source: techcrunch ai: sam altman’s space data center trash talk is what most experts already believe
level: business
sam altman and elon musk traded insults online over the weekend, with altman calling out musk for selling investors on near-term space data centers. altman's jab reflects what many experts already think: space-based computing is not a serious business yet. spacex's plans to launch orbital data centers for ai inference are a key part of its $2 trillion valuation, but specialists across startups, google's orbital compute project, and independent engineers agree the economics don't work without drastically cheaper rockets and high-volume satellite production.
musk's counter relies on starship, spacex's massive new rocket set for its 13th test flight soon. if starship becomes rapidly reusable, the business case could improve. however, even if the upcoming test recovers both stages, operational reusability is likely years away. spacex also admitted during its ipo roadshow that starship may not be fully reusable soon, requiring expendable second stages that would kill the economics for space data centers. musk's claim that flights start next year ignores the real challenge: scaling manufacturing and launches to make it profitable.
the timeline for viable space data centers likely stretches into the 2030s. while spacex could launch a single data-processing satellite next year, mass deployment remains distant. the gap between vision and reality is wide, and public market investors may be overlooking these hurdles. for now, the consensus is clear: space compute is a long-term bet, not an imminent revolution.
why it matters: for ai and data science, the hype around space data centers could misdirect investment away from more practical, near-term compute solutions.
source: techcrunch ai: sam altman’s space data center trash talk is what most experts already believe