SpaceX Isn’t a Rocket Company Anymore. The IPO filing proves it.
What $2 Trillion Actually Buys You: Chips, Satellites, AI, and a Mars Colony in the Contract
In 2008, Elon Musk was sleeping on friends’ couches. He had put his entire PayPal fortune into a rocket company that had just exploded three times in a row, and the fourth attempt was all or nothing. If it failed, SpaceX would disappear.
No backup plan, no reserve fund, nothing.
Eighteen years later, that company just filed for what is shaping up to be the largest initial public offering in recorded history. The target valuation is $1.75 trillion, with Bloomberg reporting that early investor conversations have already revised the figure above $2 trillion. The investor roadshow is confirmed for June 2026, targeting a listing around Musk’s birthday on June 28.
What’s emerging from the leaked documents is something far bigger than an IPO. What Elon Musk is developing is a unified industrial conglomerate, responsible for producing its own AI, robots, semiconductor chips, rockets, and building communication systems across Earth, orbit, and eventually Mars. The last time a single entity controlled this many capabilities across this many domains, it was a nation-state, not a private company.
Not a Rocket Company
To understand how a company that launches rockets can be valued at more than the GDP of 93 percent of the world’s countries, you need to understand what SpaceX actually is today. Most people, even those following closely, have the wrong picture.
In 2025, SpaceX launched 165 orbital missions on the Falcon 9 alone. That’s roughly one launch every two days, sustained for an entire year without pause. The company controls approximately 82 percent of the global commercial orbital launch market. For perspective: that’s as if a single airline operated 82 percent of all commercial flights on Earth.
The rest of the industry can largely watch.
But the rocket launches, as spectacular as they are, aren’t where SpaceX makes its money anymore. SpaceX is now a telecommunications company that also builds the world’s best launch vehicles. The reason for that shift is Starlink. By early 2026, over 10,000 satellites are projected to be in orbit, serving more than 10 million subscribers globally. These satellites will provide coverage for rural areas, shipping lanes, commercial aviation, military operations, and increasingly dense urban markets. Industry estimates place Starlink’s 2025 revenue at around $10 to $11 billion, with operating margins well above 50 percent. Musk himself has said that NASA will represent roughly 5 percent of SpaceX’s 2026 revenue. Everything else is commercial, driven by millions of monthly subscription payments that arrive like clockwork.
So when you hear “rocket company valued at $2 trillion,” that’s not quite what you’d be buying. What you’d be buying is a near-monopoly on orbital launch, a rapidly expanding global internet infrastructure with over 10 million paying customers, and $22 billion in already signed government contracts. Oh, and they absorbed xAI in an all-stock deal in February 2026, which means you’re also getting one of the world’s most capable large language model stacks.
The Numbers Nobody Expected
The confidential S-1 filing has leaked, and the financial picture is more complicated than anyone anticipated.
The Information reported, confirmed by Reuters, that SpaceX posted more than $18.5 billion in revenue for 2025. Not $15 to $16 billion, as previously estimated. Nearly $3 billion more than expected. But here’s the twist: the company also reported a consolidated loss of approximately $5 billion. One year prior, SpaceX had generated around $8 billion in profit. The swing from $8 billion profit to a $5 billion loss in a single year has one explanation: xAI.
Musk’s AI venture caused capital expenditures to nearly triple, reportedly climbing from around $5.6 billion in 2024 to approximately $12.7 billion in 2025. SpaceX isn’t lighting that money. It’s deploying it across infrastructure, GPU clusters, and data centers, and it needs engineering talent to compete with OpenAI, Anthropic, and Google. Meanwhile, Starlink remains massively profitable as the financial engine that funds everything else. By the end of 2026, Quilty Space’s analyst forecasts suggest Starlink will achieve roughly 16.8 million subscribers, with anticipated company revenue in the range of $22 to $24 billion.
The takeaway: the core business is strong and growing. The losses are a deliberate investment decision, not a sign of operational weakness.
The Real Reason for $2 Trillion
SpaceX doesn’t command a $2 trillion valuation based on what it does today. It commands that number because of what Starship will make possible tomorrow.
Starship is the largest and most powerful rocket ever built. In its reusable configuration, it can transport between 100 and 200 tonnes to low Earth orbit. The current Falcon 9 carries roughly 23 tonnes. That’s a fivefold to nearly tenfold increase in payload capacity per launch. But the capacity isn’t the revolutionary part. The cost is.
Today, sending one kilogram to orbit on a Falcon 9 costs commercial customers between $2,700 and $3,000 — already the lowest price on Earth by a wide margin. ULA’s Vulcan sits at around $4,000 per kilogram. Europe’s Ariane 6 ranges between $4,000 and $6,000. A decade ago, the standard was $15,000 to $50,000. Starship’s stated target is $10 to $100 per kilogram to orbit. That’s a reduction of 30 to 300 times compared to Falcon 9, and more than 1,000 times cheaper than the industry standard ten years ago.
How? Complete reusability. Since the dawn of the space age, every rocket was built for hundreds of millions of dollars, launched once, and either burned up in the atmosphere or sank to the ocean floor. That’s like Air France buying a brand new Airbus A320 for every Paris-to-Marseille flight and scrapping it on arrival. Nobody would run an airline that way, yet it’s exactly how every space agency operated for sixty years.
SpaceX started changing that with Falcon 9 by recovering the first stage. Starship pushes the concept to its conclusion: both stages return, refuel, and relaunch. The vehicle costs approximately $90 million to build. Fuel runs $500,000 to $2 million per launch. Fly it 100 times, spread that construction cost of 100 missions, add operational expenses, and you arrive at roughly $2 to $5 million per launch for 100 to 150 tonnes of payload. That arithmetic produces approximately $33 per kilogram to orbit.
On April 14, 2026, SpaceX completed the first full static fire of the Starship Version 3, with all 33 Raptor 3 engines on Booster 19 firing simultaneously, generating approximately 9,240 tonnes of thrust — more than any launch vehicle in history. The inaugural V3 flight, designated Flight 12, is targeting early to mid-May 2026. If it goes well, it lands just before the investor roadshow in June. That timing is not accidental.
The Jevons Paradox of Orbit
There is an economic concept called the Jevons Paradox that explains what happens next. When something becomes dramatically cheaper, demand doesn’t stay stable. It explodes. This is exactly what happened with commercial aviation after deregulation in the 1970s and 80s. The same travelers didn’t simply pay less for their flights. Millions of entirely new passengers flew for the first time. The market multiplied.
The same dynamic is assembling around space. At $10 to $100 per kilogram, you can place massive constellations of specialized AI satellites in orbit. You can manufacture pharmaceuticals and advanced materials in microgravity. You can deliver 100 tonnes of emergency humanitarian supplies anywhere on Earth in 45 minutes — a capability the U.S. Department of Defense is already exploring. And the biggest AI companies are collectively hitting the same wall on Earth: not enough energy, not enough cooling, not enough political goodwill to build data centers. Microsoft restarted Three Mile Island — the site of America’s worst nuclear accident — because there simply isn’t enough grid capacity. In orbit, solar panels generate power continuously with no nightfall, no electricity bill, and no angry neighbors filing permits.
ARK Invest, which holds SpaceX as the largest position in its venture fund, has argued that at sub-$100-per-kilogram launch costs, orbital data centers could provide AI compute at roughly 25 times lower cost than terrestrial alternatives. Some analysts project 5,000 to 10,000 Starship launches to build orbital AI infrastructure at scale. At $2 to $5 million per launch, that’s $10 to $50 billion — less than what the major tech companies already spend annually on ground-based data centers.
The Document That Changed Everything
On April 23, Reuters published an exclusive based on excerpts from SpaceX’s confidential SEC filing. What it revealed goes well beyond financial disclosures.
SpaceX will adopt “controlled company” status upon going public. That means the board will not require a majority of independent directors. No independent compensation committee. No independent nominating committee. Only the audit committee must be composed of independent members. Roughly 3 to 4 percent of Russell 3000 companies operate this way. Musk’s Class B shares carry ten votes per share versus one vote for the Class A shares sold to public investors, making his removal from the board effectively a self-vote.
The compensation structure is where the filing becomes something genuinely unprecedented in corporate history. Musk would receive 200 million super-voting restricted shares that vest only when two conditions are met: SpaceX reaches a $7.5 trillion market capitalization, and the company establishes a permanent human colony on Mars with at least one million inhabitants. A separate tranche grants up to 60.4 million restricted shares tied to the completion of non-Earth-based data centers capable of delivering 100 terawatts of compute per year.
These are not tweets. They are not press declarations. They are contractual conditions of executive compensation filed with the financial regulator of the United States government. Either Musk believes these goals are achievable, or he’s prepared to forfeit assets worth hundreds of billions of dollars on milestones he cannot reach. Given his track record of turning implausible bets into functioning hardware, the calculus is worth taking seriously.
Terafab and the Convergence
In March 2026, Tesla, SpaceX, and xAI jointly announced Terafab: a semiconductor fabrication complex in Austin, Texas, with an estimated budget of $20 to $25 billion. The project targets 2-nanometer process technology, one terawatt of AI compute output per year, and full integration of chip design, lithography, fabrication, memory production, advanced packaging, and testing under one roof. Intel joined the project on April 7, bringing its 18A manufacturing process. Construction began within weeks of the announcement.
The facility will produce two chip designs: one for Tesla’s autonomous vehicles and Optimus humanoid robots, and a radiation-hardened D3 processor built specifically for orbital AI satellites. Musk stated publicly that 80 percent of Terafab’s compute output will be directed toward space-based applications. Not 30. Not 50. Eighty percent.
Chamath Palihapitiya, the prominent venture investor and former Facebook executive, said in January 2026 that he did not believe SpaceX would pursue a traditional IPO. He predicted that SpaceX would eventually merge with Tesla in a reverse merger — one entity, one balance sheet. Dan Ives, one of the most widely followed tech analysts on Wall Street, has maintained his forecast that the two companies will merge by 2027, with Terafab representing the first concrete step in that operational integration. Tesla’s $2 billion investment in xAI has already been converted into SpaceX equity following the February merger. The two companies are financially linked for the first time in their history.
On prediction markets, the probability of a full Tesla-SpaceX merger oscillates between 22 and 26 percent, depending on the time horizon. That is far from certain. It is also far from fantasy.
What You’re Actually Pricing
When the IPO opens, retail investors will purchase SpaceX shares for the first time. For years, it was the most coveted private company in the world. Access required secondary markets with massive premiums, reserved for accredited investors with over $1 million in net worth or $200,000 in annual income. Everyone else was locked out. SpaceX reportedly plans to allocate up to 30 percent of the offering to retail investors, three to six times the industry norm. A dedicated event for approximately 1,500 participants is scheduled for June 11. Twenty-one banks are on a single deal. That tells you exactly how much demand the underwriters are anticipating.
For the rest of the industry, the IPO is existential in one direction and validating in another. Boeing’s Starliner has been a cascade of leaks, delays, and cost overruns. ULA’s Vulcan is more expensive and less capable than Falcon 9. Ariane Space continues to lose market share. When SpaceX raises $50 to $75 billion in fresh capital and channels it directly into Starship development and Starlink expansion, the competitive gap widens further. And yet, when the IPO filing became public, Rocket Lab rose 5.5 percent. Planet Labs jumped 11 percent. Intuitive Machines gained over 10 percent. A SpaceX IPO at this scale validates the entire orbital economy as a legitimate institutional investment class. More capital flows to the sector as a whole.
Gwynne Shotwell, SpaceX’s President and COO — the person who has run the operational side of this company for over twenty years alongside Musk — told Time Magazine last month that she was eager to see the IPO complete and described it as “a new way of running a company.” When someone at her level says that, it’s worth paying attention to what she means.
Here, we are establishing a fully integrated, interplanetary enterprise responsible for manufacturing its own AI, robots, chips, energy, rockets, and communications systems, with a presence on Earth, in orbit, and with an ambitious trajectory towards Mars. Nothing comparable exists in history. The last entity with this scope of capability across this many domains was a government.
AI connects every layer. The orbital data centers run AI. The Optimus robots are built on AI. Tesla’s autonomous driving stack is AI. Starlink will serve as the AI’s operational foundation. Terafab’s chips are the physical substrate of AI. Every piece of this ecosystem feeds back into every other piece, and that recursive integration is precisely what the IPO is asking you to price.
Whether you think $2 trillion is visionary or insane, the question isn’t really about the number. It’s about whether one company can build the infrastructure that shapes the next several decades of human civilization, from energy and communications to transportation, intelligence, and exploration. The filing says SpaceX is betting everything on yes.
Thanks for reading. Sometimes I wonder if we’re in a transition period or just watching an entire industry figure out in real time that nobody actually knows what they’re doing. Let me know your thoughts in the comments.