Since it wasn't talking about cost, it makes no assumption that anything is "free". As for perfection, I was assuming the reader could multiply 400 by whatever number < 1 was the actual efficiency.
Unstated was whether the difficulty of running breeder reactors makes this unworkable. It very well might, in practice.
I will note that the demonstrated U collection techniques (like polyamidoxime adsorbers) seem quite cheap enough for fueling breeder reactors.
