I think that 60% is fishy too. But I'm pretty sure there is still alot of power to squeeze out of the PS3 SPEs: the more parallel the architecture, the harder it is to fully exploit it.
On MS' side I do agree that probably we're going to see a shift from meat-and-potato PC-like development to a more dirty hands-on approach with the hardware.
Up to this point the x360 has sailed quite easily on having an architecture more similar to the PC and a better SDK, so that for example in multiplatform titles we've seen MS AA on the x360 vs worse AA methods (quin. supersampling) on the PS3 where MS AA has to be implemented by the devs, not sure if via pixel shaders or SPEs. Another example is the fact that ambient occlusion has still to land into the standard Sony SDK, whereas multiplatform games have been using it yet on the x360.
Microsoft may have fumbled alot of projects but they know how to cater to developers and their excellent toolchain gave them a good headstart in multiplatform titles territory. As Sony's SDK catches up and their first and second party titles tap deeper into the hardware strong points, so MS will have to "change gear" and specialize their software technology to keep up.
Good times ahead :)
edit: darn, priteshmodi stole most of my post... oh, well :)
