SpaceX Has Lost an Entire Tesla
Trending Topics Jakob Steinschaden
SpaceX stock has crashed 50% in six weeks, wiping out over $1.2 trillion — basically all of Tesla's value. The hype was AI, not rockets, and reality is catching up fast.
Six weeks ago, SpaceX went public in the biggest IPO ever, pricing at $135 a share and briefly touching $225.64 before gravity took over. On Monday it closed at $113.50 — its 13th losing session out of 16 — and Tuesday's pre-market pointed even lower, around $110. Do the math and you get a loss north of $1.2 trillion, which happens to be roughly Tesla's entire market cap, gone in a little over a month.
The valuation never made sense on rocket-and-satellite math alone. At the IPO price, SpaceX was worth $1.75 trillion against $18.67 billion in 2025 revenue — a price-to-sales ratio above 80, versus 1.5 to 3 for normal aerospace companies. What justified that gap wasn't Starlink's solid 33 percent growth. It was the idea, cemented by February's xAI acquisition, that SpaceX had quietly become a vertically integrated AI company. Investors bought the story; now they're re-reading the numbers. Q1 showed Starlink throwing off $1.19 billion in operating profit on $3.26 billion in revenue, while the AI unit burned $2.47 billion. That's not a rounding error.
Making things worse, only about four percent of shares were actually tradable at listing, so the stock has been whipping around on thin volume rather than genuine institutional conviction. Even getting added to the Nasdaq 100 on July 7 couldn't hold the price up — it fell seven percent that same day. Add a $25 billion bond sale landing right as the broader market soured on AI capex spending, and you have a stock primed to overreact in both directions.
The most exotic piece of the pitch — orbital data centers — is getting picked apart in real time. Analyst Peter Zeihan ran the numbers and found that a single one-gigawatt space-based data center would need roughly 500 Starship launches, which he flatly called a bad idea from multiple angles. Even SpaceX's own prospectus admits the tech is unproven and might never pencil out commercially. Meanwhile the actual AI money is being made on the ground: Anthropic pays $1.25 billion a month for Colossus compute, and Google is set to pay $920 million monthly for GPU access starting in October.
Opinion on Wall Street is split almost comically. Jeremy Grantham called this the craziest IPO he's ever seen and put collapse odds at 90 percent, arguing SpaceX's AI offering is mediocre at best. Yet Goldman, JPMorgan and most of the sell side maintain buy ratings with an average target near $237 — nearly double where the stock trades now. Cathie Wood's ARK is still adding shares, betting on the company's roughly 10,800 Starlink satellites and the idea that Earth-bound compute is running out of room. Operationally, at least, things are going fine — Starship's 13th flight nailed every mission objective on Friday. But the calendar is unforgiving: earnings land August 4, and two days later up to 911.5 million shares unlock, dwarfing the current float. You don't need to believe SpaceX is doomed to sell now — you just need to believe a supply wave is ten days away.
My take
This is what happens when a rocket company lets Wall Street rebrand it as an AI company without the AI business actually working yet — the market priced in a fantasy and is now pricing in the lockup calendar instead. Orbital data centers requiring 500 launches per gigawatt isn't a moonshot, it's marketing dressed up as physics, and it deserved the scrutiny it's finally getting. None of this means SpaceX is a bad company; it means investors confused 'important company' with 'correctly priced stock,' which is a very expensive mistake to make twice in one decade.
Read more about this at: Trending Topics