Judging from his response to the desktop segment question, I'm a bit concerned. I think most people are very happy with the llano and want to see future fusion tech do well, but their bulldozer flop and yield issues at GloFo can potentially be serious issues for their desktop chips and future desktop chips. If they're going to stand by the bulldozer architecture, i think it'll be a few years and multiple revisions before we finally see a decent chip, and that would mainly have to do with the software industry embracing multithreaded optimization and not the actual performance of the chips.
I think focusing on the mobile sector is fine, but the #1 priority, even above performance, is low power consumption. Unfortunately, Intel isn't having any issues at all producing chips that consume less power and still perform respectably. They're about to start producing 22nm chips before the end of the year (Ivy) and don't appear to be having any unforeseen issues.
I know I may sound like a heretic, but AMD should really reconsider and perhaps go the ARM route. I don't think dropping x86 chip power consumption to ARM levels will ever be possible. Opting to do so can put them ahead of Intel in the mobile market while still pumping out fusion for lappies. Maybe they can then pay a little more attention to us desktop folk as well