Sam Altman and Elon Musk exchanged sharp tweets over the weekend, with Altman accusing Musk of selling investors on short‑term space data centers. The spat brings renewed focus to a consensus among industry experts: orbital data centers won’t become a viable business until rockets are far cheaper and high‑power satellites can be mass‑produced.
SpaceX’s plan to launch a fleet of orbital AI inference nodes fuels much of its $2 trillion valuation, and bullish analysts tout the potential of an “orbital neocloud.” Yet engineers and entrepreneurs alike point out that even a reusable Starship—targeted for its 13th test flight as early as July 16—won’t achieve the rapid, low‑cost turnaround needed for a data‑center market. SpaceX has already warned that full reusability may be years away, and that second stages will likely be discarded, undermining the economics of large‑scale deployments.
A single high‑speed compute satellite could see orbit next year, but scaling the concept to a meaningful fleet is likely a decade‑plus away, perhaps not until the 2030s. Investors should therefore temper expectations about near‑term returns from space‑based AI compute.
