The real reason to raise an eyebrow at Musk’s wealth has nothing to do with him, and everything with what he is actually selling.
Earlier this week, we looked at some of the many arguments about Elon Musk’s status as the world’s first trillionaire – that nobody deserves to be that rich, that the assets of money-hoarding sociopaths must be taxed, that he exploited subsidies and tax breaks to make his fortune, that he is a dislikable character – and concluded that those arguments are either stupid, wrong, or quite right but beside the point.
In this, the second part of the story, we’ll consider the true reason why I believe Musk’s insane wealth is not merited.
In short, and glossing over some caveats we’ll get to in more detail later, Musk sells pie in the sky, most of which is not realistically achievable.
Common business advice is to under-promise and over-deliver. Musk does the exact opposite, and on a grand scale.
And while his unbounded ambition has produced some amazing breakthroughs, those alone are nowhere near sufficient to justify the market valuation of SpaceX, of which Musk owns 41%. (The same goes for his other major publicly listed company, Tesla, of which he owns 20%, but for the purposes of this article I’ll focus mostly on SpaceX.)
Spectacular achievements
Give SpaceX its due first, because its achievements are real, and spectacular.
It has racked up an impressive array of technological firsts. The company cracked reusable rockets when the entire aerospace establishment insisted it could not be done. Others are trying (and for the most part failing) to copy it.
SpaceX launches more than 80% of the world’s mass to orbit. Its Falcon-9 rocket now flies almost every other day, and boasts a 99.5% success rate. It has flown 652 successful missions since 2010, and landed its reusable boosters 598 times. The current record for reusing boosters is 35 times.
Starlink, its orbital communications network, is a genuine business with genuine revenue, 10.3 million subscribers, and an audited $11.4 billion to show for 2025.
None of that is hype. All of it is extraordinary. And all of it is particularly notable for being achieved by a private company, instead of a bloated and inefficient state bureaucracy.
But ultimately, the question we need to answer is what SpaceX is worth. And sadly, there is no better answer than “whatever people are prepared to pay for it”.
What is SpaceX worth?
SpaceX, as it is described in the prospectus for its initial public offering (IPO), consists of three core businesses: the space launch business, the AI business (including social media platform X), and the Starlink communications network.
Of these, only Starlink is profitable, but it is not profitable enough to counter accelerating losses in the space division, and the equally rapidly accumulating losses in the AI division.
Summary here, but the upshot is that in SpaceX, Musk was listing a business that lost over $1.9 billion on revenue of $4.7 billion just in the first quarter of 2026 alone. That reflects a mere 5% revenue growth on 2025’s $18.7 billion, and a very rapidly growing loss from $2.6 billion for the full year. Since its founding, it has lost $41 billion.
I don’t want to dwell too much on the numbers, because whichever way you slice them, there is simply no way that they shed any light on the company’s stock market valuation.
Its price-to-earnings ratio, which usually indicates how long it will take before a company’s profits justify the price of its shares, is negative. Analysts have used the absurd measure of price-to-revenue to get a handle on the company’s valuation, but that sat at a mind-boggling 94 at the listing price of $135 per share.
After peaking at $225 a share two trading days after the IPO, it fell back to a narrow range around $155, where it has been trading for a few days now. In fact, Musk is no longer a trillionaire at the time of writing.
SpaceX sold less than 5% of its total stock at its listing, raising $87.5 billion. The share price at the time of writing values the company at over $2 trillion, making it the 7th largest company in the world by market capitalisation. The companies ahead of it on the list report annual earnings between $65 billion (TSMC) and $195 billion (Amazon). By revenue, SpaceX doesn’t come anywhere near the top 1,000 largest companies.
As I said, the valuation doesn’t make any sense based on the numbers. But then, people weren’t buying numbers. They were buying vibes.
Musk’s promises
People who bought SpaceX shares bought Elon Musk’s vision. We should distinguish between things Musk merely said, and things that are included in the SpaceX prospectus.
Verbally, and on social media, Elon Musk’s ambitions are literally limitless.
At the IPO, Musk said that the goal of SpaceX is to “make Star Trek”.
“What SpaceX is all about, is to take the fiction out of science fiction and create an exciting, inspiring future for everyone,” he said. “We want to be able to take anyone who wants to go to the moon, anyone who wants to go to Mars, or anywhere in the solar system, and maybe beyond the solar system… at some point, we want to be able to take you there. Not just a few astronauts. I mean, you, literally you, whoever you are watching this. SpaceX wants to be able to take you to the moon, take you to Mars, and ultimately beyond. And I, I’m confident at this point that with the incredible team that we have here at SpaceX, that we will do that for you.”
Those are bold claims. Extremely bold claims. How do you value a “Star Trek” future, with all the literally impossible technologies that this entails?
Musk has form with bold claims. Just recently, he said: “In the future, a trillion times a trillion dollars will be spent on making antimatter to travel to other star systems. Things won’t be measure [sic] in dollars then, just mass & energy.”
Now we can create antimatter. Under ideal circumstances, the particle collider at CERN can, in theory, produce a grand total of 0.00000000000000003g of antihydrogen per year. Introduce it to hydrogen, and you get a very exciting explosion in which both particles are annihilated to produce pure energy.
It’s about 100 times more energetic than nuclear fusion, which has been 20 years away since the 1950s. Even so, 30 attograms isn’t going to get us very far.
An additional problem is that storing it is insanely difficult and energy-intensive, and involves hard-physics constraints that will be extraordinarily difficult to engineer around.
So there’s no way to value a claim like this, because it posits a hypothetical technology so far in the future that it is impossible to say when, or whether, it will be realised. You might as well be promising warp drive.
You can’t invest in this, because the return, if any, will be earned generations from now. Put differently, there are an almost limitless number of investment opportunities where capital can earn greater returns, sooner, and at less risk.
Also Elon: “…if current trends continue, I would say the economy 10xes in 10 years and have a base on the moon.”
We’ll get to the moon base idea later. In the last 10 years (well, from 2014 to 2024), the world economy grew by less than 40%. In the 10 years before that, it grew by 82%. The last time the economy increased ten-fold it took 44 years.
The highest annual growth rate recorded since 1960 was 6.6% in 1964. Even the post-pandemic rebound only reached 6.4%. The ten-year period with the highest average growth rate was 1964 to 1973, averaging 5.32% per year.
“If current trends continue,” to quote Elon, global GDP will grow at around 3% per year, as it has since 2022, resulting in total growth of 35% over 10 years.
Musk wants us to believe that growth will, over the next decade, average 25.9% per annum, which is what is required for a ten-fold increase (1,000%) in global output.
Well, we all know Musk is on drugs.
So let’s come back to Earth and read the SpaceX prospectus, because putting science fiction in a prospectus constitutes fraud.
SpaceX prospectus promises
You’d think that the prospectus would be toned down a little, since the Securities and Exchange Commission doesn’t look very kindly on selling hot air, but no.
SpaceX’s mission is, “To build the systems and technologies necessary to make life multiplanetary, to understand the true nature of the universe, and to extend the light of consciousness to the stars.”
Well, that’s nice. Except that none of those things is even remotely achievable in the foreseeable future with any technology we can plausibly develop over the next few decades.
So let’s skip past the aspirational guff that few companies would dare to put in an SEC filing, and the baker’s dozen of lovely rocket photos, and try find something more concrete to evaluate.
It commits to “build a base on the Moon and cities on other planets”.
One does not usually encounter the word “dinosaurs” in a registration statement. Here it appears as a risk thesis: we do not want humans to share their fate.
The promise carrying the actual valuation is narrower and more interesting: data centres in orbit.
The company expects to begin deploying “orbital AI compute satellites as early as 2028,” harnessing constant sunlight and the cold of space to run the AI workloads that are straining terrestrial power grids.
SpaceX believes this is “an incredibly difficult technical challenge that only we can solve at scale in the near term”.
Perhaps it is right. The rocket-and-internet business alone cannot justify the price, so a great deal rests on a technology nobody has ever operated, at a scale – up to a million satellites – nobody has ever attempted.
And that goal had better be possible, because without it, the company cannot generate the revenue required for its “future markets”.
These include a lunar economy of “cargo transport, manufacturing, and energy production,” a railgun on the Moon, defined in the glossary as a “lunar mass driver,” and manufacturing on Mars.
To its credit, the filing does not pretend these are around the corner. It concedes that they “do not exist today”. That may be the most honest sentence in the document.
Yet there are also implicit promises, contained in the incentive structure of Musk’s own compensation. It is tied to two of these stretch goals. One grant of 200 million shares vests only on the “establishment of a permanent human colony on Mars with at least one million inhabitants”.
A second, of 60 million shares, requires “the completion of non-Earth-based data centers capable of delivering 100 terawatts of compute per year”.
Each tranche also demands a soaring market capitalisation, rising to levels never seen before.
Neither objective has a deadline, and the body certifying that they have been achieved is controlled by Musk, through his super-voting Class B stock, which gives him more than 80% of the vote, and clauses that prohibit shareholders from firing him as CEO.
The prospectus makes big promises, but it likely survived regulatory scrutiny by being candid about the catch in the fine print.
The whole edifice “is highly dependent on the successful development and scaling of Starship,” and the grand initiatives “may not achieve commercial viability”.
That is a very short summary of 38 pages of risks that may scupper SpaceX’s ambitions. They include dependence on non-existent technology, accidents, regulation, legal issues, and the fact that the entire enterprise is highly dependent on one man who cannot be fired: Elon Musk.
He is not exactly the picture of health, and yet investors are betting entirely on his vision and their belief in his ability to execute at least a significant part of that vision.
The problem with data centres in space
It is easy to poke holes in very substantial aspects of the promises contained in the SpaceX prospectus.
Let’s start with the idea of data centres in space, because that is the core revenue generator that will support everything else.
There are many reasons why the idea of a million (or millions of) satellites handling compute in orbit is not feasible.
Data centres require power. They require cooling. They require a shielded environment. They require fast, high-bandwidth communication. They weigh very many tons.
All of this is problematic in space.
The cooling problem
The fundamental problem with thermal is that although space is nominally cold, it is also a vacuum, and a vacuum is an excellent insulator. That makes shedding heat harder, not easier.
On Earth, air and water carry heat away by convection and conduction; in orbit there is no medium, so a data centre can only lose heat by radiation, which is hundreds of times less effective.
At $150 billion, the International Space Station is the single most expensive thing ever built in human history. It is also the heaviest object ever flown to space, at about 450 tons.
The radiators of the ISS are about 42.5 square metres in size, and dump about 70kW of heat into space.
Now, the SpaceX prospectus talks of terawatts, but let’s rather imagine the requirements of a gigawatt data centre that would rival those of the hyperscalers of today. (Meta is building a 5GW data centre called Hyperion, due to be completed in 2027, at an estimated cost of $10 billion.)
Consider a one gigawatt AI data centre compute in space, then. Computers convert almost all of their energy into heat, so such a data centre needs to dissipate somewhere near 1GW of heat. That means you would need radiators at a scale of about 14,000 ISS systems, at, say, 30 tons each, to cool a single gigawatt data centre.
If we assume Starship is to carry this stuff into space, we’d need 2,800 flights, just to get the cooling radiators for a single gigawatt data centre to low Earth orbit.
This is so far beyond anything that has ever been done, it can be dismissed as impossible.
The power problem
But at least there’s unlimited power in space, right? Solar energy doesn’t have to pierce a thick atmosphere, so we should be golden, right?
Yeah, well, no. This notion ignores that solar panels are themselves heavy and require enormous area. The new iROSA panels on the ISS cover an area of 114 square metres, and produce a peak output of 120kW. They weigh about two tons, total. Due to orbital shadows and system losses, about a third to a half of that output is usable power.
Powering a gigawatt data centre in space would take 16,500 times more area, or about 33,000 tons, or 220 Starship flights to orbit. For a single data centre.
And we haven’t even got to the actual computer hardware.
Problems upon problems
Launch mass just kills the economics. A single gigawatt orbital centre will come in at between 250,000 and 500,000 tons – compute, radiators and solar combined. At today’s roughly $5,000 per kilogram, that means at least $1.25 trillion just to reach orbit, before counting the cost of hardware, batteries or in-orbit assembly.
That is two or three orders of magnitude more expensive than data centres on Earth.
Three further problems make the orbital case worse.
Bandwidth: AI clusters need fibre-class interconnects of terabits per second, but orbit-to-ground links rely on lasers that atmosphere, cloud and turbulence throttle far below fibre capacity. So now you have massive compute trapped behind a narrow pipe.
Radiation: high-energy particles flip bits and degrade silicon, and shielding adds the very mass that drives the price up.
Maintenance: you cannot swap a failed rack in orbit, so the only model is Starlink-style redundancy and constant relaunch, multiplying cost and adding orbital debris and atmospheric pollution.
The Moon, floated as the path to true scale, fails harder. Radiation is unshielded by an atmosphere or magnetosphere, which means you have to use far slower radiation-hardened chips at 10 to 100 times the cost.
The 14-day lunar night demands enormous batteries or reactors. Abrasive dust degrades everything. Landing costs run roughly 20 times orbital costs. The 2.6 second round-trip light delay rules out real-time inference entirely, leaving only batch processing or cold storage as a use case.
That Earthbound data centres face problems is indisputable. There are alternatives to space-based data centres that don’t create more problems than they solve. We could build data centres underwater, for example. We could build out nuclear energy to power them. Doing so would still be more expensive than today’s land-based data centres, but they would out-compete space-based data centres handily.
At the scale Musk envisions, the concept of space-based AI data centres is defeated first by engineering complexity, then by economics, and ultimately by first-principles physics.
I hope to be proven wrong, as journalists before me were wrong in declaring heavier-than-air flight or rocket propulsion in space impossible. But that is the gamble you’re taking in believing SpaceX’s multi-trillion dollar valuation.
Stretch goal: in-situ manufacturing
The stretch goals are even more implausible.
Musk speculates that much, or all, of future data centres (and eventually, habitable bases) can be manufactured in situ, on the moon (and, ultimately, Mars). This might be possible in principle, but not in the foreseeable future, on a scale that merits investing in SpaceX.
In-situ manufacturing off-world founders on the sheer depth of the industrial stack that any modern fabrication requires.
Making even a single GPU on Earth draws on a globe-spanning supply chain – ultrapure silicon, dozens of trace elements, photolithography machines that are themselves among the most complex objects humans have ever built, reliable gigawatt-scale power, clean water, and a workforce of specialists.
Each of those inputs sits atop its own pyramid of prior industry. You cannot simply transplant the tip of that pyramid without the base.
A lunar or Martian factory would have to either import nearly everything (which defeats the purpose, since the entire rationale is to escape the constraint of the rocket equation), or bootstrap mining, refining, chemical processing, energy generation, and precision tooling more or less from scratch, in an environment that is actively hostile to all of them: abrasive electrostatic dust that degrades seals and optics, vacuum or near-vacuum that complicates any process relying on convection or fluids, temperature swings of hundreds of degrees, unshielded radiation, and communication delays ranging from seconds to minutes that rule out real-time remote operation and forces near-total autonomy.
SpaceX’s own prospectus implicitly concedes the point by filing “manufacturing capabilities on the Moon and Mars” under technologies that “do not exist today”.
The plausible near-term reality is much more limited than manufacturing high-tech equipment. In-situ resources utilisation could plausibly be used to produce basics like oxygen, water and rocket fuel. Perhaps construction material could be created from regolith. But the off-world manufacture of complex finished goods is so far beyond the foreseeable future that it might as well be classed as impossible.
Stretch goal: lunar or Martian colonies
Once again, we butt up against the hard realities of physics, biology and economics.
A permanent colony on the Moon or Mars fails on all three fronts at once, and the failures compound.
Physically, neither destination tolerates human life. Mars has barely 1% of Earth’s atmospheric pressure and no magnetic field, the Moon none of either, so settlers would face unshielded cosmic rays and solar-storm radiation, temperature swings of hundreds of degrees, and – on the Moon – abrasive dust that ruins seals, lungs, and machinery.
Every breath of air, drop of water, and calorie of food must be manufactured or imported and then endlessly recycled in a sealed loop that has never been demonstrated at colony scale even on Earth.
Biologically, the deeper unknowns are unresolved: the long-term effects of very low gravity on the human body are untested, and probably injurious. We have no evidence that mammalian reproduction and development can succeed off-Earth at all, and a colony that cannot safely bear children is not a colony but an outpost.
Economically, the case is weakest of all. Settling a million people on Mars will not relieve overpopulation. It will not make anyone rich. As Musk himself admits, it has no business model.
The best historical analogy is Antarctica, which despite its far gentler climate hosts only government research stations and a trickle of tourists, not cities, or even small towns.
A self-sustaining city of a million people on Mars is not something that a company can promise, but a multi-generational aspiration resting on breakthroughs nobody has yet made, and which look mighty implausible from our perch at SpaceX’s IPO in 2026.
Musk has form
Musk has form selling vapourware.
Full self-driving cars have been a year away every year since 2015. A key reason they’re still not available is Musk’s own insistence that they can’t use Lidar, but must use only camera vision. Rivals are operating robotaxi fleets, but Tesla is not.
The Hyperloop never materialised, and the Boring Company, which was supposed to construct its infrastructure, now digs tunnels for good old-fashioned subway train systems.
Musk said in 2016 that Dragon capsules would land on Mars by 2018. No SpaceX vehicle has ever made it beyond low-Earth orbit, and Starship, on which he bet the entire business, has yet to achieve even that.
Teslas would be appreciating assets, he said, and you could run your own robotaxi network. Well, no.
How many people bought Tesla’s solar roof tiles? Tesla’s second-generation Roadster was unveiled in 2017. We’re still waiting for it to be produced.
Then there’s that godawful bakkie of his, which looks terrible, ages terribly, performs terribly, and is just not fit for purpose. Remember the bullet-proof glass demo? Yikes!
Musk’s pattern is to make aggressive-to-fantastical promises, on hopelessly optimistic timelines, on genuinely hard goals. Many collapse entirely. Some arrive years behind schedule and billions over budget. And then he uses the delivered-late successes to retroactively launder his companies’ reputations.
It’s fine to have stretch goals. It’s fine to throw money at many ideas, just to see what sticks. Failure, and failing fast, are perfectly decent strategies.
But none of that justifies SpaceX’s valuation (or that of Tesla, for that matter).
The crystal ball
Ray Dalio, billionaire investor and hedge fund manager, once said: “Those who live by the crystal ball are destined to eat broken glass.”
Elon Musk sells space exploration as a limitless frontier.
As long as people believe that, trillion-dollar valuations will hold up. As soon as the prophecy encounters reality – namely that there are very hard limits that nobody really considered, and Musk himself glossed over – the bubble will burst.
At that point, SpaceX will be re-valued at a more realistic function of feasible future earnings, and my guess is that a lot of unhappy investors will be left holding the bag.
That, ultimately, is why Musk isn’t worth his on-paper wealth. On what he’s actually selling, versus what he’s promising, the financial math just doesn’t work out.
The most important bits of what he’s selling will never happen. Much of what will happen doesn’t have a sustainable business case. And everything depends on the mythical will of a single individual, who contractually may never be replaced, and for whom there is no succession plan.
Those are risks that aren’t worth trillions, and neither is the man promoting them.
[Image: An artist’s impression of a future SpaceX moonbase. https://www.spacex.com/humanspaceflight/moon]
The views of the writer are not necessarily the views of the Daily Friend or the IRR.
If you like what you have just read, support the Daily Friend