Elon Musk "is racing," as the Wall Street Journal put it Wednesday, to take his closely held SpaceX company public — raking in billions in cold, hard cash to fund the founder's latest wild ambition: orbital AI data centers.
The "why" is simple. The "what's next" is where things get interesting. And the "how" might reveal whether Musk still intends to build a self-sustaining human colony on Mars.
The "why" is that SpaceX doesn't generate enough cash to tackle Musk's next project: orbital artificial intelligence data centers. As I'm sure you know, the plain-old Earthbound data centers already planned for construction will likely require more power than our grid can provide, prompting novel solutions like small, modular nuclear reactors that might not ever materialize.
But the sun always shines in orbit, and if anybody knows how to network a large constellation of solar-powered satellites together to provide a single service, it's the company that created Starlink.
Full Disclosure: I'm a happy Starlink customer, and I became one because it was the one action I could take to support SpaceX's Mars ambitions. Cash generated from Starlink subscriptions like mine goes into Starship development, and Starship opens up the inner Solar System to humanity.
Heat's a problem, though. All that electricity generates a lot of it — and it isn't always easy to dissipate right here on Earth. You should hear the fan run on my Mac mini when I'm ripping 4K video.
Space is cold, yes, but it's also a vacuum. Anybody who understands how a Thermos works also knows why it's so difficult to cool off hot-running equipment in space. But orbital heat dissipation is merely a technical issue, and Musk seems like a pretty bright fellow with some decent engineers on his team.
We have to divide "what's next?" into two parts.
The first part of the "what's next" is, "What happens if the AI bubble pops before SpaceX builds out its orbital data centers?"
That's an excellent question, and it makes me so glad I asked it. What I don't have is an answer, or certainly not a satisfactory one.
There's some small chance that there's no bubble in AI, because there's no limit to the human desire to have machines automate whatever we can get them to automate. And putting those machines in orbit, as Musk well knows, eliminates the power-production crunch that could pop the bubble.
Also, the AI bubble is unusual compared to the last two bubbles we've lived through. The dot-com bubble built out our internet infrastructure, enabling everything from Indian service centers for American companies to streaming video. A lot of people lost money, but the winners won bigly — and in ways that came to benefit the whole world.
The housing bubble resulted in an awful lot of McMansions getting built, and love them or hate them, they made for a comfy and very real addition to the country's housing stock.
But if the AI bubble were to pop tomorrow, this time next year, or the year after, I'm not sure what surviving infrastructure we'd get out of it — and I don't think a trillion dollars' worth of rapidly depreciating Nvidia graphics cards counts.
Hell, come to think of it, I'm not entirely certain that Nvidia would survive the flood of gently used Nvidia cards hitting the market. I know some gamers who are already drooling at the prospect. But I'm honestly and openly ignorant about the "surviving infrastructure" angle, so please correct or add to this down in the comments.
Currently, SpaceX's closely held shares are valued at over $800 billion. While there's no way to predict exactly how much cash the company might raise by going public — particularly since we don't know how much of the company they plan to sell — my rough estimate is a metric butt-ton.
Anyway, the second part of "what's next?" is what SpaceX might do with all that IPO cash if the bubble pops and their orbital data center plans come to naught. If you read this, Elon, I have a few ideas — and not all of them involve an orbital low-gravity Barbarella reboot starring Sydney Sweeney.
The "how," at long last, is how SpaceX's initial public offering (IPO) is structured.
SpaceX is closely held, rather than public, so that its major shareholders would all be on board with Musk's Martian colony vision — there are no pesky board members demanding immediate profits over the company's long-term goal.
But that might change, depending on how the IPO is structured.
When Mark Zuckerberg took Facebook public in 2012, his IPO featured a two-class share system. Publicly traded Class A shares carried one vote each, while his and other insiders' Class B shares had 10 votes per share. That allowed Zuckerberg to retain approximately 57-60% of the company's voting power with only about 22-28% economic ownership. Reportedly, Zuck feared getting ousted from his own company the way Steve Jobs effectively had from Apple in 1985.
If SpaceX has a similarly protective structure, or they sell a very limited number of shares, then it's safe to assume Musk and the small number of current shareholders intend to keep the company's Mars ambitions intact. If the IPO allows voting shares to fly out the door faster than a Falcon 9 reaches orbit, then I'd call into question whether Musk is still serious about Mars.
In either case, I've instructed my business manager to get in touch with Elon's people about that Barbarella thing.
We'll do lunch.
Recommended: The Obamas' Power, Corruption, and Lies Are Things to Behold
Like what you’re reading? There’s a lot more where that came from.
PJ Media VIP members get deeper dives, exclusive columns, and commentary you won’t see on the public side. Best of all, you can get 60% off right now with the promo code FIGHT.
Sign up for PJ Media VIP and help us keep telling the stories that matter.
Join the conversation as a VIP Member