x86 has become so complicated and difficult to design and verify that even a company as big as Intel uses the same processor core for their server, desktop and notebook chips, despite the different market needs. I think simplicity is a strong argument in favor of ARM, regardless of what the power consumption and silicon area savings might be. If it was practical to design separate processor cores for each of these markets, the server processor could have 4 threads per core and a bigger TLB with more entries for 1 GB pages, the desktop processor could have a higher clock rate and the notebook processor could have 2 instruction/clock superscalar instead of 4 instruction/clock superscalar (to save power). I think the server and tablet markets are the first places ARM can displace some x86 because a processor core more tailored to the needs of these markets could have more value to customers than the value provided by x86 compatibility.

George Baker on 1/4/12 wrote:
> All of these could be applied to an ARM server chip. In an ideal
> world, AMD and Nvidia would work together on such a chip and
> would be second sources for each other on the finished product.

I agree than AMD and Nvidia would both benefit if they worked together on an ARM server chip. They could work together on the second-generation ARM server chip, since Nvidia's first-generation chip is probably almost finished.