As someone whose done both hardware and AI, the current AI investments are at worst a repeat of the 2000 dot com boom.

They aren't wrong, but they may be premature with how terrible our compute substrate is.