How would you like 4 teratransistors on a chip? https://spectrum.ieee.org/cerebras-chip-cs3

Learn how to operate your own networks. Meanwhile, use anonymizing tools religiously.

Look at the enormous processing resources of biological brains. Human brain is 2% of body mass but 20% of baseline metabolism – this is very expensive evolutionary. Neural hardware used for LLMs or just any scientific numerics accelerator is just a bad reinvention. Your argument reminds me of Minsky’s “5 MIPS is enough for AI”. Nope. You have to track a lot of state, its relationships and refresh it all very quickly. Computation is expensive.

Your points are adressed by Ed Zitron.

The AI hardware is dead after 3-5 years.

AI will remain a massively parallel numerics affair with enormous data sets and monstrous memory bandwidth and network crossection. And according energy consumption. Jevon’s paradox will eat any efficiency improvements.

The node sizes are backwards computed from the unit area of a one bit SRAM cell. It used to be the physical size of the smallest structure, e.g. a transistor gate.

I agree it’s highly misleading. Also, people forget that the original Moore’s law publication mention constant doubling time of transistors per area per unit of cost. So people keep pointing at multichip, thinned die stacking and 5 nm WSI like Cerebras to argue Moore scaling didn’t die a long time ago, but ignore costs.

I’m not buying shit unless I have to. And then I’ll likely buy used.

Meh. You can’t find tech talent in niches. E.g. senior K8s skills are hard to find.

Go back to the Ukraine. I hear fans of Bandera and Shukhevych need warm bodies to fertilize the rich chernosem, so go join https://en.wikipedia.org/wiki/International_Legion_(Ukraine)

Is the Pope catholic?

Ukraine has been too busy killing their own citizens since 2014 to now stop being a NATO proxy against Russia. Demographically, they are already dead.

I could use some of them. But the power bill will make my eyes bleed.

You can buy a used Pixel. Motorola has some useful hardware, but this is about freedom, not best price performance.

GrapheneOS, too.

https://en.wikipedia.org/wiki/The_Machine_Stops

All I need is a baseline open hardware EV. Fat chance, of course. So I guess I have to buy something used, today older than 8 years and counting.

Big companies license Copilot for less than 25 usd/month per seat. Don’t tell me it covers the ops cost, even for mixed calc.

I see you have no clue. You will learn, eventually.

The EU is busily building the Fourth Reich, so don’t expect help from there.