The best I can guess is that the author is mixing up load balancing through reserve energy and efficiency losses. The amount that is reserved for load balancing is also not anywhere near the size they are talking about.