The Real Story About Data Centers in Space

This month, SpaceX went public and made Elon Musk the world's first trillionaire — on the strength of an idea that doesn't exist yet: data centers in space.

This week, one of the most successful investors alive called that idea a fantasy. Masayoshi Son, who built SoftBank betting on the future, told his shareholders the whole thing makes little sense. The point of putting data centers in orbit is to save on electricity, he said — and electricity is the cheap part. The chips are what cost the money.

Same idea. Same month. Worth a trillion dollars or worth nothing, depending on who held the microphone.

I was already chewing on this a few days earlier, sitting in a Space Capital event listening to Jonny Dyer, the CEO of Muon Space. The phrase "data center" came up the way it comes up everywhere now — like weather. But Dyer didn't pitch the warehouse. He pointed at the thing underneath it.

Here's the part the coverage tends to skip: what is a data center in space, actually.

A data center is just a building full of computers. It's the thing that answers your search, your prompt, your map. Right now those buildings sit on Earth, and they are hungry — for land, for water, for power. A data center in space is the same building, in orbit, running on sunlight. That's the whole concept. The cloud, which was always just someone else's computer in a warehouse, finally goes where the warehouse can sit anywhere.

Why would anyone want that? Three reasons, and they're real. The sun never sets in orbit, so the power is free and constant. Earth is running out of welcome — nobody wants a data center drinking their town's water, and the grid is already strained. And launch finally got cheap and reliable enough to make the math worth running at all.

Why would anyone doubt it? Also three reasons, also real. Space is cold, but it's a vacuum, so the heat the computers throw off has nowhere to go — you need radiators the size of the spacecraft just to keep them from cooking. Radiation chews up the electronics over time. And launch, even now, still costs five to ten times too much for the numbers to close. You can't send a technician to orbit to swap a bad part.

Both lists are true at once. That's usually the sign you're asking the wrong question.

Because here's what Dyer actually said, and what I haven't been able to put down: whether or not the data centers happen, space has already stopped being a destination and started becoming infrastructure. Not a place we visit. A layer we use — like the internet, like the grid — running quietly under ordinary life.

And you won't see it. That's the entire point.

The same drop in launch cost that made orbital computers thinkable has already put something else overhead: instruments that read the planet itself. Muon's FireSat constellation watches Earth in six bands of infrared, and it doesn't take pretty pictures — it reads heat. A fire five meters across. A gas flare. The warmth a city gives off after dark. Soon, every point on the planet, every twenty minutes.

You will never see that satellite. You'll just get the alert — the evacuation order, the air-quality warning, the reason your flight was grounded — and never know it came from something reading the Earth's heat from orbit. The resource we'll pull down from space won't be compute. It'll be attention.

I come from people who built flying machines everyone called impossible, right up until they weren't. This is the same story, later in the chapter. The argument over data centers will sort itself out — Son or Musk, this decade or next. But the bigger shift already happened, the way these things always do.

Quietly. Overhead. Before most of us thought to look up.