Back when the PPro was launched there was a low double-digits percentage of the die taken up with the x86 translation hardware (for turning x86 instructions into internal, RISC-like micro-ops). Now that number is some small fraction of a percent. It kills me that I still read this nonsense.
The reason that Intel didn't get into cell phone chips is because margins were (and are) crap, and Intel is addicted to margins. The reason everyone else went ARM is because a) licenses are dirt cheap, and b) everyone knows that Intel is addicted to margins, and they know about all of Intel's dirty tricks and the way Intel leveraged Windows to keep its margins up in the PC space and screw other component vendors, so everyone has seen what happens when you tie yourself to x86 and was like "no thanks".
Of course Intel didn't want to make ARM chips for Apple (or anyone else), because even if they had correctly forecast the volume they'd have no control over the margins because ARM gives them no leverage the way x86 does. If Intel decides it wants to make more money per chip and it starts trying to squeeze Apple, Apple can just bail and go to another ARM vendor. But if Apple went x86 in the iPhone, then Intel could start ratcheting up the unit cost per CPU and all of the other ICs in that phone would have to get cheaper (i.e. shrink their margins) for Apple to keep its own overall unit cost the same (either that or give up its own margin to Intel). Again, this is what Intel did with the PC and how they screwed Nvidia, ATI, etc. -- they just ask for a bigger share of the overall PC unit cost by jacking up their CPU price, and they get it because while you can always pit GPU vendors against each other to see who will take less money, you're stuck with Intel.
(What about AMD? Hahaha... Intel would actually share some of that money back with you via a little kickback scheme called the "Intel Inside" program. So Intel gives up a little margin back to the PC vendor, but they still get to starve their competitors anyway while keeping AMD locked out. So the game was, "jack up the cost of the CPU, all the other PC components take a hit, and then share some of the loot back with the PC vendor in exchange for exclusivity.")
Anyway, the mobile guys had seen this whole movie before, and weren't eager to see the sequel play out in the phone space. So, ARM it was.
The only mystery to me was why Intel never just sucked it up and made a low-margin x86 chip to compete directly with ARM. I assure you there was no technical or ISA-related reason that this didn't happen, because as I said above that is flat-earth nonsense. More likely Intel didn't just didn't want to get into a low-margin business at all (Wall St. would clobber them, and low-margin x86 would cannibalize high-margin x86), and by the time it was clear that they had missed the smartphone boat ARM was so entrenched that there was no clear route to enough volume to make it worthwhile, again, especially given that any phone maker that Intel approaches with a x86 phone chip is going to run the other way because they don't want x86 lock-in.