This has been tried by Cray like 15 years ago and the university of Colorado 11 years ago; there were also other initiatives, all failed.
I'm not excluding that now we have more budget/time/AI to make it work (if it can), I'd just say exercise caution reading this (very AI slop) article (and anything that looks too good to be true)
So also 1000x smaller integration
I'm not excluding that now we have more budget/time/AI to make it work (if it can), I'd just say exercise caution reading this (very AI slop) article (and anything that looks too good to be true)