not necesserily. depends entirely on the algoritms and approaches employed by the devs and artists. if we assume there's a set of 'sweet spot' visualisation techniques that yield good visuals at max rate, then i would venture to guess that the work-load under such scenarios would be mainly on the GPU.
Thanks I thought the work on the CPU would need to scale equally too, are those algoritms and approaches more at the level of shading and such (and less vertex creation)?
Althought I would prefer strong CPU and weaK GPU as most of next gen feature that I like seems to come from the CPU (dinamic AI, dinamic animation, physics...).
of course if you write a sw raytracer for a wii title things may be a bit tough : )
I would be happy with fast raycasting for AI .