According to The Rundown AI, SpaceX has acquired xAI with the goal of moving AI data centers off Earth and into space. Elon Musk's thesis is that terrestrial power limitations cannot sustain the exponential growth of AI compute, so SpaceX plans to deploy one million satellites as orbital compute nodes. The strategy involves Starship launches every hour, each delivering 200 tons of cargo, with a projected 100 gigawatts of AI capacity added annually. Musk predicts space-based compute will become more cost-effective than traditional data centers within two to three years. The long-term vision includes Moon factories, Mars settlements, and leveraging solar energy on a massive scale. With competitors like Google's Project Suncatcher also exploring space-based compute, SpaceX is uniquely positioned to commercialize this emerging AI infrastructure, according to The Rundown AI.

According to Sundar Pichai on Twitter, Google’s Project Suncatcher is pioneering the deployment of Trillium-generation TPUs—purpose-built for AI—into space, aiming to create scalable machine learning compute systems powered directly by solar energy. Early research cited by Pichai confirms these TPUs survived radiation tests simulating low-earth orbit, demonstrating their robustness for extraterrestrial AI workloads. The project, developed in collaboration with Planet, is scheduled to launch two prototype satellites by early 2027. This initiative signals significant business opportunities in space-based AI compute, offering the potential to overcome terrestrial energy limitations and unlock new markets in edge AI and satellite-based machine learning applications (source: Sundar Pichai, Twitter, Nov 4, 2025).