One cannot overestimate the precarious nature of ARMs existing situation.
1.) ARM foundries are still unable to produce enough 28nm capacity to satisfy demand.
2.) Opinions of 20nm planar technology have now reached a consensus level that the technology is noncompetitive with FinFET and therefore doesn't have the economics to be viable.
3.) The ARM foundries are therefore left with two equally ugly options. Stick with a doable move to 20nm planar leading to production of a inferior product that leaves Intel a huge opening into the mobile world. Or attempt the almost impossible two process level jump to 14/16nm FinFET that would take three years even if successful.
4.) The other option which appears to be the most reckless due to the extreme cost is to do both at the same time. This appears to be the approach that has been settled upon.
There is almost no chance that this is going to end well for ARM. The stakes are immense and no ARM roadmap appears to have a good chance of success. Only two foundries remain players at this point in time due to the huge costs of building new fabrication plants: Samsung and TSMC. It appears that this will soon be just Samsung. With Samsung competing with its own customers, this leaves no viable ARM foundry choice for companies such as Apple.
ARM has reached the fabrication tipping point and Intel's 2013 roapmap provides them little opportunity to avoid the abyss...