $400 billion private valuation as recently as last fall. When a company can raise unlimited money and pay out its insiders while private, the traditional reasons to go public simply expire.

Here is what that means for returns, quantified by the private markets industry itself. Andreessen Horowitz, a firm whose entire business is private access, published the numbers: companies that went public between 2014 and 2019 generated more than 80% of their market capitalization after listing. The most recent cohort generated over half of its market capitalization while still private. Forge Global, a marketplace for pre IPO shares, found that for companies that reached unicorn status and later listed, the median appreciation between the unicorn round and the IPO ran 65.7% per year. That 65.7% a year is the return stream that used to belong to public shareholders. It has not vanished. It has been rerouted.

Uber (UBER) was the prototype. It went public in 2019 at $45 per share and roughly $75 billion, after a decade in which all of the hypergrowth happened privately. The stock broke its offer price on day one and needed more than a year to sustainably reclaim it. By the time the doors opened to the public, the elevator had already made the trip.

And the arithmetic only gets worse from here. A company that lists at $438 million can return a hundredfold inside public markets. A company that lists at a trillion cannot; there is not enough market on Earth. When issuers arrive fully grown, the power law outcomes that retail investors once compounded are structurally pre captured. That is not pessimism. It is multiplication.

The Mechanics of the Pop

Now zoom in on the listing day itself, because the transfer does not wait for the aftermarket. It happens at the opening bell, in plain sight.

An IPO is priced through bookbuilding. The underwriters set the offer price the night before trading, and the shares at that price are allocated overwhelmingly to institutional clients. The public's first available price is the opening print, which on a hot deal opens 20% to 50% higher, sometimes far more. Every headline you read about a stock "soaring in its debut" is measuring from a price the public never had access to. Ritter has a name for the aggregate gap between the offer price and the first day close: money left on the table. It is real wealth, and it flows from the issuer and its selling shareholders to the allocated funds positioned to flip.

The 2025 revival made the mechanics unusually visible. Across the roughly 100 US IPOs priced above $100 million last year, the average first day return was 24%, according to Dealogic. Strip out just two deals, Figma (FIG) and Circle (CRCL), and the average falls to 8%. The pop is not evenly distributed. It is a lottery, and the lottery tickets were handed out the night before.

Figma is the cleanest specimen. It priced at $33 per share, a $19.3 billion valuation, and closed its first day near $47 billion, up roughly 250%. Circle more than doubled on day one. Chime (CHYM) rose 59%. Then the fades began. The pops across the class faded almost universally, with the average name settling well below its first day close as institutions flipped their allocations into retail demand. By late December, Figma sat among the large 2025 listings down more than 73% from their highs.

Map the flows and the picture is uncomfortable. The allocated buyer paid $33 the night before and sold into the open. The public buyer paid roughly triple the offer at the bell and absorbed the decline. Nothing about that sequence was hidden, illegal, or even unusual. It is simply what the structure produces when demand is marketed to one group and access is granted to another.

The Aftermath

Suppose you skip the pop entirely and simply buy at the end of day one, the first honest price available to everyone. The data on what happens next has been consistent for as long as anyone has measured it.

Ritter's 1991 study in the Journal of Finance, the paper that named the anomaly, tracked 1,526 US IPOs from 1975 to 1984. Purchased at the first day's closing price and held for three years, the IPOs returned 34.5%. A matched sample of comparable companies returned 61.9%. The underpricing that creates the pop is a short run phenomenon. The long run belongs to underperformance.

Modern cross sections repeat the shape. A Nasdaq analysis of three year post IPO returns found nearly two thirds of IPOs trailing the market, with 64% behind by more than ten percentage points. Only about 29% outperform, and the winners are violently concentrated: the top decile earns an average market adjusted return above 300%, the ninth decile 75%, the eighth 25%, and it decays from there. The average IPO return is a lottery distribution, a few jackpots stapled to a long tail of losses. Composition makes it worse: in the frothiest recent vintages, roughly 80% of IPOs arrived at the exchange unprofitable, and Ritter's data shows that revenue scale at listing predicts long run performance far better than the size of the pop does.

The class of 2025 is the live sample. By the last week of December, 14 of the 20 largest IPOs of the year traded below their offer prices, and 11 of the 20 had fallen more than 40% from their intraday highs. Renaissance Capital calculated that the ten largest deals averaged a 13% return from the offer price, and only by excluding Venture Global (VG), the year's biggest early flop, did the group reach 23%. Remember what the offer price means: those modest gains accrued to the allocated buyers. Measured from the first day closes, the prices the public actually paid, the class was a sea of red, and it remained one into this spring.

Even the winner tells the story. CoreWeave (CRWV) priced at $40, below its range, in an undersubscribed deal that needed Nvidia (NVDA) to step in as an anchor buyer. It ran to roughly $187 by June 2025 and traded back near $89 by this spring. The single best name in the entire cohort round tripped more than half its peak. That is what winning looked like.

SpaceX in Real Time

Which brings us to the perfect specimen. SpaceX is 24 years old. It is the company that proved, more thoroughly than any other, that you never have to go public: two decades of private rounds, scheduled tender offers that cashed out employees and early holders on demand, and a $400 billion private valuation as recently as last fall, all without filing a single quarterly report. Then it absorbed xAI in an all stock deal in February, a transaction we traced in The Musk Shell Game, filed its S-1 in May, and marketed the largest offering ever attempted at a $1.75 trillion target, roughly 94 times revenue, against a claimed $28.5 trillion addressable market.

The tape since: priced at $135 on June 12. An all time high of $225.64. Then the slide. On July 15 the stock closed at $134.85, its first finish below the offer price, and it has stayed below since, at one point dropping six straight sessions while roughly a trillion dollars of market value evaporated from the peak. About 185 million shares are sold short, near 29% of the public float, roughly $25 billion in bets, and the shorts are up an estimated $8.7 billion in five weeks. There are buyers too: ARK bought the dip across four funds. The next Starship test flight is scheduled for July 23, and the first earnings report lands August 4. Those catalysts may move the stock violently in either direction. They will not change who has already been paid.

Because now sort every SpaceX shareholder in history into two buckets. Bucket one: everyone who bought privately, in any round across two decades, at valuations from the millions to $400 billion, plus everyone allocated shares at $135 the night before trading. Every member of that bucket is above water or already out, most by multiples. Bucket two: everyone who bought in the open market above $135. That bucket is the public. It is the only bucket losing money, and it took five weeks.

Notice what did not have to happen for this outcome. No scandal. No failed launch driving the decline. Not even a bad quarter, because the company has not yet reported one. The stock simply stopped finding new buyers at prices above what the sellers had already received. When a company arrives at the exchange fully priced for a $28.5 trillion future, the marginal public buyer is not purchasing growth. They are purchasing the seller's exit, at the seller's price. We asked in May who this IPO was actually for. The market has spent five weeks answering: it was for the people selling it.

What Could Go Wrong

This thesis has real counterarguments, and they deserve full strength.

The anomaly is contested. The long run underperformance result is sensitive to method and sample. A National Bureau of Economic Research study extending the record back to 1935 found that the underperformance largely disappears under alternative measurement approaches, including calendar time and factor model tests. Some of the measured gap may be an artifact of hot issue windows and benchmark choice rather than a permanent law of markets. Ritter's own work shows the damage concentrates in young, unprofitable companies listed during high volume years, which means it is partly avoidable by selection.

The winners are buyable in public. CoreWeave opened below its offer price. The single best return of the entire 2025 class was available to anyone at the opening bell, no allocation required. And the concentration data cuts both ways: because a top decile of IPOs returns 300% or more, a diversified basket can capture the jackpots without predicting them. An investor who owned the whole cohort, pops and flops alike, fared far better than one who chased only the hyped names.

Selection improves the odds. Profitable issuers with substantial revenue have historically performed dramatically better than money losers. Nobody is compelled to buy the average IPO, and the average is what the grim statistics describe.

The exclusion is eroding. Retail vehicles now reach private markets. One publicly traded crossover fund that held SpaceX before the listing reported a 27% second quarter return on exactly that strategy, and secondary platforms and interval funds keep widening access. If the objection is that the public cannot touch private growth, that is becoming less true each year. Worth noting, though, what is for sale in those venues: pre IPO shares offered to the public are, by definition, the shares earlier holders wish to sell. Adverse selection does not vanish because the venue changed.

Every trade has two willing sides. If IPO opening prices systematically overpay, the remedy is available to everyone: do not pay them. A structure can only transfer value from participants who volunteer. That rebuttal is fair, and it is also the quiet concession of this entire article. The defense of the modern IPO is not that it creates value for its buyers. It is that the buyers consented.

What the Data Suggests

None of what follows is a recommendation. It is what the record shows about how the game is scored.

First, measure from the open, not the offer. Reported IPO returns describe a price the public cannot buy. The relevant question for a public investor is never how the stock did from the offer price. It is how it did from the first print anyone could actually trade, and by that measure the modern record is bleak.

Second, time has been the public buyer's only reliable edge. Lockup expirations around six months, the first few earnings reports, and the general first year shakeout have repriced most of these names far below their debut euphoria. CoreWeave was available below its $40 offer at the open. Figma round tripped. SpaceX broke issue in five weeks. Across the data, waiting has historically cost little and saved a great deal.

Third, fundamentals at listing separate the cohorts. Revenue scale and profitability at the IPO have predicted long run survival far better than first day enthusiasm. The companies that reward public shareholders tend to be the ones that arrive with a business, not a story.

Fourth, the question that cuts through every roadshow: who is selling, and why now? A 14 year old company in a world with $8 trillion of private capital does not list because it needs your money. It lists because someone inside wants your price. The answer to that question is usually the entire analysis.

Finally, a note on how we handle this at the platform level. Wealth Engine Pro's systematic scores sit out fresh IPOs until filing history accumulates, because a company with one public quarter has no scoreable record. No numbers, no score. The same discipline applies to the decision an IPO asks you to make on day one.

The Bottom Line

The initial public offering once marked the start of public value creation. Today it marks the settlement of private value creation: a median 13.5 years of private compounding at 65.7% a year, a first day pop reserved for the allocated, and an aftermarket that has trailed the market for as long as academics have measured it. The structure did not break. It inverted. The public used to be the partner in the growth. Now the public is the liquidity event.

No villain is required for this reading, and none is alleged. When companies can grow to full size on private capital, the one remaining reason to list is liquidity for the people already inside, and every incentive in the pipeline, from the late round investors to the underwriters to the founders, points toward selling at the highest price the public will bear. SpaceX is not the cautionary exception to the modern IPO. It is the model executed at maximum scale: every private holder across two decades paid, every open market buyer above $135 underwater, five weeks in.

The narrative says an IPO is an invitation. The data says it is a receipt. At Wealth Engine Pro, we evaluate companies on what they are, not on what a roadshow promises, and the same rule applies to the offering itself. Before buying the ticker, it is worth asking what the numbers say about the transaction. Usually, they say it already happened, years ago, at prices you were never offered.