SpaceX CEO Elon Musk has pushed back against concerns regarding the company’s capital expenditure efficiency, stating that terrestrial data centers will be "peanuts" compared to the scale of compute planned for orbit. The comment addresses a recent analyst estimate suggesting SpaceX needs to spend roughly $470 billion over the next five quarters to meet its infrastructure goals.
What Happened
Musk responded on X to a post by Milk Road Stocks citing analyst Kyle Reidhead, who estimated the $470 billion figure based on SpaceX’s Q3 earnings call target of 13 gigawatts or more of compute by the end of 2027. Reidhead’s breakdown suggests Nvidia captures 83 to 89 percent of this spending. Currently, SpaceX ended the quarter with 1.7 GW of nameplate compute and guided for $35 billion to $40 billion in fourth quarter capex.
While Musk did not attach a specific number to his rebuttal, the company’s financial trajectory supports his claim of high efficiency. CFO Bret Johnsen noted during the earnings call that new compute pays for itself in under a year at current pricing, a window Musk said could shrink to six months next year. Based on this pace, Reidhead argues that approximately $230 billion of the estimated spend could come from operating cash flow, and $60 billion from SpaceX’s $90 billion cash pile, leaving about $180 billion to borrow. Musk confirmed last week that SpaceX had already closed a $40 billion debt raise to purchase GPUs.
The ultimate goal, according to Musk, is the transition from ground-based infrastructure to orbital compute. Replying to charts derived from Sequoia partner Shaun Maguire’s model, Musk wrote that "Terrestrial compute will be peanuts compared to orbital compute," predicting that orbital compute deployed per month will likely exceed terrestrial levels around the end of 2028 and far surpass them in 2029. Musk added regarding the model’s projections, "I think we can do better."
Why It Matters
This strategic shift highlights the critical role of the Starlink ecosystem and future orbital infrastructure in supporting the AI boom. The orbital plan relies on "Starmind," a network of up to 1 million AI satellites flying in clusters of about 10, linked by roughly 10 terabits of bandwidth. Deployment of thousands of these units is planned to start as soon as late 2027 from the Gigasat factory in Bastrop, Texas. Each Starmind spacecraft is expected to carry more than 250 kW of solar power.
The feasibility of this ambitious roadmap depends heavily on launch volume. Maguire’s model assumes 250 Starship flights in 2028, a significant increase from two flights this year and 38 next year. Financially, SpaceX has guided for $33 billion to $37 billion in first quarter 2027 revenue, more than double the $14.58 billion booked in Q3. Musk reposted this revenue guidance on Sunday morning, underscoring the company's confidence in its growth trajectory despite the massive capital requirements.
The Bottom Line
Musk’s dismissal of terrestrial data centers as insufficient underscores SpaceX’s pivot toward becoming a dominant player in orbital AI infrastructure. By leveraging high-capex efficiency and rapid Starship deployment, the company aims to surpass Earth-bound compute capacity by 2029, transforming its business model from launch services to a comprehensive space-based computing utility.