Google: Space Data Centers Need 1,800 Starship Launches

Google has offered a sobering estimate for the future of space-based data centers, saying SpaceX's Starship would need to launch around 1,800 times before orbital computing infrastructure becomes viable. The figure, reported by TechCrunch, underscores how far the concept remains from practical deployment. Google has already sent its first advanced chip into orbit to test the idea, but the company's own math suggests that reaching scale will require a launch cadence far beyond what is currently routine. Space data centers are promoted as a way to sidestep some of the constraints facing terrestrial facilities. In orbit, servers could potentially rely on abundant solar power and avoid competition for land, water, and cooling capacity on Earth. As AI workloads grow, those constraints are becoming more pressing, prompting interest in unconventional solutions. Yet the obstacles are enormous. Launching hardware into space is expensive, risky, and energy-intensive. Satellites must also be maintained, shielded from radiation, and kept cool in a vacuum, where heat dissipation works very differently than in a data center on the ground. Upgrading or repairing equipment is far harder. The 1,800-launch estimate puts the challenge in stark terms. Even if Starship dramatically lowers the cost per kilogram and increases flight frequency, achieving that many launches would require a mature, highly reliable, and routine space transportation system. It would also demand a supply chain capable of producing and integrating vast amounts of computing hardware for orbit. For now, Google's orbital chip test is an experiment, not a product roadmap. The company appears to be exploring whether space-based computing could eventually complement Earth-based infrastructure, rather than replace it. The concept may become more attractive if launch costs fall and AI demand continues to soar. But the estimate makes clear that orbital AI infrastructure is still a long-term bet, not an imminent answer to data center capacity problems.