"expected" payoff is the key, here. He's not just dividing a billion by 3.
His math is very suspect in the first place, because 1 in 2 million is roughly what you'd get if you divide the number of billion-dollar exits by the population of the U.S. That's an odd way to slice it; if you want something more representative of an individual person's odds, you're better off dividing the number of billion dollar exits by the number of companies started that believe they have a shot at a billion dollars (which might be in the tens to hundreds of thousands, but certainly not 2 million), and then you end up with figures roughly equivalent to his AEPs, but that ignores that many of those founders are college or grad students and the outcome is getting that job at Google after all.
Even if my excuses for him are correct, I think you all have made a valid point that this is not a really productive way to measure :) Nonetheless, the lesson "don't start a company to get rich" is, I think, a good thing to say.
He says:
"Even if you do raise money and sell a company or take it public"
Given that statement "sell a company or take it public" it would appear that we are definitely talking about people that have gotten to a particular point (where the group narrows quite a bit). Not the chance of getting to that point after starting a company.
For sure the probability is low for getting to the point of taking your company public (or selling) but once you take your company public in no way could he be arguing that the economic benefit is only $300k. Something doesn't make sense here.