2026-06-19
By Vadym · Generated with Boba, curated by me
SpaceX went public on June 12th at $135 a share, closed its first day at $161, and crossed a $2 trillion market cap before the week was out. Largest IPO in history. Four days later, the company announced it was buying Cursor — an AI coding tool — for $60 billion in all-stock. The stock ran to $225 intraday before pulling back to around $192.
That sequence tells you almost everything about what SpaceX is trying to become.
For most of its history, SpaceX was the company you couldn't invest in. It stayed private while consistently doing things that looked impossible — reusable rockets, autonomous drone-ship landings, internet from orbit. The secondary markets filled the gap. On platforms like Hiive, SpaceX shares changed hands at $212 apiece in July 2025, implying a $400 billion valuation. By December, an insider share sale pegged it at $421/share and $800 billion. By the time the S-1 landed in May this year, the IPO target was $1.75 trillion.
That's a 4x valuation expansion in under twelve months, before a single day of public trading. The Hiive final pre-IPO price was $149.92 — almost exactly where the stock opened on day one at $150. Secondary markets, it turns out, are reasonably efficient when there's enough trading interest.
The IPO priced at $135, opened at $150, closed at $161. Goldman, Morgan Stanley, and Bank of America led 23 underwriters total. Initial float was around 5% of total shares — a tight float, which matters for what comes next.
The company Wall Street just valued at $2 trillion has two real businesses. One is working. The other is the reason for the valuation.
The one that's working: Starlink. The satellite internet service generated $11.4 billion in revenue in 2025 — up 49% year over year — at a 39% operating margin. That's real money, from a real product, serving 10.3 million subscribers across 164 countries. Starlink represents about 61% of SpaceX's total revenue and is the only reason the company's consolidated finances look as good as they do.
The consolidated numbers are less clean. SpaceX posted $18.7 billion in total revenue against a $4.9 billion net loss and has accumulated $41.3 billion in deficits. The drag comes from two places: Starship R&D (around $3 billion a year burning into the future) and losses from the xAI segment that shows up in the consolidated books. Starlink is profitable. SpaceX as a whole is not.
There's also a compression problem inside Starlink. Average revenue per user has dropped from roughly $99 in 2023 to around $66 today — a 33% decline over three years. SpaceX is adding subscribers fast (analysts project around 17 million by end of 2026), but they're doing it by cutting prices and expanding into lower-yield tiers. Volume is growing; unit economics are shrinking. At some point those two curves need to cross — and they haven't yet.
The launch business — Falcon 9, Falcon Heavy, and eventually Starship — is the other pillar. SpaceX is the dominant commercial launch provider globally, with a manifest that includes NASA Artemis, DoD classified payloads, and Starshield (a dedicated military constellation that's the government-facing variant of Starlink). The details on Starshield are intentionally opaque — SpaceX doesn't break out segment revenue — but it represents meaningful government contract backlog that de-risks the revenue profile at least partially.
Here's where it gets interesting.
SpaceX buying Cursor for $60 billion is not an obvious move for a rocket company. Cursor is an AI-native coding tool with $4 billion in ARR that has been growing fast. The deal is all-stock — meaning SpaceX shareholders absorb the dilution, and the existing Cursor investors get SPCX shares. The announcement came four days after the IPO. The timing was not accidental.
What SpaceX is doing is visible in layers. The autonomous landing systems on Falcon 9 and Starship are genuinely sophisticated machine learning problems — real-time guidance, navigation, and control under conditions that don't have clean training data. Starshield is a defense AI play, not just a connectivity play; the government wants edge compute in orbit, and SpaceX has the orbital infrastructure to deliver it. The xAI segment already in the financials signals that the Musk-connected AI interests aren't cleanly separated from the SpaceX entity.
Cursor adds enterprise AI software revenue to a business that currently generates almost all its revenue from hardware and connectivity. It's an attempt to reframe SpaceX as a vertical stack — launch infrastructure, orbital connectivity, AI applications — rather than a company that makes rockets and sells internet.
That reframing matters enormously for the valuation multiple. Infrastructure companies trade at different multiples than launch providers. AI companies trade at different multiples than infrastructure companies. If the market accepts SpaceX as an AI-infrastructure company with a monopoly on orbital real estate, a 90x price-to-sales ratio becomes a story you can tell. If the market decides it's a capital-intensive aerospace company with a good internet business, Morningstar's $780 billion fair value estimate looks less crazy — and the current stock price looks about 2.5x too expensive.
The bull case for SpaceX rests on Starship — the fully reusable heavy-lift rocket that's supposed to dramatically reduce the cost of getting mass to orbit. The long-term thesis is that cheaper access to orbit transforms industries: satellite deployment, space tourism, point-to-point Earth transport, eventually Mars. If Starship delivers on its cost targets, the addressable market for everything SpaceX does expands enormously.
The problem is that Starship has a timeline problem. More precisely, Elon has a timeline problem. He has a consistent and documented pattern of announcing dates that slip by years while the underlying technology eventually arrives. Full Starship operationalization has been "next year" for multiple consecutive years. I don't think this means the vehicle won't work — it clearly is working, in test flight form — but pricing the bull case in dog years is the prudent adjustment.
Specific risks worth naming:
Lockup unlock supply overhang. The IPO prospectus includes an early unlock trigger: if SPCX closes above $175.50 (130% of the IPO price) on 5 of the 10 trading days before Q2 earnings, an additional ~10% of shares unlock early. The measurement window falls in late July and early August. With the stock having already traded well above that threshold, this trigger is live. That's a lot of supply that could hit the market right as the post-IPO excitement fades.
ARPU compression vs. subscriber growth math. Starlink's unit economics are moving in the wrong direction at the same time subscriber counts are growing. The bet is that volume wins. It might — but the margin profile that makes Starlink look like a good business today is not the same margin profile you get at 50 million subscribers if you've cut prices 40% to get there.
Key-man risk, extreme edition. SpaceX without Elon Musk is a different company than SpaceX with Elon Musk. The orbital infrastructure, the manufacturing, the culture — all of it is inseparable from him in ways that most key-man risks aren't. This is not unique to SpaceX as a risk category, but it is more acute here than almost anywhere else in public markets.
Political exposure. Musk's visibility in the political sphere creates regulatory exposure in markets where SpaceX wants to operate. EU regulatory friction, market access questions in Asia, and the inverse problem in the US — where close government relationships can turn into liability if administrations change — all add noise to what should be a clean infrastructure story.
The competition exists. Blue Origin's New Glenn is flying. Rocket Lab's Neutron is in development. Chinese state launch providers are advancing. None of them are SpaceX — the gap is real and wide — but "no serious competition" is no longer accurate.
SpaceX built something genuinely important. Reusable orbital launch was an engineering breakthrough that most serious people in aerospace said wasn't economically viable. Starlink is a real product with real margins serving real demand in places where terrestrial internet doesn't reach. The company has a track record of delivering things it promised, even if the timeline always slips.
The AI reframing with Cursor is aggressive, and I think it's partly real and partly valuation theater. The autonomy systems are genuine. Starshield is a real AI-in-orbit play. But buying a coding tool for $60 billion — 15x revenue — days after the largest IPO in history reads like a signal sent to investors as much as a strategic decision. The market heard it: the stock ran 40% in days.
At $2 trillion, you're paying for Starship delivering on its cost targets, Starlink scaling to 50+ million subscribers while defending its margin profile, and the AI label doing sustained valuation work rather than fading back to aerospace multiples. Any of those assumptions being wrong means the stock has significant downside from here.
SpaceX's own S-1 claims a $1.6 trillion total addressable market for Starlink connectivity alone. Morningstar's realistic estimate is $129 billion. The gap between those two numbers is the entire valuation debate. One of them is right. You need to decide which one before you think about the stock.
The timing for this post isn't accidental either. SpaceX being public changes what's possible. For the first time, the infrastructure play is accessible without a secondary market account and a tolerance for illiquidity. That's genuinely new. Whether the current price reflects that access fairly is a different question — and one I'd answer with considerably more skepticism than the opening-week enthusiasm suggests.
The infrastructure is real. The price reflects the story. And Elon's timelines — price in dog years.