Elon Musk's SpaceX IPO values the company at $1.75 trillion, making it the largest initial public offering on record. The headline number masks the real story. SpaceX's AI division lost $6.4 billion last year, yet investors are pricing in an orbital data-center infrastructure play that could dwarf traditional space launches.
The bet hinges on deploying a million satellites designed to serve as distributed compute nodes for AI workloads. This constellation would create redundant, low-latency infrastructure for training and inference operations. The valuation has more than doubled since December, suggesting markets are absorbing this shift from aerospace to AI infrastructure.
SpaceX's rocket business remains profitable and essential, but it serves as a delivery mechanism for the larger vision. Starlink satellites, originally positioned as global broadband, function as foundational hardware for an orbital compute grid. The company's AI arm bleeds cash today because it's building toward monopoly-level control over space-based compute capacity.
This contrasts sharply with Apple's opposite bet. While Musk pushes compute into orbit, Apple optimizes on-device processing and keeps data local. Apple's neural engines reduce reliance on cloud infrastructure entirely. Both strategies hedge against cloud compute bottlenecks, but from opposite directions.
The timing matters. GPU scarcity and data-center build-out costs plague AI companies racing to scale model training. Orbital infrastructure sidesteps terrestrial real-estate constraints and regulatory friction around data centers. A million satellites offering distributed compute eliminates single points of failure and latency penalties for certain workloads.
The IPO values SpaceX's AI arm despite current losses because the long-term revenue model is theoretically unlimited. Renting compute slots on orbital infrastructure to AI labs, cloud providers, and enterprises could generate recurring revenue streams orders of magnitude larger than launch contracts.