SpaceX seeks US approval to deploy one million satellites for orbital AI compute

SpaceX has submitted an application to the U.S. Federal Communications Commission seeking permission to deploy as many as one million small satellites in low-Earth orbit configured to perform on-orbit data processing for artificial-intelligence workloads. The filing reframes satellites from communications relays to active compute nodes and emphasizes potential reductions in terrestrial power and water consumption compared with conventional data centers. SpaceX did not include a specific rollout timeline or detailed cost model in the submission; instead the filing outlines the architecture's intended capabilities and asks for regulatory clearance to operate large numbers of processing-capable craft. The company argues the system could serve billions of users and would require new spectrum allocations, cross-operator coordination, and low-latency edge-to-space software stacks. Engineers and independent observers point to major technical hurdles: launching and deploying very large counts of mass-bearing satellites, heat rejection for dense electronics in vacuum without conventional liquid cooling, radiation-hardening of processors, and long-term maintenance or replacement strategies in a congested orbital environment. Space-traffic management and collision-avoidance planning are central concerns—so are interference and visibility impacts on optical and radio astronomy that previous constellations have already highlighted. Parallel research from academic teams proposes alternative forms of orbital compute infrastructure—most notably tethered systems with thousands of modular nodes that share power and radiator capacity along long cable structures. That tethered approach aims to reduce the need for millions of separate launches and to use passive tether dynamics for orientation, local thermal management and redundancy, while still relying on laser optical links for high-throughput data transfer. Both the constellation and tether concepts claim potential environmental trade-offs by shifting energy and water demands off Earth, but they transfer engineering and governance burdens to space: different architectures change debris profiles, failure modes and rendezvous/servicing strategies. Regulators reviewing SpaceX's filing are likely to press for detailed debris-mitigation plans, spectrum-management commitments, and evidence about safe operations that protect other space users and scientific observing. Commercial viability will hinge on continued reductions in launch cost per kilogram, robust thermal and radiation solutions for sustained compute, and clear economic comparisons with terrestrial cloud providers. The filing signals intensifying industry and academic interest in relocating at least some AI workloads to orbit, and it brings forward policy questions about traffic management, shared-resource governance, and protections for scientific sensing as multiple technical pathways compete for regulatory and market acceptance.