SpaceX’s only problem is finding more space to work with

SpaceX’s IPO marks the moment the space dream became a mainstream capital markets story. For decades, it has sold a vision of interplanetary civilization on behalf of the wider space industry. In one sense, not much has changed. That vision remains central to its identity. What has changed is that the market is no longer treating that ambition as distant science fiction. No company raises more than $85 billion on vision alone. SpaceX has reached the public markets as something rarer: a company whose long-term mission is extraordinary, but whose near-term infrastructure is already becoming essential. That belief is not built on hype. It is built on a refusal to accept the old aerospace consensus and the ability to execute a new vision. SpaceX put reusability at the center of launch architecture and forced the rest of the industry to rethink what a rocket company could be. Then it built Starlink, and forced the world to rethink telecommunications too. Now, as it moves into orbital data center, it will need to rethink the orbits at which satellites fly as well. The vast majority of the more than 10,000 satellites in space currently operate in low Earth orbit (LEO), roughly 500 to 700 kilometers above the Earth. That number is set to increase dramatically, with SpaceX applying to put up to 1 million new satellites into space. The reality is that, if orbital data centers are to become viable, we will need to think seriously about operating satellites even lower, in very low Earth orbit, or VLEO, roughly 200 to 300 kilometers above the Earth. That is because LEO is already becoming crowded. In addition to the satellites already in space, tens of thousands of pieces of debris are being tracked in orbit, with millions more fragments too small to track. Starlink satellites executed around 300,000 collision-avoidance maneuvers in 2025, a 50% increase from 2024. If we move from 10,000 satellites to one million in LEO, the number of maneuvers needed to avoid collisions could become extraordinary. That is not an argument for writing off orbital data centers. The investment case is compelling, given the pressure AI is placing on the world’s electrical grids. Eric Schmidt, the former CEO of Google, has predicted that AI could account for 99% of total electricity generation, while some estimates suggest annual spending on grids will need to double to $970 billion by 2050. If SpaceX is right that orbital data centers can help solve this problem, it could end up controlling a significant share of global AI infrastructure within five to ten years. SpaceX is also likely to hold that advantage for a long time, given no other company can yet match the scale, complexity and cost profile of its Falcon 9 and Falcon Heavy rockets. However, there has been far less discussion about where these satellites will go, and what happens if they collide with space debris. The Kessler effect, long discussed as a theoretical risk, describes a scenario where one collision creates thou…