The biggest problem (aka challenge) is capital.
In order to continue to drive the technology nodes smaller and smaller you need tons of money.
Billions of dollars and a huge market you can serve. 
5 billion or maybe 10 or one day even 20 billion dollars will need to be put on the table. 
The game is open. 

A chip manufactured in 5 nm will need a huge audience. Otherwise you simply can’t afford to build a new chip or even a new version (aka revision) of it.

Intel knows why they are investing in FPGA start-ups. 
Expect future Intel chips to be more flexible.
What will count is that you produce as many chips of one and the same device (respectively design) as you possible can.

Intel x86 in mobile devices? 
According to sources who have tested the first x86 driven smartphone ARM and x86 are sort of on par. 

That is as of April 2012. 
In several years the picture may well look different. 
x86 for mobile applications may take the lead since Intel may get themselves into a position where they can manufacture devices in incredible volume but still offer some flexibility – which spells: programmable hardware. One device will cover several applications (which essentially was the original idea of the Intel microprocessor anyway).

The smaller the process technology nodes become the more Intel will take the lead.

I expect Apple to move to x86 for iOS. Not today. Not tomorrow. But the day may well come.

My two cents …