Agree, with one exception: pg isn't arguing about "success", he's arguing that single founder companies (which he further qualifies to mean "and that have no early employees") will not, statistically speaking, grow into large companies.
Of course, pg's stacking the deck by disallowing early employees for solo founders (which he does by claiming that early employees are "founders", and thus, support his multi-founder hypothesis).
I do think it's correct to say that companies without employees are rarely very large, but this doesn't seem all that insightful.
The more interesting question is why pg disallows solo founders with early employees? The vast majority of large, successful companies actually fall into that camp, and not the multi-founder camp, of which Google is a large, successful example.
To pg, a "founder" is apparently an early employee who (a) wants the company to succeed, (b) plays some kind of "important" role (e.g. is not a janitor), and (c) has some skin in the game to benefit from "success".
I think that's far too broad, and that we should call them early employees with equity (i.e. what they are), not "founders" (what pg apparently wishes them to be).