Intel is quite unconvincing on a lot of key aspects (10nm, 7nm, performance after spectre family mitigations, etc) but to me the biggest danger is their inability to commit and persist on pretty much any topic which isn't their mainstream CPU line.

They sold Xscale, they sold their modem division (admittedly it was terrible, but I refuse to believe it was unfixable, certainly Apple didn't).

This creates hesitation in prospective customers of anything which isn't their mainstream CPU business, reducing them to a one trick pony. And that pony is now having its lunch eaten from both below (ARM) and above (Epyc).

Their operating profit and their reserves mean little if prices go down - they are used to fat, lofty margins and I'm not sure they still know how to execute without them.

Think about what happens to the shale oil businesses once crude prices approach their extraction costs. You implode very quickly no matter how big you are.