SpaceX after the IPO: from 225 to 142 dollars. What really happened

On 9 June, three days before the listing, we wrote something simple: when whoever produces a forecast stands to gain if that forecast comes true, you need to look at the incentives before the numbers.

We had avoided predicting SpaceX’s collapse. We had chosen to ask a question instead. Why were almost all the institutions describing the IPO as a once-in-a-lifetime opportunity also involved in the funds, the placement or the fees generated by that deal?

Then SpaceX went public.

The shares were priced at $135, opened at $150 and closed the first day at $160.95. Four days later they reached $225.64. On 3 August they fell as low as $104.83. At the close on 1 September they were worth $142.23: just above the offer price and almost 37% below the peak.

In less than three months, the market added almost a trillion dollars to SpaceX’s value and then took it away again.

It would be easy to call it a failure. It would also be wrong. The numbers tell a more interesting story: the market first bought SpaceX as a single promise, then began to value the companies it contains separately.

SpaceX is three companies under one ticker

The first quarterly report since the listing exposed the structure that the IPO narrative had held together with a single word.

Today SpaceX comprises a cash-generating satellite connectivity business, a space infrastructure that requires continuous investment and an artificial intelligence segment capable of growing at extraordinary speed while absorbing capital just as fast.

In the second quarter of 2026 total revenue reached $7.814 billion, almost double the previous year. The net loss narrowed to $541 million. At first glance, the results beat expectations. The problem arises when you lift the bonnet.

Starlink is the part that actually pays

The Connectivity segment, built around Starlink, generated $4.291 billion in quarterly revenue and $1.656 billion in operating profit. Starlink reached 12 million active lines, double the figure for June 2025. It is the most legible part of the group: recurring subscriptions, real demand, margins already visible.

There is, however, a less photogenic figure. Average monthly revenue per subscriber has fallen from $85 to $66. International growth broadens the base, while cheaper plans lower average monetisation. Starlink is growing strongly, then. Yet each new customer is, on average, worth less than the last.

This does not destroy the model. It explains why twelve million subscribers and a dominant network can sustain SpaceX without being able, on their own, to justify a market capitalisation of close to $1.9 trillion.

More rockets fly than customers pay for

The space segment posted $962 million in revenue and an operating loss of $542 million in the quarter. SpaceX completed 37 Falcon launches, but only 10 primarily carried a customer’s payload. The other 27 mainly served the internal system. Even the period’s only Starship launch was classified as internal.

It is an important distinction. The number of launches measures industrial capacity. Launches paid for by external customers measure the market. SpaceX can break operational records while spending more than it earns from space transport, because much of that capacity is used to build the group’s other businesses.

Starship remains the element capable of changing every forecast. Until commercial frequency becomes repeatable, however, that capability still belongs to the future. And on the stock market, the future always has a present price.

xAI has turned the promise into Capex

The part that has truly changed SpaceX’s financial profile is xAI, acquired in February. In the second quarter the AI segment generated $2.561 billion in revenue, more than triple the previous year, and an operating loss of $1.257 billion. The growth is real. Its cost is even greater.

Of the $18.369 billion in total Capex for the quarter, $15.828 billion was absorbed by AI. A year earlier the entire group had spent $2.825 billion. In twelve months, quarterly capital expenditure has risen more than sixfold.

This is the figure that unsettled the market after the results. Data centres, GPUs, energy and compute infrastructure require money today to generate revenue that will have to prove it can last tomorrow. The contracts announced may fill capacity in the short term. The open question concerns margins, continuity and the time needed to recoup the investment.

We had already compared AI to the American railways of the nineteenth century: a technology can transform the world and destroy capital along the way. SpaceX concentrates this contradiction in a single set of accounts. Starlink funds the present. Starship builds the scale. xAI consumes the capital needed to reach an even greater promise.

The stock market has gravity too

The fall from $225 to $142 cannot be explained by the income statement alone. In the first few days, only a limited portion of the shares was in circulation. Scarcity, combined with the enormous attention on the IPO, pushed the price far faster than the fundamentals could change.

In August, the first restrictions preventing early investors from selling expired. More than 900 million shares became tradable on 6 August, more than were placed in the IPO. Further tranches followed. Being eligible for sale is not the same as being sold. Even so, the price began to contend with a larger supply and with shareholders who had come in years earlier at far lower valuations.

That is why $142 is not a final verdict on SpaceX’s value. It is the provisional point where a company with real results, gigantic expectations, new public data and a growing number of available shares meet.

The barber has merely changed his coat

Here we return to the thesis of the first article.

The banks involved in the IPO collected around half a billion dollars in fees. Shortly afterwards, many of the same institutions published bullish ratings and price targets ranging from $200 to $800. Citi set $200 in the near term and a trajectory capable of exceeding $900 in its most favourable scenario.

It could happen. SpaceX owns assets that no other company controls in the same combination. Yet a range that runs from bearish scenarios below $100 to optimistic cases above $900 bears little resemblance to a valuation. It resembles a statement of how much we still do not know.

The barber, meanwhile, has not disappeared. He has merely changed his coat: from the offering prospectus he has moved on to analyst reports. And the structural dependence on a founder remains embedded in every forecast. Musk retains voting control, decides how capital is allocated across three very different businesses and remains, at one and the same time, the main multiplier of confidence and its most concentrated risk.

What remains of the first analysis

In June, our point was about consensus. Today, the data allow us to add something.

SpaceX has an extraordinary, profitable connectivity business built around Starlink. It has a space infrastructure without equal, still costly to turn into external revenue. Finally, it has an AI business that is growing fast and burning amounts of capital hard to imagine even for one of the most ambitious companies in the world.

All three things can be true. A fourth can be true as well: the price paid to own them may build in many years of flawless execution before that execution actually happens.

The stock may return to $200, reach $300 or approach the $900 indicated by the most optimistic scenarios. It may also fall back below the IPO price. Guessing the next number matters less than understanding which assumptions are needed to reach it, who wrote them and who cashes in when you decide to believe them.

Hair grows back.
Capital, more often than not, does not.

---

N.B. The following article is for information purposes only. The information contained herein does not in any way constitute financial advice. Before investing, we encourage you to do your own research (DYOR) or to consult a certified financial adviser.