Reading SpaceX as Compute, Not Just Rockets

With SpaceX’s S-1 on file and a June listing in view, Monte Investments declines to read it as a rocket IPO. The hardware case is real enough — the desk counts 652 launches with roughly 85% on reusable boosters, the lever that has taken launch cost down by a claimed 65–85% — but reuse is treated as the enabler, not the thesis. What it enables is everything stacked on top.

The stack is where the read gets interesting. Starlink is framed as distribution (the desk cites about 10.3m subscribers and a $19.6bn spectrum arrangement with EchoStar); the xAI combination is framed as compute, pulling the Colossus data-centre effort inside the same entity. Put together, Monte’s SpaceX is less a launch provider than a vertically-integrated infrastructure compounder — and the piece pushes the logic one step further, to compute that does not stay on the ground.

“Orbital AI inferencing may be the next frontier as a component of SpaceX’s broader infrastructure theme.”

Monte Investments

That is the sentence that turns a launch story into a compute story. Whether or not orbital inference arrives on anyone’s timetable, the framing is the useful part: distribution in low orbit, compute in the same corporate tent, and a roadmap that treats the satellite layer as future data-centre real estate rather than a comms business.

Why it’s worth your time

Monte’s habit is to start from a theme and end on a company, and this is a clean example: a contrarian frame on a name everyone will be looking at for the wrong reason. It sits naturally beside the way Closelooknet reads the build-out — compute, power and the physical layer underneath the model — so if you find inference economics interesting, this is the same question pointed at orbit.