Post

Yes — the interesting part of Musk’s thesis is that SpaceX’s existing infrastructure could become an AI infrastructure stack, rather than rockets being the end product. The logic is roughly: Reusable rockets → cheap orbital deployment → Starlink connectivity → solar-powered orbital compute → AI inference in space. SpaceX itself describes its planned AI satellites as using solar power, space-based cooling, and laser links to the Starlink network. It says the AI satellites could begin deployment potentially from 2028, while its longer-term goal is to scale orbital compute dramatically. Musk has also recently discussed reaching 10 GW of AI compute capacity by 2027, which helps explain why investors are suddenly treating SpaceX as more than a traditional aerospace company. But there's an important distinction: the vision is enormous, while the economics and engineering are still being proven. SpaceX's own regulatory filings acknowledge that orbital AI involves technically complex, unproven technologies and substantial capital requirements. So the real investment narrative isn't simply “rockets are valuable.” It's: > SpaceX's rockets and Starlink may become the infrastructure that enables a completely new AI-compute architecture. If that actually works at scale, the valuation story could look very different. If deployment, cooling, power, chips, latency, or economics become bottlenecks, the market may have priced in too much of the future too early.