The World’s First Trillionaire
I’m sure you’ve seen the news. Now that his company SpaceX has gone public, in the world’s most valued and impactful IPO, the sale has made its owner, Elon Musk, the world’s first trillionaire.
By inflation-adjusted purchasing power, Musk is now probably among the top richest individuals ever recorded, maybe number one. Even by share of the global economy, he’s near the top, though probably still below historic rulers like Augustus. Still, he’s on the short of list of the most powerful individuals in human history.
I have some thoughts on the matter, mostly about the company itself. Many of you will disagree. Some of you might be angered by my take. I’m sure I’ll hear about it in the comments.
Now, I attempted to buy a single share of this IPO, just so I could get my hands on the prospectus. I did not receive any shares. I like to think it’s because “they” Googled me and saw how much I dislike most of what Musk does and stands for.
But I did enjoy reading the prospectus, which I will talk about below.
First, I need to say something. While I do not like Musk, I am deeply impressed by the overall innovation and trajectory of SpaceX. Musk’s initial vision was to reduce the price of moving mass into space, for the sole purpose of creating a realistic highway to Mars. And he was wildly successful. Compared with the Space Shuttle, SpaceX’s Falcon 9 rocket reduced the cost of putting mass into orbit by about 94-95%. The famous heavy lift vehicle from the 1960s, the mighty Saturn V rocket that put men on the Moon, cost about $40,000 (in today’s money) to put a single kilogram in low Earth orbit. The Falcon heavy does it for about $2000/kg.
And the new SpaceX workhorse, the Starship and its variants, might reduce that cost to $100/kg or even lower. This is due to a streamlined manufacture, launch, retrieval, and re-use system. It’s quite brilliant. Starship, by the way, is the biggest flying machine ever built by human hands. When the bugs are finally worked out, it is slated to be launched hundreds of times per month… perhaps more. That is a staggering amount of heavy lift capacity. A staggering amount.
This low cost opens the door to a heretofore unseen launch economy, servicing commercial, government, and military clients. This is an interesting distinction from how that other billionaire spaceman Jeff Bezos structured his rocket company, Blue Origin. The latter was initially presented as a space tourism company, with distant plans for more adventurous offerings, like lunar exploration. But Bezos funded most of it initially from his own personal wealth, essentially lining up a highway of a $1 billion of cashed Amazon stock every year to fuel the endeavour.
Musk, on the other hand, does not seem to have ever put his own money at risk. SpaceX was always funded by commercial and government contracts, with seemingly most (if not all) revenue re-channeled into product development. Some people dismiss SpaceX due to its reliance on government contracts, which is a criticism that I find odd. Government contracts are not grants or subsidies.
The US government is increasingly dependent on commercial launch capacity, as NASA’s own heavy launch vehicle, the SLS, is a financial boondoggle whose launch cadence cannot match the current or planned SpaceX cadence. SpaceX is the only operational American system capable of routinely launching NASA astronauts into orbit from U.S. soil. Meanwhile, all signs point to an accelerating desire for the US military to lean on proven commercial launch partners. In other words, I think it’s a reliable revenue stream.
Now, I know that many reading this are bristling at the very existence of a commercial space entity. Superficially, I see your point. But practically, this is an unavoidable development; we can debate this point at a later time. I’ve written in the past about the civilian scientific and economic spin-offs from the development of space, so I won’t go into that here, either. There are valid concerns, however, about the politicization and militarization of space, as well as fear of Kessler Syndrome, which becomes more and more likely as launch cadence increases. So, I acknowledge these criticisms, though they are not the point of this article.
Okay, so let’s talk about the prospectus. The thing about the SpaceX IPO is that investors aren’t getting one company, they’re getting at least two. In essence, they’re getting the rocket company and Musk’s artificial intelligence company, xAI, which is home to Grok, known for calling itself “MechaHitler.”
Yeah, this will end well.
Within the xAI assets, though, are two enormous data centres: Colossus 1 and Colossus 2. Colossus 1 is arguably the largest AI supercomputer currently operating. Colossus 2 is poised to become the largest AI data center ever built.
Colossus 1 currently consists of roughly 200,000 interconnected Nvidia GPUs, making it the largest coherent AI training cluster publicly disclosed. Colossus 2 is even larger. Current estimates suggest more than 500,000 GPU equivalents, with planned expansion to roughly 1 million GPUs. Its eventual power demand of 1–2 gigawatts. (That’s like the residential power demands of the city of Montreal.) If completed as envisioned, Colossus 2 would dwarf anything publicly announced by any of the competitors.
Currently, both Google and Anthropic rent computing power from Colossus to the tune of billions of dollars per month. This means that the financial heart of the SpaceX package is not the rocket company, but the data centres. Dedicated AI companies might not be profitable yet, and maybe they never will be. But AI infrastructure investment is already driving the lion’s share of US economic growth, and is clearly profitable.
But wait. There’s more. As I think everyone is aware, no one wants data centres in their community. They are water and energy hogs, and some people even fear they have deleterious health effects. (Though there is no convincing evidence of that yet.) That has led to loud voices in the field to suggesting that space-based data centres might be the future.
However, there is also a lot of chatter online about how such a thing is not feasible, mostly because in space cooling is done entirely by radiation, whereas on Earth it’s done by convection. And radiation is far less efficient.
But I do not think this is the non-starter that many argue. First, the obvious advantages are that, in space, solar energy is abundant and continuous; there are no clouds; there is no atmosphere absorbing sunlight; and waste heat can be radiated directly to space without the need for resources like water. What you need are very large radiator arrays. In space, that is not impossible to build. The physics is workable, the engineering is feasible, what is in doubt is the current economics of the plan.
The economics become favourable only if:
- Launch costs fall dramatically.
- Hardware becomes extremely lightweight.
- AI compute becomes valuable enough to justify orbital deployment.
It seems that SpaceX is betting on those eventualities, with the heavy lift capability of Starship at the heart of the wager. Indeed, SpaceX recently unveiled plans for an 11-million-square-foot “Gigasat” factory in Bastrop, Texas, intended to manufacture AI-equipped satellites that function as orbital data centres. Musk’s stated goal is to eventually produce gigawatts of space-based compute annually. So there you have it.
The question is, can they build centres in space that are cheaper than a terrestrial facility powered by a nuclear reactor? Today, the answer is almost certainly no. By the 2030s, if Starship achieves airline-like launch economics and AI demand continues its current trajectory, the answer becomes less obvious.
The other variable in play is the lunar colonization effort. With NASA’s Artemis program in full swing (sort of), China well on schedule to start their own Moon base, SpaceX is looking like it might prove to be the single indispensable lynchpin to the American lunar effort. It’s not surprising that Musk has his sights set on a sustained lunar presence, independent of NASA’s Moonbase plans.
If lunar materials can be used to locally generate orbital data centres, then mass driver technology could be employed for inexpensive launches, down to pennies per kilogram. Suddenly, constellations of space-based data centres start looking very economically feasible.
This is, of course, highly speculative. But if it pans out this way, SpaceX would be worth tens of trillions of dollars in 20-30 years.
I admire SpaceX technology and have waited all my life to see sustained human presence on the Moon and beyond. That aspect of it all is very exciting to me. But, let me be clear…
Elon Musk owns Xitter (X, formerly known as Twitter), which is a hotbed of hate and disinformation. He created Grok, trained on the venom of Xitter, which fancies itself “MechaHitler”. He created DOGE, which is likely killing untold numbers of people globally. His politics are divisive and cruel.
Furthermore, the prospects of AI superintelligence, housed in orbital brains outside the reach of governments and citizens, introduces a new dystopian layer to the Apocalyptic concerns for the headlong rush into AI development voiced by so many experts.
So, these are my mostly jumbled thoughts about the SpaceX IPO. It’s by all accounts an historic and important moment. Is the value of the company inflated and oversold? Most definitely. But there’s no denying that there is the potential for something truly transformative here. My worry, though, is that that might not be a good thing.
