Here is the thing about the SpaceX headline: Musk never said the company would make $1 trillion in revenue by 2030. That number is the projected size of the entire global space economy by 2034, according to industry analysts. Someone confused a total addressable market with a company forecast, and the headline machine ran with it.
It does not matter. The real number is worse for the bull case and more interesting for anyone trying to figure out what they actually own.
SpaceX went public on June 12 at $135 a share, opened at $150, and closed at $161, up 19 percent on its first day. That valued the company at roughly $2.1 trillion. It was the largest IPO in history. Musk became the first person whose paper wealth exceeded a trillion dollars. All of that happened on $18.7 billion in annual revenue and a $4.9 billion net loss.
The math does not add up unless you buy the AI story.
SpaceX has three businesses. Starlink, the satellite internet service, is the engine - it generated more than $11 billion of that $18.7 billion in 2025, and it is the only segment that looks like a mature cash business. The Space segment, which covers launches and government contracts, is the brand origin but not the growth story. Then there is the AI segment.
The AI segment brought in $3.2 billion of revenue in 2025 and lost $6.4 billion operating. It spent $12.7 billion building infrastructure. In the first quarter of 2026, the segment did $818 million - still burning through cash at a faster clip than it generates. This is a business that lost twice what it earned last year.
Yet Morgan Stanley projects that AI segment will produce $190 billion in revenue by 2030. Goldman Sachs says $322 billion. Evercore ISI is more generous - $755 billion by 2031. SpaceX's own prospectus claims a $28.5 trillion total addressable market across its businesses, with "the lion's share" being enterprise AI.
That is a 60-to-200x revenue increase in five years from a division that currently loses money. No AI infrastructure company has ever done that. Nobody has ever done that in any technology segment.
The way to think about SpaceX is not to ask whether the rocket business justifies a $2 trillion valuation. It does not. The way to think about it is to ask whether the AI business can become the thing Wall Street says it will be, and whether the market is pricing that outcome as a probability or a certainty.
Because the stock is priced as if it is certain.
$2.1 trillion on $18.7 billion in revenue is roughly 112 times sales. Aswath Damodaran at NYU, who teaches valuation, put it bluntly: this is a company with small revenues and large losses, and investors are paying a hundred times revenues for it. Morningstar's DCF model lands at $780 billion - less than half the IPO valuation. CFRA analyst Keith Snyder initiated coverage with a Sell and a $115 target, below the $135 IPO price. Fortune calculated that the actual addressable market may be closer to $129 billion, not the $1.6 trillion the prospectus implies.
None of this means the AI thesis is wrong. I have been wrong about AI before. The point is that the stock does not leave room for being wrong.
Here is the structure of the bet. You are buying three businesses under one ticker. Starlink is real and growing - revenue grew 33 percent from 2024 to 2025 - but growth slowed to 15 percent year-over-year in the first quarter of 2026. The launch business is a cash-generating franchise but it is not a trillion-dollar business. The AI segment is where the optionality lives and where the risk lives. You own the upside of Musk becoming an AI infrastructure magnate. You also own the risk that $12.7 billion in annual capex does not produce $190 billion in revenue.
That is not a conventional investment. It is a bet that one person's judgment about capital allocation in AI infrastructure is better than the judgment of every other public AI infrastructure company, multiplied by the leverage that his rockets provide. The rockets are real. The Starlink network is real. The AI bet is the variable.
I suspect the people who make money on this are not the ones who bought on day one because the IPO was the biggest in history. They are the ones who figure out whether the AI segment is a real business or a capital expenditure story wearing a revenue disguise. The way to test that is not to look at TAM slides. It is to watch whether the AI segment's operating loss narrows relative to revenue. If it does not narrow, the capex is buying dreams, not margins.
The lockup on early shares expires in September or December. Options trading began last week. The FOMO mechanics are already working.
The question for the person who owns SPCX is simple: do you believe the AI segment will narrow its loss and scale its revenue at the pace Goldman and Morgan Stanley assume? If yes, the $161 price might be cheap. If no, you bought a $129 billion business at $2.1 trillion.
Neither answer requires faith in Musk. Both require watching the quarterly numbers from the AI segment more closely than anyone else will.