SpaceX's $1.75 Trillion IPO: At 100x Revenue, You're Buying Three Companies and a Rocket That Hasn't Flown

Source: My First Million | Published: 2026-06-12T10:55:28Z

SpaceX is going public at a 100x price-to-sales ratio, and a huge chunk of that $1.75 trillion valuation is paying for a Starship that hasn't successfully flown and space data centers that don't exist yet.

You're Not Buying One Company—You're Buying Three Stapled Together

SpaceX currently generates $18 billion in annual revenue with $6.6 billion in adjusted EBITDA, but burned through $8 billion in cash last year. Break it apart and it's really three completely different businesses: space launch, internet service (Starlink), and AI (xAI + Colossus + forthcoming space data centers). Then there's Twitter—the "uninvited party guest" that got thrown in at $25 billion.

The launch business is dominant on a categorical level: 85% of global launch missions and 80% of orbital mass go through SpaceX. It has driven the cost per kilogram to orbit down by 50 to 100x. But launch itself isn't the profit center—it's the railroad, the infrastructure connecting everything else.

Starlink Is the Money Printer

Starlink grew from zero to 10 million paying subscribers in four years, hitting $11 billion in annual revenue with a 63% EBITDA margin—and it's recurring revenue. No real competitors, massive cost advantages. This business checks every box investors dream about.

But growth is slowing. Q1 2025 numbers are no longer dazzling. As Starlink expands into underserved regions like Africa and India, average revenue per user is declining—these markets need internet but can't pay the thousand-plus dollars a year that American users do. Where Starlink's ceiling actually lies depends on how much value it can extract from these regions.

Direct to Cell: The Next Growth Story

There's a direction that doesn't get enough attention but has enormous potential: Direct to Cell—satellite signals connecting straight to phones, no dish required. SpaceX has already partnered with T-Mobile to provide texting, calls, and data in areas with no ground coverage.

The business logic is straightforward: the global mobile and home internet market is roughly $2 trillion, and carriers have almost zero differentiation—whether you pick T-Mobile or AT&T basically comes down to whose spokesperson you like better. Starlink introduces a genuinely different value proposition for the first time: Our service works everywhere. Theirs doesn't. If this becomes a $3 to $10 add-on per phone plan, the scale gets very real.

Twitter's "Failing Forward" Playbook

Twitter's ad revenue is half what it was pre-acquisition. Current annual revenue sits at $1.8 billion, plus $1 billion from subscriptions and payments—$2.8 billion total, well below the $4.5 billion at the time of purchase. By conventional standards, this was a failed acquisition.

But Elon turned it into data fuel for Grok. When Grok couldn't keep pace with Anthropic and ChatGPT, he pivoted again—turning Grok's AI infrastructure, Colossus (the world's largest GPU cluster), into a rental business. Over the past two months, Anthropic signed a $1.25 billion per month deal and Google signed one for $920 million per month. Google, itself a top-tier data center builder, is now renting compute from Colossus.

These are short-term agreements cancellable with 90 days' notice—potentially worthless in five years. But the "failing forward" pattern is clear: Twitter data feeds Grok, Grok isn't competitive enough so rent out the infrastructure, and each failure becomes a stepping stone to the next move.

Space Data Centers: Insane or Inevitable

The core narrative in SpaceX's investor presentation is space data centers. It sounds like science fiction, but there's a very real driving force behind it: The biggest obstacle to building data centers on the ground isn't engineering—it's permitting. Red tape is harder than rocket science. Companies with the money, equipment, and engineering talent can't get the permits.

Elon's reasoning: satellites in space are powered by the sun, and the heat generated by chips dissipates naturally through radiative cooling—no need for the expensive cooling systems ground-based data centers require. If space inference costs 50% to 200% less than terrestrial, SpaceX becomes the Saudi Arabia of the compute era—the largest, lowest-cost "energy" supplier, except instead of pumping oil, it's serving AI tokens.

The analogy is powerful but also dangerous. Saudi oil is already in the ground. SpaceX's space data centers don't exist yet. Everything hinges on whether Starship works.

Starship Is the Linchpin

Starship's payload capacity is 7 to 10x that of Falcon 9—Falcon 9 delivers 10 satellites per launch, while Starship can deliver 70. But it hasn't flown successfully yet. On a podcast, Elon said the goal is 30 launches per day, operating like an airport. When pressed—"you can't even launch once right now"—he replied: "Airplanes went through the same thing."

History is on his side. Betting against Elon's technical capabilities has proven to be one of the most expensive wagers you can make—even when you're right in the short term, you're almost always wrong in the long run. But without Starship, space data centers are a fantasy and Starlink expansion decelerates. A large chunk of that $1.75 trillion valuation is paying for a rocket that hasn't flown yet.

42% Ownership: A Miracle After Two Decades of Fundraising

Rockets are among the most capital-intensive businesses on Earth, and after twenty years of continuous fundraising, Elon still holds 42% of shares and 85% of voting rights. For comparison, Box founder Aaron Levie had just 4% left at IPO—and he was building cloud storage, not rockets.

This means buying SpaceX stock gives you effectively zero governance rights. This is a company where Elon calls every shot, as the mission statement makes clear: "Make life multiplanetary, understand the true nature of the universe, and extend the light of consciousness to the stars." Compare that to your company's mission statement and feel the gap.

That Adjusted EBITDA Nobody Can Make Sense Of

SpaceX's S-1 relies heavily on "adjusted EBITDA," adding back stock-based compensation and depreciation among other items. For a capital-intensive company that spent $20 billion in capex last year, depreciation is a very real cost. Buffett and Munger called this kind of thing "bullshit earnings"—if you're paying people in stock, why do you keep adding that expense back every quarter? Does it just disappear?

The S-1 has some other interesting details: the company holds roughly $2 billion in Bitcoin, and Bitcoin price declines are listed as an adjustment item. Also, if you search for "420" in the filing, it's everywhere—from equity pricing to valuation figures. Elon's obsession with this weed-culture meme has seeped into the legal documents.

The Mars Award: The Most Absurd Compensation Package in History

Elon's base salary is zero. His entire compensation comes from two milestone-based awards.

The first is the "Mars Award": 1 billion shares of SpaceX stock (roughly $135 billion at current prices), contingent on market cap reaching $7.5 trillion and establishing a self-sustaining colony of at least one million people on Mars. Both conditions must be met simultaneously.

The second is the "AI CEO Award": 300 million shares, contingent on market cap reaching $6.5 trillion and delivering 100 terawatts of compute annually from non-terrestrial data centers. For reference, the entire U.S. power grid currently has a total capacity of about 1 terawatt. He needs to build the equivalent of 100x the American grid—in space.

These numbers are so absurd they sound like a joke. But the last time he set "absurd" targets at Tesla, every analyst who laughed got proven wrong.

Who's Getting Rich Off This IPO

Second-largest shareholder Antonio Gracias holds roughly 7% and is expected to cash out around $90 billion. He's been Elon's close friend for twenty years, helped optimize production lines at Tesla and SpaceX in the early days, and personally lent Elon money when Tesla was on the brink of bankruptcy—without asking for any equity in return.

Gigafund's story is even more interesting. After Luke Nosek left Founders Fund, he and his partner did something that seemed "not sophisticated enough" at the time: raise a fund that only invests in Elon's companies. That's it. Everyone thought it was too simple—not what a serious fund should do. But in hindsight, the most profitable strategy of the past 15 years was buying Google, Facebook, and Amazon and then doing nothing. People chase complexity and cleverness, but the biggest returns often come from strategies so simple you'd be embarrassed to say them out loud.

The Ontario Teachers' Pension Plan invested in SpaceX in 2019 and is expected to net $12 billion post-IPO—roughly $33,000 per teacher in the fund.

SBF Could Have Been the Biggest Winner

If Sam Bankman-Fried hadn't illegally invested customer funds and ended up in prison, his portfolio today would be staggering: his early Anthropic stake would be worth $80 billion, SpaceX $15 billion, Robinhood $5 billion, Cursor over $3 billion (he was likely one of the earliest investors), and Solana $5 billion. Over $140 billion total—enough to make him one of the greatest investors of all time. These assets were sold off on the cheap during bankruptcy proceedings, much like when Tim Draper bought Silk Road Bitcoin from the U.S. government at a discount—the real winners are the people who found treasure in the rubble.

This Company's Biggest Risk Is Its Founder's Lifespan

Look at SpaceX through a traditional valuation framework and you'll conclude "way too expensive." A 100x price-to-sales ratio, negative free cash flow, and a core growth story that depends on technological breakthroughs that haven't happened yet. A Munger simulated by Claude would say: Good business, absurd price, put it in the "too hard" pile.

But Elon's companies aren't normal companies. When you discount future cash flows far enough out, the biggest variable is no longer whether the technology is achievable—it's how long this man will live and how long he can maintain his execution capacity. It's a strange investment thesis: the more confident you are that the engineering risk is manageable, the more the remaining core risk starts to look like an actuarial problem.