As AI Expands to Orbit, Can Ground-to-Space Links Keep Pace?

AI workloads are increasingly spreading from data centers to satellite systems in low Earth orbit, requiring high-capacity data links between ground stations and space. The atmosphere creates a major bottleneck for these ground-to-space optical communications, as turbulence, clouds, and water vapor distort laser signals and cause data loss. Rather than only reacting to atmospheric degradation with tools like adaptive optics, a proposed solution uses an atmospheric digital twin—a continuously updated computer model fed by weather data and real-time network sensors—to forecast communications performance and shift traffic to clearer stations before problems occur. This predictive approach could transform space-to-ground optical links from unreliable best-effort connections into managed resources that operators can plan around, much like power and cooling, making distributed and orbital compute practical for AI infrastructure.