I have a question that most likely corporate relations will not answer, but if they have personal opinions, I think we are all interested.
I think we all aggree on that AMD has a severe lag in CPU performance compared to the competition, while in terms of graphics power it's clearly just the opposite. Since AMD has already server products that feature multiple APU chips, what's keeping AMD from creating similar mobile boards? Synchronization between the two entities on the CPU side is a solved problem. The two ways to go about the IGP is either create a cross-fire like connection between the two using HyperTransport, or have them aggregate their computing power in a unified way (similar to the CPU). This latter is clearly trickier, yet much more rewarding.
This way AMD could create true 8 core mobile systems, that could cut short the lag in CPU power, and even the most powerhungry mobile APUs don't exceed 45 Watts, which is far from impossible to dissipate in a notebook, even when doubled. Just a plain HD8970M consumes more power. The power of two IGPs added together could satisfy the needs of any hardcore gamer on the go.
In my opinion the idea is so simple and so great that surely this has passed the minds of all the people in the industry, and as far as research goes, I would think AMD has put the most thought into multi-socket APU boards. Is the idea a dead-end? Or are there just too many obstacles on the way? Or simply the time has not come yet to synchronise APUs in this manner with all the quick evolution of GPU architectures? What's keeping mobile products with dual-socket solutions to arise?