For pure horsepower and balls out performance, Intel is the way to go. But as mentioned, for entry to mid-level and OEM systems, AMD will be the platform of choice.
Actually, with the way the AMD and Intel product offerings and performance differences are today, it reminds me of the way AMD and Intel were before the Athlon64. Back then, AMD firmly held the entry and mid-level offerings and were easily identified as the an underdog favorite offering "close-but-no-cigar" performance comapred to Intel but also at a cheaper price. For the budget enthusiast and comparing the cost of an original Athlon compared to a P3, the Athlon was the way to go. Aside from the processor, Intel compatible mobos were also far more expensive than AMD compatible mobo; but that was also when VIA was the AMD chipset of choice.
Given that Nehalem is going to require a new chipset and socket, while it may beat the pants off 45nm AMD quad-cores, there are many folks with AM2+ mobos who (much like the Skt939 and Skt775) who can just drop in a new chip and update the BIOS to increase their cpu power.
At best, the B3 steppings will (hopefully) redeem AMD and get them back on track to producing issue free products. And, hopefully the new CIO and Marketing guy can and will do a better job at dealing with any issues that may arise.
It's been noted that Intel skipped creating a monolithic quad-core at 65nm because of (possibly) the same issues that AMD has experienced. But there are two sides to the 45nm monolithic quad-core coin. On one hand AMD has had extensive experience with and IMC, monolithic design, and now implementing an L3 cache. Whereas Intel has extensive experience producing high yields of 45nm chips. However, it would be niave to think that Intel will not suffer their own "growing pains" implementing 45nm, monolithic quad-core, and IMC, as well as an L3 cache. So with that said, the fanbooys can claim Intel this or AMD that, the fact remains that we won't know for sure until the chips are available for review and testing.