I'm not bold enough to say he's wrong, but I don't think it's much of a reach to suggest he's failed to account a truly monumental level of sampling bias in his 'research'. He hears a lot about Go being the future because he hangs around with lots of Go users. As he says himself, his company probably has the most information about Go users outside of Google or Github - but he doesn't have information about trends in other languages and that's what you need in order to predict that Go is going to be big.