Not extreme? Have you seen his interviews? I guess his wording and delivery are not extreme, but if you really listen to what he's saying, it's kinda nuts.

That Dyson sphere interview should've been a wake up call for the OpenAI faithful.

I understand what GP is saying in the sense that, yes, on an objective scale, what Sam is saying is absolutely and completely nuts... but on a relative scale he's just hyping his startup. Relative to the scale he's at, it’s no worse than the average support tool startup founder claiming they will defeat Salesforce, for example.