$SPCX is moving from Pentagon contractor to Pentagon infrastructure layer. The Pentagon is discussing a multibillion-dollar AI-compute deal with SpaceX, and the talks could still fall apart. The number matters, but the stack it would complete matters more. What SpaceX actually owns SpaceX already launches national-security payloads and supports military communications and missile tracking, so compute would add a fourth critical layer to the same relationship. After absorbing xAI, it also gained Grok, the Colossus data centers and a large installed base of Nvidia hardware. The company has since signed compute agreements with Anthropic and Google, with stated run rates totaling roughly $26 billion a year, although termination rights mean that figure is not locked-in revenue. This is no longer a rocket company with an AI side project. It is building a chain that runs from launch vehicles to satellites, connectivity, data centers and potentially the workloads running on top. Why the market should care For shareholders, the underwriting framework is changing. Launch provides strategic access, Starlink provides recurring connectivity revenue, and AI compute adds a capital-intensive infrastructure business with demand from hyperscalers and government. The Pentagon’s proposed $30 billion Artificial Intelligence Arsenal reinforces the point: high-end chips and compute are being treated less like ordinary IT spending and more like strategic capacity. The real moat is vertical integration, because SpaceX can combine launch economics, satellite networks, power procurement, data-center construction and government relationships across contracts that competitors usually attack separately. A Pentagon compute deal would therefore matter beyond near-term revenue, since becoming embedded in military AI infrastructure creates switching costs that are operational and political, not just technical. The Pentagon’s paradox The Pentagon says it wants to reduce dependence on any single technology provider, yet adding SpaceX for compute could deepen dependence on one company across launches, communications, tracking and AI infrastructure. Multi-cloud procurement does not eliminate concentration risk when the same vendor controls several adjacent systems that are difficult to replace during a crisis. The same integration that strengthens SpaceX’s competitive position also raises the stakes of outages, contract disputes, governance problems and shifting political relationships. Bottom line: this is not a cloud contract bolted onto a rocket company; it is SpaceX trying to become the infrastructure layer beneath U.S. military power, and the market should price both the moat and the concentration risk.