SpaceX executives recently informed investors that they aim to launch initial demonstrations of orbital artificial intelligence computing infrastructure by late 2027. This timeline is earlier than the "as early as 2028" deployment date disclosed in the company's IPO filing. Gwynne Shotwell, President, and Bret Johnsen, CFO, stated these initial deployments would be demonstrator systems to validate the technology before a broader commercial rollout. This orbital computing effort is a central element of SpaceX's long-term growth pitch, as the company claims it possesses the "only company with a commercially viable path to building orbital AI compute at scale."
SpaceX anticipates the first version of these AI satellites to be powered by Nvidia chips, with the computing power of one satellite potentially equivalent to an Nvidia GB300 rack. CEO Elon Musk has indicated that building orbital data centers is not a particularly difficult engineering challenge, as much of the necessary technology is an extension of what is already utilized in the Starlink satellite network. These AI satellites are expected to be physically larger than current Starlink satellites but less complex in design, featuring extensive solar cells, radiators, and high-speed laser links for communication.
The initiative is being financed through a massive $75 billion fundraise as part of SpaceX's initial public offering, which values the company at over $1.75 trillion. This makes it the largest IPO in market history. SpaceX has sought regulatory approval to launch up to 1 million space-based data-center satellites. The success of this ambitious plan heavily relies on Starship, the fully reusable rocket, which is still years behind Musk's initial targets for full reusability, a critical factor for economically viable large-scale deployment.
