SpaceX just dropped its S-1 SEC filing, and it answers the question everyone in tech has been asking: what does the SpaceX IPO filing actually reveal? The short answer is a lot — including $18.67 billion in 2024 revenue, a bid to raise $80 billion or more, and an audacious claim to have identified the largest total addressable market in human history. Oh, and somewhere along the way, SpaceX quietly became an AI company. Here's everything that matters from one of the most anticipated IPO filings in a generation.
What Does the SpaceX S-1 Filing Actually Reveal?
SpaceX published its investor prospectus this week, giving the public its first detailed look at the financials behind Elon Musk's sprawling enterprise. The company reported $18.67 billion in revenue for 2024, a figure that surprised some analysts — including Dan Primack, who noted on CNBC that the business was smaller than he expected given the hype surrounding it.
02:15
Hosts react to the SpaceX S-1 being called the most enjoyable IPO read in years — and what that says about the company's positioning
Watch at 02:15 →
But the revenue number is almost beside the point. The S-1 is being widely praised as one of the most enjoyable reads in recent IPO history. Investor Kevin Kwak called it "the most enjoyable S-1 read in a long time — reads so easy, like sci-fi or fiction." That cuts both ways, of course: if you're a bull on space and technology, it's inspiring. If you're a skeptic, the same sentence sounds like a warning label.
The document opens with stunning, rarely seen photography before diving into dense text — a deliberate choice that feels like SpaceX held these images in reserve for exactly this moment. The company has grown from a startup that nearly went bankrupt to one of the world's most valuable private companies, with over 22,000 employees as of March 31st. It controls technologies that competitors and even nation-states haven't been able to fully replicate.
How Much Is SpaceX Raising — and When?
SpaceX is targeting $80 billion or more from its stock sale, with the rumored date circling around June 12th — roughly 20 days from the filing. If it hits that number, it would dwarf Saudi Aramco's $26 billion raise in 2019, currently the largest IPO in history.
05:40
The TAM slide breakdown: SpaceX claims the largest addressable market in human history at $28.5 trillion
Watch at 05:40 →
Goldman Sachs is leading the deal as left lead in the joint book-running managers, which is itself a big story. Morgan Stanley's Michael Grimes reportedly returned to the bank in part to win this mandate, making Goldman's victory a notable upset. At this size, though, essentially every major bank is getting a piece of the action.
One analyst described what it means to be a Goldman analyst on this deal: "It's not just the biggest IPO of all time. It's not just an incredibly complex structure with multiple businesses. You're also reporting to Elon Musk — and he's going to ask for things more aggressively than almost any other CEO." The SpaceX IPO is a career-defining event for everyone involved.
Among the biggest winners: Luke Nosek, who left Founders Fund to start Gigafund — a vehicle specifically focused on investing in SpaceX rounds. Conviction, as one observer put it, will do that.
Is SpaceX's $28.5 Trillion TAM Claim Realistic?
The S-1 includes what many are already calling the greatest TAM slide in IPO history. SpaceX states plainly: "We believe we have identified the largest actionable total addressable market in human history." They estimate a quantifiable TAM of $28.5 trillion, broken down as follows:
12:30
Breaking down the AI pivot: SpaceX's capital spend on AI is now 3x its spend on space
Watch at 12:30 →
- $370 billion in space-enabled solutions
- $870 billion in Starlink broadband
- $740 billion in Starlink mobile
- $2.4 trillion in AI infrastructure
- $760 billion in consumer AI subscriptions
- $600 billion in digital advertising
That last line — $600 billion in digital advertising — raised eyebrows immediately. Is that for X, Elon's social media platform? The filing doesn't fully clarify. The AI infrastructure numbers are more defensible given SpaceX's Colossus data center and its deepening partnership with Anthropic. Notably, for sizing purposes, SpaceX excluded China and Russia from its global estimates — a caveat that could look very different over the next 25 years depending on geopolitics.
The bull case valuation from one AI-assisted analysis puts SpaceX at $1.1–$1.5 trillion at fair value, with a bull scenario reaching $1.7–$1.9 trillion if AI infrastructure margins are strong and Starship unlocks new markets. A $2 trillion print would mean the market is assigning $800 billion to $1 trillion to the orbital AI compute story alone — possible as IPO mania, but aggressive.
Is SpaceX Now an AI Company With Some Rockets?
This might be the most striking revelation in the filing. According to the S-1, SpaceX's capital spend on AI is 3x its capital spend on space. As one observer summarized bluntly: "It's an AI company with some rockets."
That's a wild pivot for a company that spent 20 years as a rocket business, then evolved into a satellite internet provider with Starlink — a logical extension given its launch capacity. But the jump to AI feels different. The Colossus data center, built under the xAI umbrella, has scaled faster than almost anyone anticipated. And SpaceX's partnership with Anthropic is the clearest signal yet of how serious this pivot is.
22:10
Tyler Cosgrove walks through the Erdős unit distance problem and what OpenAI's solution actually means
Watch at 22:10 →
Tom Brown, co-founder of Anthropic, confirmed the companies are expanding their partnership and scaling up GB200 capacity on Colossus 2 throughout June. Anthropic is spending over $1 billion per month on compute — annualizing to roughly $15 billion per year. For context, SpaceX's total 2024 revenue was $18.67 billion. That one partnership is transformative. Overnight, SpaceX has become one of the largest neoclouds on the planet, and Anthropic is reportedly paying above-market rates for that compute.
Looking back, the narrative construction around this feels coordinated: late last year, conversations about space-based data centers suddenly emerged across investor circles. Valuation floats followed. Then came the Anthropic partnership, even as relations between Elon Musk and Anthropic had previously been more combative. Whether strategic or organic, the result is a company with a radically different story than it told 18 months ago.
Is Anthropic Actually Profitable in 2025?
Yes — and the numbers are staggering. The Wall Street Journal reports that Anthropic's Q2 revenue is set to reach $10.9 billion, up 130% over the previous quarter, and the company is on track to post its first operating profit. The WSJ's own URL reportedly calls it "mindblowing growth."
This puts to rest, at least for now, the persistent skepticism about whether large language model companies can ever be profitable. The argument that "AI will never make money" has taken serious damage. Dylan Patel and others have argued that leading models will eventually be able to raise prices because they're driving outsized economic value for customers — and the Anthropic numbers seem to support that thesis.
For those keeping score at home: Anthropic's valuation is up 173% since the start of the year. OpenAI's is up 67%. February 2026 Substack posts declaring generative AI a scam have aged... poorly.
Did OpenAI Just Solve a Decades-Old Math Problem?
In a week full of massive news, this one might have the longest tail. OpenAI announced that a general-purpose internal model solved Problem 90 from the Erdős problem set — the planar unit distance conjecture — without scaffolding or specialized math tools.
Here's the simplified version of the problem: given n points on a 2D plane, what is the maximum number of pairs of points that are exactly one unit apart? The Erdős conjecture held that this number scaled below a certain bound. OpenAI's model proved that for infinitely many values of n, the actual maximum exceeds that bound — disproving the conjecture with a novel 18-page proof.
What makes this significant isn't just the result. It's how it got there. This wasn't brute-force inference costing millions of dollars. Estimates suggest the computation cost hundreds to thousands of dollars in inference compute. The proof is genuinely novel — not a recombination of existing papers — and mathematicians including Terence Tao have called it impressive. Less than a year ago, frontier models were at IMO gold level. The pace of progress is accelerating, not plateauing.
Why Is the White House Investing $2B in Quantum Computing?
Rounding out the week's tech news: the White House announced $2 billion in grants to quantum computing companies, including $1 billion to IBM, with the government taking equity stakes in nine quantum computing firms. This isn't a grant in the traditional sense — it's an investment, meaning American taxpayers now hold a basket of quantum computing equity.
Rigetti Computing (not "spaghetti computing," despite what you may have heard on the show) is among the beneficiaries and saw its stock surge roughly 30% on the news. The move signals a federal commitment to not ceding quantum leadership to China — a strategic bet with a potentially very long time horizon.
The Week's Bigger Picture
This was a week that made even skeptics reconsider their priors. SpaceX is going public at a scale that will redefine the IPO record books. Anthropic is profitable and growing at a pace that defies the "AI is a bubble" narrative. OpenAI solved a math problem that stumped human mathematicians for decades. And the U.S. government is now a quantum computing investor.
The technology industry, as it turns out, is not out of news.
