I'm starting to think that some companies just chase the most expensive way to do something possible knowing there will be some niche demand for it that will pay a lot. Because it is so expensive you can then have a lot of room to start chipping away at costs.
But that Boltzmann equation isn't that good a news. It means you want to run some of the system so it is white hot. Whether you use heat pumps to concentrate that or not, the hotter you can run the source side the more efficient it will be. That means you want silicon that can run hot. That means you'd want a silicon carbide transistor. Lol.. building a silicon carbide graphics card or ASIC does sound expensive. But Tesla would be at an advantage because they use silicon carbide semis in their cars. But that's typically used for power electronics instead of logic.
It really is trying to find the most expensive chore. But I guess with AI existing you have to chase the impossible, because if it is easily possible a clanker will be doing it with zero human support soon. And then where as a human do you fit in?
Very interesging idea. I've listened to most of it. Seems the one could build 100x less expensive data centers in Utqiaġvik (formerly Barrow) Alaska with natural cooling most of the year, with 100x better security for it.
I like the idea of industrializing space* - but intelligently. That don't seem smrt.
(* My Bittersweed Seeds screenplay gets into a more practical, cheaper way to colonize space. But it's still a huge project.)
Also, any electronics in space are one solar flare, super nova, or atmospheric nuke EMP away from getting wiped out.
If I was the ruling class I'd be building them under mountains near water dams to cool and generate at the same time. Or on a thorium powered ocean liners or subs.
Also, I suspect it might be more efficient to copy a bunch of hard drives to distribute to other centers physically rather than networks. I'd imagine this physical swapping could be a full time sub-system.
Log in
I'm starting to think that some companies just chase the most expensive way to do something possible knowing there will be some niche demand for it that will pay a lot. Because it is so expensive you can then have a lot of room to start chipping away at costs.
But that Boltzmann equation isn't that good a news. It means you want to run some of the system so it is white hot. Whether you use heat pumps to concentrate that or not, the hotter you can run the source side the more efficient it will be. That means you want silicon that can run hot. That means you'd want a silicon carbide transistor. Lol.. building a silicon carbide graphics card or ASIC does sound expensive. But Tesla would be at an advantage because they use silicon carbide semis in their cars. But that's typically used for power electronics instead of logic.
It really is trying to find the most expensive chore. But I guess with AI existing you have to chase the impossible, because if it is easily possible a clanker will be doing it with zero human support soon. And then where as a human do you fit in?
Very interesging idea. I've listened to most of it. Seems the one could build 100x less expensive data centers in Utqiaġvik (formerly Barrow) Alaska with natural cooling most of the year, with 100x better security for it.
Yes.
I like the idea of industrializing space* - but intelligently. That don't seem smrt.
(* My Bittersweed Seeds screenplay gets into a more practical, cheaper way to colonize space. But it's still a huge project.)
Also, any electronics in space are one solar flare, super nova, or atmospheric nuke EMP away from getting wiped out.
If I was the ruling class I'd be building them under mountains near water dams to cool and generate at the same time. Or on a thorium powered ocean liners or subs.
Also, I suspect it might be more efficient to copy a bunch of hard drives to distribute to other centers physically rather than networks. I'd imagine this physical swapping could be a full time sub-system.