[CamperBob2]

Why do they need to be built in populated areas at all?

[yellowapple]

Shorter employee commutes? Shorter last-mile shipping distances? Lower latency to/from local customers? Closer proximity to points of intersection of fiber backbones? Closer proximity to existing electrical/water/sewer infrastructure?

[CamperBob2]

Shorter employee commutes? Shorter last-mile shipping distances? Lower latency to/from local customers? Closer proximity to points of intersection of fiber backbones? Closer proximity to existing electrical/water/sewer infrastructure?

But we're being sold a vision of putting them in low-earth orbit. That means, among other things:

- They don't need to be situated anywhere near their customers

- They don't need a lot of employees to babysit the hardware, or in fact any at all

- They don't need water. Radiative cooling is evidently just fine by itself, even without convection or conduction

- They don't need any networking infrastructure beyond what satellite IP links can provide

- They don't need anything but localized photovoltaic power

Every argument for putting data centers in space applies equally to putting them literally anywhere on Earth.

[yellowapple]

Orbital datacenters are still hypothetical, at best.

In any case:

>They don't need to be situated anywhere near their customers

They are situated “near” their customers… assuming those customers are also Starlink customers. That's really the only remotely-decent reason (IMO) to put datacenters in orbit: to fulfill the same role for space-based customers (including satellite Internet users) that “edge” datacenters do for geographically-local customers.

[CamperBob2]

Orbital datacenters are still hypothetical, at best.

We're about to find out just how hypothetical they are, since the valuation in the upcoming SpaceX IPO is heavily tied to that particular pipe dream.

[saghm]

Cynically, so that tech companies can get their power usage subsidized by the local population through their higher energy bills.

[yellowapple]

That's easy enough to fix by charging datacenters at a higher rate than residential customers. Most electrical utility companies already have separate residential/industrial/commercial rates, specifically to prevent large-scale consumers from spiking small-scale consumers' prices.

Here in Nevada, NV Energy's in the process of getting state PUC approval for datacenter-specific “large-load electrical service agreements” specifically to ensure datacenters foot the bill for the infrastructure and generation buildouts needed to support them. Hopefully it goes through, since that seems to me like the exact right way to go about it.