I don't quite agree with this. Intel definitely did focus on "architecture innovation". It's just that their whole philosophy with P4 was to pipeline the shit out of it so they could reach insane GHz numbers, and to mitigate the downsides as much as possible with a super-advanced branch prediction unit.
I think I simplified it too much . I certainly didn't mean to infer that Intel sat on their laurels and just tried to jack up the overall clock speed. It may be more worthwhile to say that they "bet on the wrong horse." I also referred to the A64 X2, where you can argue that the changes started in the A64 with its on-die memory controller, which was eventually used (in their own implementation of course) by Intel as well.
Although, it seems where this discussion stemmed from was the belief that my whole "mobile sector is taking off" statement means that it is going to run into problems like Intel did with their "one-trick pony." I don't believe this is the case at all. Although, the player that I'd like to keep an eye on right now is nVidia. The thing is... they kind of got slapped in the face with the iPad 2 coming out. I recall seeing how poor the GeForce ULV was in Tegra 2 when it barely beat Samsung's Hummingbird in some tests (if I remember correctly), so I had a feeling that it wouldn't be doing too hot against some new contenders.
So with that... what are they going to do next? They had the benefit of essentially "innovating" the desktop GPU market back with their original GeForce, but right now they've got some good competition with PowerVR and Adreno.
Just looking at Wikipedia, there are still two more GPUs similar to the one in the iPad 2 (544 and 554) and then there's the series 6 that will eventually be coming (in time for the iPad 3 perhaps?):
http://en.wikipedia.org/wiki/PowerVR#Series_6
As I've mentioned, the mobile sector is pretty exciting for me right now. While I may not have the best understanding of it all, I do rather enjoy technology!