• SpaceX/xAI combination complicates equity story
• Starlink remains compelling space infrastructure play

Its finally here, the most important listing of modern times, the largest listing ever attempted at a huge USD 1.75tn valuation, at a multiple of 92x sales, which will force investors to question everything they once thought they knew about IPO mathematics.

But SpaceX’s S-1 filing has prompted more numeric head scratching than many thought it would, especially around the decision to bolt xAI onto SpaceX, a move that threatens the viability of founder Elon Musk’s galactic dreams.

SpaceX’s S-1 reveals a highly cash generative space infrastructure business on one side and a cash consuming black hole on the other.

Starlink, the gold-plated segment of SpaceX, produced USD 11.4bn of revenue in 2025 and USD 4.4bn in operating income; it grew subscribers to 10.3m across more than 160 countries.

One investor looking at the IPO called the connectivity business a “cash machine”, praising both its moat and scale.

He also was effusive in praising Musk’s space infrastructure business and deals the billionaire is doing providing compute infrastructure for other artificial intelligence companies, such as the recent partnership between SpaceX and Anthropic, which sees the latter paying USD 15bn a year to use SpaceX data centers.

“If you look at the partnership with Anthropic and then expand that to orbital data centers, combined with Starlink and its amazing free cash flow, you have a possible monopolistic infrastructure play that nobody else can do,” the investor noted.

However, the infrastructure that supports Starlink needs serious investment.

The satellite internet provider is at, or very close to, capacity – meaning it needs more satellites in the air to continue to grow.

Slower cash generation in 1Q26 points to this capacity constraint, particularly as the business pushes into emerging markets at a lower price point, causing its average revenue per user to drop.

If SpaceX was using all its capex firepower to fix its Starlink capacity issues, that would be one thing – but it isn’t. The S-1 has revealed that most cash coming into the business is being used to feed the beast of xAI.

xAI black hole

In the first quarter, xAI lost USD 3.00 and spent USD 9.40 in capex for every USD 1.00 of revenue.

As a whole SpaceX lost USD 4.28bn in 1Q but made USD 4.69bn in revenue.

In the first quarter of 2026, SpaceX spent USD 10.1bn in capex, with USD 7.7bn going to xAI, USD 1.3bn to Starlink and just USD 1bn to space – despite the clear financial imperative to deliver more satellite payload for Starlink to monetize.

“The financials in the S-1 were a little disappointing,” noted a second investor looking at the deal. “The losses last year were massive and are accelerating this year. xAI in particular is a drag on the business. The AI spend is enormous.”

SpaceX’s 1Q26 capex haul is also almost half of the USD 20.7bn of capex across the whole of 2025, a rapidly accelerating number. That year, USD 12.7bn was spent on xAI – 61% of the total. That percentage increased to 76% last quarter.

While xAI will now be supported by an extra USD 15bn from Anthropic annually, at the current rate of spending that will last less than two quarters. Also, extrapolating across all of 2026, the cash spent on xAI will grow from USD 12.7bn to USD 30.8bn, a 142.5% increase.

And to repeat: this is all at the same time the space business needs investment to continue to roll out its Starship program, vital for the launch of next generation satellites which will allow Starlink to ramp up capacity, plus Musk’s dream of establishing orbital data centers.

All of this points to an extraordinary need for continuous capital injections, and the USD 75bn SpaceX is going to be raising at IPO, assuming an entirely primary raise, could be burnt through very quickly indeed.

At this rate, the SpaceX IPO might be the first raise in a perennial cycle of equity raising and cash burn, making its astronomic multiple of 92x sales perhaps even harder to justify on a purely mathematical basis.

We have had loss-making stock market debutants before, but few IPO candidates attempt to list at such an enormous gap between capex spend and losses as SpaceX.

The most recent example could be electric vehicle player Rivian, which listed in 2021 with a huge cash burn rate.

While the two companies are far from comparable operationally – and it’s certainly true that Starlink’s quality, the profile of SpaceX and Musk, and the moat around that business place it in another category – the cash flow parallels may conjure financial specters to haunt the minds of more conservative investors.

Source: Dealogic

“You can see from the numbers that xAI carries a massive capex load, which isn’t unusual for large language models,” said a banker who has worked on some of the company’s private fundraising rounds. “If you look at OpenAI’s capex spend, or Anthropic’s, these companies are burning billions, hundreds of billions, to grow.

“What’s striking is that the majority of SpaceX’s capex is now coming out of xAI, not out of the space business, which is arguably far more capital-intensive and dilutes overall valuation.”

The investor looking at the deal noted that SpaceX was essentially two businesses in his mind, the hugely attractive space infrastructure and connectivity business via Starlink, and the AI business that consumes capital at an alarming rate.

He noted that for now he could look past the concerns on the latter because of the qualities of the former; however, xAI would soon need to prove that it could grow revenue fast to avoid imposing too much cost on the rest of the business.

Math vs Musk

As reported last week, investors are fighting to get involved in SpaceX, fearful of missing out on an allocation in likely the largest and most high-profile IPO of all time.

All the market participants speaking to ECM Pulse found it hard to wrap their heads around SpaceX’s financials revealed in the S-1, especially given the 92x sales target is already impossible to justify on a fundamental basis.

The only way anyone seems to be able to make sense of SpaceX and support an investment of this magnitude is to view the IPO through the lens of Elon-omics rather than traditional deal metrics.

“You know there are going to be real investors that want to do real math, right? And the hard part is with an Elon company, you can’t necessarily do that,” said a banker on the IPO.

The second investor noted that heavy retail books in the IPO, with hordes of individual investors also likely to buy shares in the aftermarket, are unlikely to care about fundamentals. They will be buying the Musk hype.

To its credit, SpaceX is not shy of asking investors for their faith in the growth story, rather than dangling distribution baubles.

SpaceX is explicit in its IPO filing that it will not be returning cash to shareholders in the form of dividends due to the capital-intensive needs of the business.

The first investor noted that in his opinion SpaceX was unlikely to achieve free cash flow for some time, if ever, so investors were only buying the business based on the equity story.

An xAI-less SpaceX would have revenue of around USD 15.5bn, extrapolating 1Q revenue across the whole of 2026; at a USD 1tn valuation (the value of SpaceX at the time of the xAi merger), that would deliver a multiple of around 65x sales.

That might be justifiable for a possible monopolistic infrastructure provider powering most of the world’s AI economy. If SpaceX can then hit its most ambitious goals, the vertical economics of including xAI in the business may begin to look like genius.

Much of this may be determined on whether SpaceX can begin to initially deploy orbital data centers in 2028, as outlined in the S-1.

While this supports the narrative, it is high risk.

Orbital data centers are reliant on continued investment in the Starship program, that program hitting its own ambitious technical goals, and regulatory approvals.

This is to say nothing of the technical feasibility of such untested technology once in orbit, or whether the economics prove to be preferable to earth-based data centers.

But with the founder’s 85% voting control, investors in the SpaceX IPO are passengers on Elon Musk’s galactic ambitions.

They must hope he’s not flying too close to the sun.