SpaceX’s public-market debut was supposed to put rockets, satellites and artificial intelligence under one soaring growth story. Instead, its first earnings report left investors wondering whether the company’s future is in orbit—or in the costly, crowded business of renting out computing power.
The numbers arrived first. Second-quarter revenue rose 92% year on year to $7.81 billion, led by $4.29 billion from Starlink connectivity and $2.56 billion from AI, while the space unit brought in $962 million. Losses narrowed to $541 million, but shares fell more than 8% after hours as enthusiasm gave way to concern over the scale of the investment still to come.1
That split is central to the debate. SpaceX’s AI revenue more than tripled from a year earlier, powered largely by compute deals with companies including Anthropic and Google. Yet the division lost $1.5 billion in the quarter, while total capital spending reached $18.37 billion.2 Musk’s answer is speed: “We’re building AI compute capacity at scale faster than anyone else,” he told investors.2
The company then made its hardware wager explicit. Musk said SpaceX would build “exclusively on Nvidia” because it considers the Vera Rubin architecture the best AI computer, a decision that shuts out rival chip suppliers and concentrates supply risk in a single partner.3 He echoed that position on X, writing: “SpaceX has committed to using Nvidia GPUs exclusively because they are the best.”
4 In a separate repost, he promoted Nvidia-powered satellite compute as AI infrastructure heading to orbit.5
Skeptics see less a space company than a telecom-and-neocloud hybrid. Starlink remains the only profitable operating segment, while AI spending alone hit $15.8 billion in the quarter—far above spending on space and connectivity—and the compute business faces the familiar risks of obsolescence and price competition.6
Musk is selling a much larger horizon: data centers in space, heavier Starlink launches and a $100 billion annual revenue run rate. Wall Street, after the first report, is asking a more immediate question: how long that future can remain a loss-making one.