SpaceX goes public on Friday. The rocket is the easy part of the pitch.

The most valuable company ever to go public did not get there on rockets. It got there on a flat antenna, about the size of a pizza box, bolted to roofs and RVs and ship railings in places the cable never reached. That is the thing to keep in front of you this week, while the renderings of Mars scroll past. SpaceX prices its IPO on Thursday and starts trading on the Nasdaq on Friday under the ticker SPCX, and at the number it is chasing — roughly $1.75 trillion — you are not paying for the launch business. You are paying for the dish.

The headline figures are the kind that get a deal called historic. SpaceX is targeting $135 a share across about 556.6 million shares, a raise near $75 billion. That is the largest initial public offering anyone has ever run, by a wide margin. The valuation would make SpaceX worth more than any listed aerospace or defense company, and put it in the same conversation as the biggest names in technology. Those are real numbers and they will dominate the coverage. They are also, for the purpose of deciding whether the thing makes sense, the least interesting numbers in the filing.

Follow the revenue, not the flame

Here is the line that actually explains the company. In 2025, SpaceX reported about $18.67 billion in revenue and a net loss of roughly $4.9 billion. Read those two figures together and the romance drains out fast: this is a company spending heavily enough to lose nearly five billion dollars in a year, and the market is being asked to value it at more than ninety times that year's sales.

So where does the value live? In one segment. Starlink — the satellite-internet business, the dish on the roof — accounts for about 61 percent of SpaceX's sales and is, by the company's own telling, its only consistently profitable line and its fastest-growing one. The connectivity segment turned a profit of around $1.19 billion last quarter. Starlink crossed 10 million subscribers in February, after roughly doubling its base over the course of 2025, from about 4.5 million to 9 million. That is the engine. Everything else — the Falcon launches, the Dragon flights, the Starship program that gets all the cameras — is either the cost of building that engine or an option on a future the IPO is not, strictly, asking you to pay for yet.

The launches everyone watches are the cost line. The dish nobody photographs is the revenue line. The IPO is a bet that the second keeps outrunning the first.

This is the inversion I keep coming back to on this beat. The launch business is the part that inspires, and inspiration, in a prospectus, is usually code for unprofitable. The launch business does not close on its own math; it loses money. What makes it work is that cheap, reusable launch is what put thousands of Starlink satellites in orbit at a cost no competitor can match, and Starlink is what pays for the launches. The rocket is the means. The subscription is the business. Investors who buy SPCX because they love Starship are, whether they know it or not, buying a broadband company with an extraordinary distribution advantage.

The retail door is the real novelty

What makes this offering genuinely unusual is not the size. It is who is being let in. Elon Musk has pushed to allocate as much as 30 percent of the IPO to retail investors — ordinary people buying through Robinhood, Fidelity, and Charles Schwab. In a normal deal of this scale, retail gets maybe 5 to 10 percent; the rest is carved up among institutions before anyone at a kitchen table gets a look. A 30 percent direct-to-retail allocation on the largest IPO in history is, as far as the record shows, without precedent.

Read that as a financing decision and it is shrewd. A large, enthusiastic retail base is sticky, supportive, and less likely to dump the stock at the first bad quarter than a hedge fund rebalancing its book. It also turns millions of customers and fans into shareholders, which is its own kind of marketing. But read it as a risk transfer and it is worth pausing on. The people most likely to buy SPCX on sentiment — because they admire the rockets, because they want a piece of Mars — are precisely the people least equipped to price a company that loses five billion dollars a year and trades at ninety times sales. The carve-out that looks like democratization is also the mechanism by which the most speculative part of the story gets sold to the least protected buyers.

There is a structural catch underneath, too. SpaceX is selling a slice; founder control and a small public float mean the shares that do trade can swing hard on thin volume, and index funds cannot simply scoop the stock up the way they would a normal large-cap until inclusion rules are met. A thin float plus a fervent retail base is a recipe for a stock that moves on mood. That cuts both ways on day one, and it is the part of the prospectus that the excitement is least likely to price.

What the proceeds are really for

Follow the use of proceeds and you find the next decade's spending, not last decade's profits. The money is earmarked to fund Starship development and a large compute build the filing refers to as Colossus 2 — the data-center side of Musk's empire, where SpaceX, Starlink, and his AI ambitions increasingly blur together. That blur is itself a thing to watch: the cleaner the Starlink cash machine, the more it is being asked to underwrite bets well outside satellite internet. Buyers should be clear which company they are funding — the profitable broadband operator, or the holding structure pouring that broadband's cash into rockets and GPUs.

None of this is a reason the deal fails. Starship, if it works at scale, lowers the cost per kilogram to orbit again, and a lower cost per kilogram is the only number that has ever actually changed access to space. Cheaper launch means more Starlink satellites, bigger ones, replaced more often, which means more capacity to sell. The flywheel is real, and it is the most impressive industrial story in the business. The question an IPO forces is not whether the flywheel is impressive. It is whether $1.75 trillion already assumes a decade of it spinning without a slip.

The booster that picked a bad week

And then there is the timing, which the company did not choose and cannot spin. Last week, on Starship's twelfth flight, the Super Heavy booster delivered the upper stage toward orbit and then, during the boostback burn that brings it home, appeared to lose an engine to an explosion. By the standards of a hard-engineering program, an anomaly on the way down after a successful delivery is not a catastrophe; this is how iterative development is supposed to look, and SpaceX has recovered from worse. By the standards of a roadshow, it is the worst possible reminder, arriving the worst possible week, that the part of the company the retail buyers love most is still genuinely experimental.

It is a useful reminder precisely because it separates the two companies inside SpaceX. The Starlink business did not notice the booster anomaly; the dishes kept working, the subscriptions kept billing. The Starship program is where the risk and the romance both live. Pricing the stock means deciding how much of Friday's number is the steady broadband operator that already works, and how much is the experimental rocket program that just scattered an engine over the Gulf. The honest answer is that most of the valuation is the former dressed in the imagery of the latter.

The Starlink business did not notice the booster blow an engine. The dishes kept billing. That is the company you are actually buying.

Does the math close?

So does it close? On the part that pays the bills, the case is stronger than the skeptics allow. A broadband business growing toward and past 10 million subscribers, profitable at the segment level, with a cost structure no rival can touch because it owns the cheapest ride to orbit — that is a real, durable, valuable company. If you were buying Starlink alone at a sensible multiple, the conversation would be easy.

You are not buying Starlink alone, and you are not buying it at a sensible multiple. You are buying the whole structure — the profitable dish, the money-losing launch business, the experimental rocket, and an expanding compute bet — at a price that already credits the flywheel for spinning cleanly through the 2030s. The figure assumes Starlink keeps compounding, Starship works at scale and on something like schedule, and the launch monopoly holds long enough for the economics to mature. Each of those is plausible. All three, priced to perfection, on a thin float, sold heavily to buyers shopping on sentiment, is a lot to ask of a single Friday.

The pizza-box antenna built the most valuable IPO in history. Whether it can carry a $1.75 trillion valuation is a different question, and it is the only one worth asking this week. The rockets are the easy part. They always were. The hard part, as ever, is the invoice.