All podcasts / My First Million / Summary

The most simplified breakdown of the SpaceX IPO on the internet

2026-06-12 - 68 min - source - Read full transcript

Sam Parr (host)Shaan Puri (host)

Key insights

SpaceX is really three stapled-together businesses, not one.

The hosts break the company into launch services, Starlink (internet connectivity), and X/xAI (the AI and social bet), noting investors buying the IPO are effectively buying a launch company, an ISP, and an AI lab wrapped into a single ticker.

spacex-business-model

Starlink, not rockets, is the actual profit engine of SpaceX.

Starlink has quadrupled in two years to 10 million paying subscribers and about $11B in revenue at roughly 40% EBIT margins, with few real competitors and low ongoing cost once satellites are launched - the hosts call it the textbook 'cash cow' inside the larger company.

starlink-economics

The $1.75 trillion IPO price is roughly 100x revenue and only makes sense if you believe in Musk personally, not in conventional multiples.

One host frames it as 'the cult of Elon' versus rational-multiple investors; the argument is that Musk-run companies get priced on a 'price to Elon ratio' rather than price-to-earnings or price-to-sales, because the market is betting on execution, not current financials.

elon-musk-valuation-premium

The entire bull case hinges on an unproven rocket: Starship.

Starship, meant to carry 7-10x the payload of the Falcon 9 and dramatically cut launch costs again, has not yet been made to work reliably. Both Starlink's next leg of growth and the space-data-center thesis depend on Starship succeeding, which the hosts flag as the single biggest open risk in the filing.

spacex-business-model

SpaceX's next bet is building AI data centers in orbit, powered by free solar energy and free radiative cooling.

The pitch is that satellites already run on solar power and shed heat into space without active cooling, so streaming AI compute from orbit could undercut ground-based data centers on cost - if the physics and the build-out actually pencil out, which is unproven at scale.

space-based-ai-compute

The real obstacle to earthbound data centers is regulatory red tape, not technology - which is part of why Musk is going to space instead.

The hosts argue it is currently easier to solve orbital rocket engineering than to get local approval (e.g., a county permit) for a new data center in the U.S., given anti-data-center backlash and permitting delays; they compare this dynamic to colonial America's frustration with distant, unresponsive governance.

space-based-ai-compute

X (Twitter) is a financial drag Musk salvaged by cannibalizing it for AI, then monetizing spare compute.

X's ad revenue is about 40% of what it was before Musk's 2021 acquisition, and total X revenue (ads plus subscriptions) is down from $4.5B to $2.8B. Musk used X's user data to differentiate Grok, and when Grok still lagged Anthropic and OpenAI, he began renting out the Colossus GPU cluster - reportedly worth $20B+ in deals to Google and Anthropic combined - turning a weak consumer product into an AI infrastructure landlord business.

spacex-business-model

Betting against Elon Musk's technical execution has historically been a losing trade, even though his timelines are consistently wrong.

The hosts note a pattern across Tesla and SpaceX: skeptics who bet Musk could not deliver on hard engineering promises (reusable rockets, the original Tesla pay package targets) have repeatedly lost, even though he is chronically late on stated dates - the risk isn't whether he can build it, it's whether the timeline or his personal survival cooperates.

elon-musk-valuation-premium

Musk's new SpaceX pay package sets deliberately extreme, mostly binary targets.

The 'Mars Award' pays out one billion shares (~$135B at IPO price, growing with the stock) only if SpaceX hits a $7.5 trillion market cap AND a self-sustaining Mars colony of one million people. The 'AI CEO Award' pays 300 million shares if the company reaches $6.5 trillion in value and delivers 100 terawatts of compute per year from non-Earth data centers - about 100 times the entire current U.S. electrical grid's output.

elon-musk-valuation-premium

Musk has retained an unusually large ownership stake for a company this capital-intensive, at this age.

After 20 years and many funding rounds in one of the most capital-hungry industries possible, Musk still owns about 42% of SpaceX and controls roughly 85% of the vote - contrasted with Box founder Aaron Levie, who owned only about 4% of his company at IPO despite running a far less capital-intensive software business.

ipo-cap-table-winners

Investors who simply backed 'every Elon company' outperformed more sophisticated strategies.

Gigafund, started by Founders Fund alum Luke Nosek, built its entire thesis around indiscriminately investing in whatever Musk was building next. The hosts frame this alongside historical examples (holding Google/Facebook/Amazon, or Bill Gates not selling Microsoft stock) as evidence that picking an obvious long-term winner and holding beats constant sophisticated repositioning.

ipo-cap-table-winners

The IPO is expected to create thousands of new millionaires beyond the usual executive and VC winners.

SpaceX's S-1 reportedly implies the IPO will create over 4,000 new millionaires, including cafeteria and other blue-collar workers who received employee stock options - alongside big institutional winners like the Ontario Teachers' Pension Plan, whose 2019 investment is expected to return about $12B, or roughly $33,000 per teacher in the fund.

ipo-cap-table-winners

Books referenced

The Hitchhiker's Guide to the Galaxy - Douglas Adams - brought up while debating whether Musk's habit of pricing things at $420/$42,000/etc. is a nod to the book's '42' joke or just a weed reference (they conclude it's the weed reference)

Media referenced

Elon Musk biography - other - referenced in passing (title/author not stated on air) regarding Antonio Gracias's early role helping Tesla and SpaceX fix production bottlenecks
Wall Street Journal article on SpaceX blue-collar workers - article - cited as evidence that rank-and-file SpaceX employees are genuinely inspired by the company's mission, not just chasing a paycheck
Entourage - show - the Ari Gold paintball-gun-layoffs scene used as an analogy for how Musk reportedly reset the stalled Starlink team in Seattle

Companies

SpaceX - subject of the episode; IPO discussed in detail, valued at roughly $1.75 trillion
Starlink - SpaceX's satellite internet unit, described as the actual cash-generating core of the business: ~10 million subscribers, ~$11B revenue, ~40% EBIT margins
X (Twitter) - acquired by SpaceX for roughly $250B; ad revenue now about 40% of pre-acquisition levels, seen as a drag Musk offset by using its data to bootstrap Grok
xAI / Grok - Musk's AI company/model, described as behind Anthropic and OpenAI but propped up by X's user data and the Colossus compute cluster
Tesla - referenced as the template for Musk's earlier 'impossible' pay package and for his pattern of building factories that build factories
Cursor - SpaceX reportedly bought or holds an option to buy Cursor for $60B despite Cursor doing only $3-4B in revenue, framed as an AI-compute alliance play
Anthropic - renting compute from SpaceX's Colossus data center for roughly $1B+/month; also referenced as a company Sam Bankman-Fried was an early investor in
Google - signed a deal reportedly worth ~$12B/year to rent compute from SpaceX's Colossus data center despite being a major data center builder itself
TerraFab - Musk's planned chip factory, described as an attempt to get ahead of a chip production bottleneck for AI
Gigafund - VC fund started by Luke Nosek (Founders Fund alum) and Steve, built almost entirely around backing every Elon Musk company; expected to be one of the IPO's biggest winners
Ontario Teachers' Pension Plan - invested in SpaceX in 2019; the IPO is expected to net about $12B for the fund, roughly $33,000 per teacher
Box - cited for comparison: founder Aaron Levie owned only about 4% of Box at its IPO, versus Musk's ~42% of SpaceX after 20 years

Techniques and frameworks

Price-to-Elon ratio - the hosts' half-joking framework that Musk-run companies trade at a premium multiple purely because of who runs them, independent of standard valuation metrics
Failing forward - used to describe how Musk repeatedly repurposes a struggling asset (X's declining ad business feeding Grok's training data; Grok's shortfall justifying the Cursor tie-up) into the next bet
Pessimists get to be right, optimists get to be rich - framing used to explain why the hosts default to an optimistic read on ambitious technology bets despite frequent failures in the space
Adjusted EBITDA skepticism - hosts flag SpaceX's addbacks (stock-based comp, a Bitcoin markdown, one-time depreciation) as the kind of non-GAAP massaging Buffett and Munger dismissively called 'bullish earnings'

Summary

Sam Parr and Shaan Puri use SpaceX's IPO filing as an excuse to explain, in plain language, what the company actually is and whether it's worth the hype. Their framing from the outset: SpaceX is not one business but three stapled together - a dominant rocket-launch operation (80-85% of all payload sent to orbit), the Starlink satellite internet business, and a bolted-on AI/social bet spanning X (Twitter) and xAI. Starlink is the clear standout: roughly $11B in revenue, ~40% margins, 10 million subscribers and no real competition, funded by the launch business's collapsing cost-per-kilogram. The company is going public at around $1.75 trillion, a figure the hosts repeatedly frame as pricing in "Elon" as much as fundamentals - a "price to Elon ratio" rather than a conventional multiple.

The episode's most technical thread is the case for data centers in space: the pitch is that orbital chips get free solar power and free radiative cooling, potentially making space-based AI compute cheaper than anything buildable on Earth, where the real bottleneck is regulatory permitting rather than engineering. That bet - along with Starlink's next growth leg - depends entirely on Starship, SpaceX's next-generation rocket, actually working at scale, which it has not yet done. The hosts are candid that this is the single biggest open question in the filing, while also noting that betting against Musk's eventual technical execution has historically been a losing position, even though his timelines are reliably wrong.

They spend real time on the messier parts of the empire: X's ad revenue is down to about 40% of its pre-acquisition level, and Grok still trails Anthropic and OpenAI - but Musk turned that weakness into a new business line by renting out the Colossus GPU cluster to Google and Anthropic for a combined $20B+ in deals, effectively becoming an AI infrastructure landlord to his own competitors. They also flag the S-1's use of "adjusted EBITDA," which strips out items like stock-based compensation and a Bitcoin markdown, comparing it unfavorably to what Buffett and Munger dismissed as "bullish earnings."

Cap-table trivia rounds out the episode: Musk still owns about 42% of SpaceX and 85% of the vote after 20 years of fundraising, an unusually high retention rate compared to founders like Box's Aaron Levie. Gigafund, a firm built around simply backing every Musk company, stands to be one of the IPO's biggest winners, as does the Ontario Teachers' Pension Plan off a 2019 bet. The filing projects the IPO will mint over 4,000 new millionaires, including cafeteria workers with stock options. The hosts close on Musk's newly disclosed SpaceX pay package, which conditions huge share grants on near-impossible milestones - a self-sustaining million-person Mars colony, or delivering compute equal to 100x the entire U.S. electrical grid from off-planet data centers - framing it as classic Musk: set an absurd target, miss the deadline, and still eventually get closer than anyone thought possible.

Notable Quotes

"It's not about the price to earnings ratios. It's not about price to sales. It's about, you know, the price to Elon ratio. And if it's an Elon company, you're going to have 10 times the price you would probably otherwise have in a company." - Shaan Puri

"You're not buying SpaceX. You're buying three different companies that are stapled together. You have the space launch business, the internet connectivity business, and the AI business." - Sam Parr

"It's a wonderful business at a silly price... You can admire it from afar. You didn't need to own it." - Sam Parr, paraphrasing the "Charlie Munger" verdict he had Claude generate on the business

"Pessimists get to be right and optimists get to be rich. If you hang out in Silicon Valley long enough, you realize that pessimism is a losing strategy when you're around innovation and technology." - Shaan Puri

"Betting against Elon's technical ability has proven to be like the most unprofitable bet you could make, even if you're right for a year or two, you're eventually wrong." - Sam Parr