Lower watts/kg needed for a given speed because more total watts produced. grams already explained it. Not many folk produce >5 W/kg for extended periods. Lots of people can do 3 W/kg.
That's assuming that bigger automatically means more power - enough to neutralise the weight/aero disadvantage and then some to spare. Sadly that isn't true. I'll agree that a bigger frame on which to attach muscle and lever them provides the potential for more power - but it's not automatic. I'm the same size and weight as Andre Greipel but can't manage even half his peak power. I wish.
I know a superbly fit 100kg Zwifter, every inch of his body is finely honed muscle - and on anything beyond a 100 metre sprint my 85kg flab wins because all I have is leg muscle, not the extra 15kg of huge arms, abs and pecs.
Size a rider can't do much about apart from good aero, but weight they can. For any rider, reduce weight equals go faster. Yes, there are limits. In G's last book his discusses the difference between his track weight and road weight. He also mentions discovering his weight limit, the point at which lossing any more weight causes a power drop - presumably the point at which he's lost everything else that isn't muscle.
The working of size and weight can at least be modelled. Doing the same for the human body and power is much harder, much less predictable and absolutely not guaranteed.
ps. the average weight of the UCI pro peleton in 2017 was 70kg, but I guess they don't know anything.