Isn’t because hype is cheap, and reality requires actual work?
There is no amount of work that can overcome the fact that space is a terrible place to have to overcome heating or cooling issues or to do any kind of maintenance.
This one was always dead in arrival.
This lesson became clear for me when I saw what it took to shield the JWST from the Sun’s radiation. The outermost IR shield was at near room temp, in space. Removing waste heat is a major problem for data centers, which would not be made any easier by miles and miles of insulating vacuum. smh.
Subsea data centers already exist and work way better but they aren’t as flashy.
Once you get below the thermocline temps are very stable at 1-4⁰C usually. Cooling is literally free down there and you can at least haul them up with a tender ship if you really need access (modern hardware is redundant enough you would only do so when a bunch of stuff fails simultaneously)
Edit: I was wrong about Google operating them. Removed it
Subsea data centers already exist and work way better but they aren’t as flashy.
I mean this is also one of those fools errand thing. I think every sailor knows that anything in the ocean is a constant battle to keep it from corroding and getting absolute beat to piss. They make almost as little sense as space based data centers.
The issue with data centers isn’t their existence per se, at the end of the day they’re basically a high energy consumption warehouse, not that dissimilar than a large refrigerated storage facility. The problem with data centers is the politics they represent, how they are being rolled out, what they are being rolled out for, and who they’re being built to benefit.
But at the end of the day, building something that requires regular upkeep, access, maintenance, cooling, etc… it will never make sense to do those things in a place which is more expensive than a place in which it is less expensive to do those things. You don’t need to take even one mental step further into a prospective approach if it doesn’t make the process more affordable in some manner.
The one benefit is they’re below the wave action layer at the sea bottom where it’s cold and stable. Infrastructure like this isn’t new due to offshore drilling for oil and gas processing providing a lot of the technological base and logistics, so making high reliability hands-off infrastructure hundreds of meters below the surface is mostly well understood.
A normal data center is actually pretty low maintenance, which is why they might employ a few dozen people full time, and that’s mostly as Button pushers or cable pluggers. You can design them to be remotely administrated especially for cloud computing or web hosting. The hardware doesn’t change just the tenants and their software.
Apparently it made enough financial sense to do these prior to the AI boom when data center builds were more calculated risks that needed enough traffic to justify them. Deciding to put them in space is just pure AI bubble hype though… It’s not nearly as sexy to say “we’re building AI data centers at the bottom of the ocean so they can be close to subsea fiber optic cables and get free cooling!”
Apparently it made enough financial sense to do these prior to the AI boom when data center builds were more calculated risks that needed enough traffic to justify them. Deciding to put them in space is just pure AI bubble hype though… It’s not nearly as sexy to say “we’re building AI data centers at the bottom of the ocean so they can be close to subsea fiber optic cables and get free cooling!”
My understanding of undersea datacenters were that they were high visibility pilots which never resulted in any materially significant number of installation. Not any different than whats going on with the space borne hype. Them being below wave action isn’t relevant. They could be at the bottom of a lake for all that matters. The at-sea datacenters are just as much goofy hype as the spaceborne ones.
