There was an interesting article posted here perhaps 9 months ago about CISC vs. RISC computing and the various tradeoffs. Ultimately, in the desktop space, the conclusion was that RISC was a bit better (maybe 2x) - but the development expertise wasn't all there and a 2x gain wasn't enough to drive scale.
In the mobile space, though, a 2x gain means A LOT. A lot more than on the desktop, where you can kind of trade off performance and power. In mobile, battery technology is what it is - you can't just add power. In that environment, if RISC really has an approximate 2x generic performance advantage, allocable to power, computational speed or whatever - then Intel will have to jimmy their platform around substantially to be competitive. Cool if they do it, but it means a really complex design and possibly manufacturing process. We'll see.
There is a certain elegance to using the "natural" solution to a problem. Given how much mobile stuff is already optimized for ARM, Intel is amusingly behind the curve on software - just like they retained a powerful lead in desktop in part for the same reasons.