I will try to elaborate why there is no perfectly balanced system:
code you write for ps4 is written for fixed hardware which can't be changed. the hardware uses caches and ram to deliver data as fast as possible to the computing units as they are always held back by waiting for data. sony and ms try to reduce this with what they think will do so: sony gddr5-ram, ms move-units or other tricks which have yet to be reveiled. apu also helps, of course.
if you really want to use perfectly balanced hardware then you have to write your algorithms, bring it to hardware and make an asic of it. but this is not happening here and is also only done for special cases.
every component other than the computing units are a bottleneck for computing units. that's why caches grow and ram-speed and latency is also growing.
that does not mean ps4 is not powerful but far away from perfectly balanced and far away of 100% usage. the system itself doesn't allow for that, in every computer.
