The Cost and Network Limits of Space-Based AI Compute

2026-07-15

AI Summary

This study evaluates the feasibility of large-scale AI data centers in low Earth orbit as an alternative to terrestrial facilities. Accounting for key differences between orbital and ground-based systems—including launch costs, power supply, thermal dissipation, radiation exposure, and atmospheric re-entry—it proposes a mesh network architecture based on laser inter-satellite links. The work introduces, for the first time, dual-cut bandwidth, dual-cut strength, and roofline models to systematically assess the performance of space-based AI computing. Comparative analysis with Clos networks reveals fundamental limitations of the orbital environment for training large language models: while orbital data centers can support AI inference tasks, they fall significantly short of ground-based counterparts in terms of cost efficiency and scalability for training state-of-the-art foundation models.

Abstract

This paper evaluates whether large-scale AI data centers deployed in low-Earth orbit (LEO) could become a cost-effective alternative to terrestrial facilities. The analysis compares orbital and ground-based systems across launch cost, power generation, cooling, radiation exposure, and atmospheric reentry, as well as compute-network performance. A key distinction is the shift from terrestrial Clos networks to space-based mesh networks using laser inter-satellite links. Using bisection bandwidth, bisection intensity, and roofline-style models, we show that while LEO-based inference may be feasible, training frontier-scale LLMs in orbit is unlikely to be competitive with terrestrial data centers.

Problem

• space-based AI
• low-Earth orbit
• data centers
• large language models
• cost-effectiveness

Innovation

• space-based AI
• laser inter-satellite links
• mesh networks
• bisection bandwidth
• roofline model