The Multiplanetary Trade

SpaceX filed its S-1 on May 20, ran a quick roadshow starting June 4, and prices after Thursday's close. Trading starts Friday, June 12 on the Nasdaq under SPCX. They went with a fixed price — $135 a share, no price-range games — selling about 556 million shares to raise around $75 billion at roughly a $1.75 trillion valuation. Biggest IPO ever, about three times Aramco's record.

The run-up has been wild: $350 billion in late 2024, $800 billion at the December tender, $1.25 trillion after the xAI merger in February, and now $1.75 trillion. Yes, the last leg more than doubled in six months — but the $250 billion xAI deal happened inside that window, so it's not quite apples to apples. The S-1 also showed what that deal costs: revenue topped $18.5 billion in 2025, with a net loss of almost $5 billion.

If you're writing a $1M check, two structural things matter. One, this is a controlled company, period. Dual-class shares give Elon about 85% of the votes on roughly 42% of the equity, and they'll lean on Nasdaq's controlled-company exemptions. You're buying the economics, not a say. Two, the deal is built for retail like nothing before it — about 30% of the offering, $22.5 billion or so, goes straight to retail brokerages. Triple the usual. Lockups open September through December. So the listing is real. The only open question is what Friday's open does to a fixed $135 price.

165 Falcon launches in 2025. A record for the sixth year straight, more orbital flights than the rest of the planet combined — a rocket roughly every 53 hours. They've flown 68 more this year through June 8. The whole machine runs on reuse: one Falcon 9 booster just flew its 35th mission. Every reflight spreads a ~$30M first stage over another paying launch, which is why estimates put SpaceX's internal cost around $600–650/kg while they charge ~$2,700/kg. Competitors are trying to break even at SpaceX's price. SpaceX makes money at its cost.

The cleaner way to see it is upmass. In 2025, SpaceX put up roughly 2,213 of the ~3,020 metric tons the whole world sent to orbit — about 73%. All of China, the number-two launch power on Earth, managed maybe a tenth of that. At some point it stops being market share and starts being market ownership, and it compounds while every rival flies expendable or close to it.

Starship is act two, and it's been messy but it's moving. Twelve flights so far. They've caught the booster with the tower and reflown it with most of its engines intact. Flight 12 in May flew the new V3 vehicle and deployed 22 payloads — though they lost the booster on the way back, and they still haven't caught the ship itself. V3 is supposed to do 100+ tons to orbit, fully reusable. If that works, the engineering estimates say under $200/kg and eventually under $100/kg — a 10–30x drop in the cost of getting to space. To be clear, those are targets, not numbers anyone has hit. But the moat isn't the targets, it's the decade of flight data nobody else has. Blue Origin reportedly lost a year-plus to its pad explosion in May. Vulcan, Ariane 6, Long March — all still expendable.

Forget the rockets for a second. What the S-1 actually shows is a subscription telecom growing at startup speed. Starlink did $11.4 billion of revenue in 2025, up about 50%, and hit 10.3 million subscribers across 164 markets by the end of Q1. It's also the only part of SpaceX that makes money — $7.2 billion of EBITDA at a 63% margin. That one line is what turns this from a launch company priced on dreams into a real recurring-revenue business.

There are three growth tracks. Consumer broadband keeps scaling as the dishes get cheaper — though watch the trade-off: ARPU has dropped from $99 a month in 2023 to $66 now, because growth is shifting to cheaper markets. They're winning on volume, not pricing power. Then direct-to-cell: 650+ satellites up, live in about 22 countries, T-Mobile's service commercial since last July, carriers signed everywhere from Japan to Ukraine. And then government — $4.2 billion of enterprise and government revenue in 2025, plus a $2.29 billion Space Force contract in May. Sticky money, but politically exposed, and it's about 20% of total revenue.

And here's the honest version of the "Starlink pays for Mars" story: the $7 billion of EBITDA covers Starship's ~$3 billion a year of R&D just fine. It doesn't cover the whole ambition. Capex hit $20.7 billion in 2025, plugged with $26 billion of new debt and preferred — that's what the IPO money is actually for. So no, Starlink doesn't pay for Mars today. What it is, is a real 60%-margin annuity growing 50% a year. Every quarter it compounds, the speculative stuff gets cheaper to own.

When I first sketched this thesis, "X, xAI and SpaceX as one ecosystem before 2030" was a prediction. Now it's a filing. The dominoes went down fast: xAI swallowed X in March 2025. SpaceX put $2 billion of corporate money — not Elon's — into xAI that July. xAI raised $20 billion at ~$230 billion in December. And then on February 2, SpaceX just bought xAI outright, all stock, $1.25 trillion combined. (Worth saying out loud: an all-stock deal between two companies the same guy controls isn't exactly price discovery.) Grok and X are now, in Elon's words, the AI division of SpaceX. That combined company is what lists tomorrow.

The stated logic is orbital compute. In January, SpaceX filed with the FCC for up to a million solar-powered satellites running as an orbital AI data center. The roadmap is surprisingly concrete — Starlink V3 sats running Grok inference late this year, prototype 150 kW compute satellites in early 2027, a million-square-foot satellite factory ramping through next year. The bull case is obvious once you see it: one company owns the rocket, the network, the model, the distribution, and increasingly the power supply. AI's bottleneck is electricity, and these are the only people with a plausible way around the grid.

Tesla is the asterisk, and it's a telling one. Tesla shareholders voted down an xAI investment in November — and Tesla put $2 billion in anyway a few months later, structured around the failed vote. That tells you two things at once: the convergence is real (Grok ships in Teslas, Tesla now holds equity in what became SpaceX), and the governance is exactly as Elon-dependent as the skeptics say. I've tried to keep the line between fact and extrapolation visible here — the mergers and filings happened; the orbital data-center economics are riding on hardware that has never flown.

First, be clear about where the base is, because it moved fast: $350 billion in December 2024, $800 billion a year later, $1.25 trillion after the merger, $1.75 trillion tomorrow. Now the uncomfortable part. 100x from $1.75 trillion is $175 trillion. That's about one and a half times the GDP of the entire planet, and roughly 34 Nvidias. At a generous 30x earnings, a $175 trillion SpaceX has to earn about $6 trillion a year — more than every US company combined earns today. To do it in 20 years you need ~26% compounded, every single year, for two decades, from the biggest IPO base ever.

And the precedents don't help as much as people think. Sure, companies have done 100x in 20 years — from small bases. Apple was a ~$50 billion company in 2006. Amazon was around $16 billion. Nvidia was single-digit billions. Nobody, ever, has even 10x'd from a trillion. And honestly, the 100x-from-cheap trade already happened — just to other people. Whoever bought the December 2024 tender is up about 5x in eighteen months. Buying at $135 on Friday means buying after the repricing, not before it.

So what does a great outcome actually look like from here? Maybe 5–10x over 20 years — a $9–18 trillion company by the mid-2040s, two or three times bigger than anything that exists today. Even that needs Starship to work, Starlink to keep compounding despite falling ARPU, and xAI to win real share. And it pencils out to roughly 8–12% a year. Index-fund returns wearing a rocket suit. The thesis can be right — cheap mass to orbit might genuinely be the platform of the century — and the 100x can still be basically impossible from this entry point. Believing in the company and being sober about the price aren't opposites. That's the whole discipline.