The New Space Race Nobody Saw Coming

Will the next great infrastructure race be fought 500 kilometres above earth?

- Rohit Chintapali

ELON MUSK has never been shy about thinking in civilisational terms. But even by his standards, the proposition he has been advancing with increasing conviction is remarkable: that data centres belong not on Earth but above it.

The practical argument is straightforward. Land is expensive. Power grids are strained. Cooling a data centre in the Arizona desert requires staggering volumes of water. In space, solar energy is uninterrupted, cooling is a function of radiating heat into an infinite vacuum, and the real estate is, for the moment, free.

Musk's Starlink constellation, which has placed more than 10,000 satellites into low Earth orbit since 2019, was originally about internet connectivity. Increasingly, it feels now like the scaffolding for something larger, an orbital computing network that internal SpaceX projections suggest could eventually scale to one million nodes, making it the largest distributed data centre in human history, answerable to no municipal power authority and visible from every point on Earth simultaneously.

Jeff Bezos, who has spent an estimated $10-20 billion building Blue Origin into a credible space venture, is pursuing a parallel logic, but through a different architectural path. In March, Blue Origin submitted plans to the US Federal Communications Commission for Project Sunrise, a proposal to deploy 51,600 data entre satellites in low Earth orbit.

The Economics of Orbit

So is any of this economically viable? The honest answer, for now, is: only if you own the rocket.

SpaceX's structural advantage is almost unfair because it manufactures its own launch vehicles, every satellite it deploys goes up at cost, with no margin paid to a third party. For everyone else, launch costs remain the governing constraint. The economics only work at scale, and scale only works if you already have the rockets.