Analysis
SpaceX's first earnings report as a public company delivered a genuine beat and a stock decline in the same breath. Revenue came in around $7.8 billion for the quarter, nearly double what the company generated a year earlier and ahead of the $6.9 billion analysts expected, with adjusted EBITDA tripling as rockets, Starlink and AI compute infrastructure all beat internal targets, according to [Fortune](https://fortune.com/2026/08/04/elon-musk-spacex-earnings-trillion-revenue-target-capex-stock-reaction/).
The Capex Number That Spooked Investors
The number that actually moved the stock wasn't revenue -- it was capital expenditure. SpaceX spent $18.4 billion in the quarter, with nearly $16 billion of that going directly into AI compute infrastructure, well above the $13.2 billion Wall Street had modeled. That's a company reinvesting essentially all of its earnings growth back into AI infrastructure buildout, a spending pace investors are increasingly unwilling to reward on faith alone this earnings season.
“## The Capex Number That Spooked Investors The number that actually moved the stock wasn't revenue -- it was capital expenditure.”
Musk's $1 Trillion Pitch
Musk used the earnings call to move SpaceX's internal target for $1 trillion in annual revenue forward a full year, from 2031 to 2030, and floated a 'non-zero chance' of hitting it by 2029. He also pitched building robots on the Moon as part of the justification for directing so much capex toward AI compute rather than core launch and Starlink operations, a plan he acknowledged sounded 'totally nuts' even as he defended it.
A Familiar Pattern This Earnings Season
Shares fell more than 5% anyway, extending a pattern that's shown up across AI-infrastructure-adjacent earnings this week: AMD and Uber both beat their trailing quarters and still sold off on forward spending or guidance concerns. SpaceX's case is more extreme because the company is now explicitly pitching a decade-scale, trillion-dollar vision to justify near-term capex that's growing faster than revenue.
The Real Risk
The bear case here isn't subtle: $16 billion in one quarter on AI compute is a bet that frontier-lab demand for that capacity keeps compounding at the same pace indefinitely, and any slowdown in Anthropic, Google or other compute customers would leave SpaceX with expensive infrastructure and a much harder trillion-dollar case to make.
What to watch: how SpaceX's AI infrastructure revenue specifically performs relative to this quarter's spend in the next earnings report, and whether the moon-robot plan produces any disclosed contract or customer commitment beyond Musk's own framing on the call.
