SpaceX trades 5 per cent below the IPO mark it once beat by $900bn

Peak-to-trough and IPO-to-trough are two different measurements, and the trillion belongs to the first. SpaceX priced at $1.8 trillion ^[1], buying after the listing took it as high as $2.7 trillion ^[2], and by the mid-July low near $1.4 trillion the fall from that peak was about $1.3 trillion ^[1]. Measured from the price the deal was actually struck at, the low sat about $400 billion below it ^[2]. The $900 billion in between ^[13] came from the initial surge of buying ^[2].

The stock now sits about 5 per cent under the offer ^[4], which implies roughly $1.71 trillion ^[3], while the S&P 500 gained nearly 4 per cent over the same stretch ^[4]. That is a relative gap of about nine points ^[4]. Nine points is what an offer-price buyer has given up against the index.

The cash flow statement explains more than either valuation does. First-half capital expenditure of $28.5 billion was up 308 per cent on the same six months of 2025 ^[5], which puts the prior-year half near $7 billion ^[7], and 86 per cent of the new figure went to AI ^[6], about $24.5 billion ^[5], much of it into the Colossus data centres ^[14]. Annualised, that capex line runs at $57 billion ^[14]. Evercore ISI estimates it could reach $360 billion by 2030, most of it AI ^[12].

Revenue of $12.5 billion in the first six months ^[7] annualises to $25 billion ^[8]. Musk said SpaceX could generate $3.5 trillion of revenue by 2033 ^[8], which is 140 times that base ^[9] and needs about 103 per cent compound growth for seven straight years ^[10]. In a separate post he wrote: "You don't seem to understand that SpaceX will be worth more than the rest of Earth if we accomplish our goals." ^[9] Management puts the addressable market for the AI business, Grok plus sold data centre capacity, at $26.5 trillion ^[10], and the current revenue base is 0.09 per cent of it ^[11]. Microsoft's Azure, the working comparison the Motley Fool reaches for, does $100 billion a year ^[11], so the 2033 figure is 35 times it ^[12].

My read is that the market has accepted the offer price and declined the aftermarket one, and that the thing which has to be earned is the capex. The other reading, or rather the more interesting version of it, is that data centre capacity gets contracted before it is poured, in which case the $24.5 billion is prepaid revenue and 5 per cent below the offer is a discount. The second half decides between them: revenue annualising well above $25 billion while capex growth comes off 308 per cent supports the discount case, and capex doubling again without revenue following puts the $1.8 trillion mark itself in play. The Motley Fool's own explanation for the faltering price is that investors do not believe the spending spree will pay off ^[13].