SpaceX, Nvidia Target 2027 Launch for Orbital AI Data Centers

SpaceX is deepening its push into artificial-intelligence infrastructure by choosing Nvidia as the exclusive computing platform for its AI services. The alliance extends beyond terrestrial data centers into orbital computing, where SpaceX’s satellite and launch capabilities could open a new market for AI processing. The move may reshape competition among chipmakers and emerging cloud providers as power availability and access to advanced accelerators become critical constraints on industry growth.

SpaceX plans to expand its Nvidia-based computing capacity to nearly 10 gigawatts by the end of 2027. The companies are also developing Starmind, a space-based data-center project whose first computing satellites equipped with Nvidia GPUs are scheduled to launch in 2027. The announcement lifted Nvidia shares 3.4%, while rival Advanced Micro Devices fell 7.04%, reflecting investor concern that an exclusive SpaceX-Nvidia architecture could narrow AMD’s opportunities in a potentially large new market.

Generative AI is driving up data centers’ power and cooling requirements, prompting SpaceX founder Elon Musk to extend computing infrastructure into Earth orbit. The concept combines Starlink’s satellite deployment expertise with solar power and optical communications networks. If SpaceX can overcome cooling and maintenance constraints, the project could create a formidable moat in AI hardware and launch capabilities that rivals would struggle to replicate.

SpaceX has unveiled AI1, its first AI data center satellite, featuring a modular design and massive solar arrays. The company aims to deploy 1 GW of space-based computing capacity in 2027. Responding to engineers who questioned the feasibility of dissipating heat in a vacuum, Musk pointed to the more than 10,000 SpaceX satellites already operating in orbit. The project could also generate potential orders for Taiwan’s ODM and optical communications supply chains.