1 article tagged with this topic
A Reddit proposal to run huge local models on 16–32 SSDs says more about hardware cost pressure than a new path forward.
