29
AI data centres — in SPACE! Why DCs in space can’t work
(pivot-to-ai.com)
Big brain tech dude got yet another clueless take over at HackerNews etc? Here's the place to vent. Orange site, VC foolishness, all welcome.
This is not debate club. Unless it’s amusing debate.
For actually-good tech, you want our NotAwfulTech community
@burble @techtakes
Why not submerge them in a river for far less money.
Microsoft had an underwater datacenter-in-a-container test a while back.
"How do we cool these things?"
"idk, boil the oceans?"
I'm surprised lagering caves, except for servers instead of beer hasn't come up yet.
He touches on that in the blog post, and gives us this link: https://www.itpro.com/infrastructure/data-centres/microsoft-scrapped-its-project-natick-underwater-data-center-trial-heres-why-it-was-never-going-to-work
Didn't Meta, then Facebook, open a datacenter somewhere in the Arctic circle and leave it exposed to the environmental cold? Until the planet runs out of cold, an eventuality whose impending rapidity I'd rather not ponder, that seems a viable approach.
Underwater datacenters make cooling very effective and maintenance nearly impossible, so you have to treat the container data centers essentially disposable. That's only viable with economy of scale big enough to be an xkcd comic punchline. I guess Microsoft found that even they are not quite there yet. Also most computers don't tolerate seawater quite as well as they tolerate air.
@bitofhope
Pretty sure space datacenters are also going to be even more disposable.
Oh yea absolutely. Underwater datacenters have one upside (cooling) and massive downsides (everything else, more or less). Space datacenters trade that upside into yet another downside, make the downsides even bigger and add a few extra downsides for good measure.
imagine a $100b GPU data centre with the disposability of a cubesat