The real difference for us server folk is that tablet CPUs are always going to be behind the curve in terms of raw power. Likely, that will result in server boards accelerating further in to the 'massively multicore' direction they are going now. Which is okay with me; 1024 chickens, for what I do, actually work better than four strong oxen.
The home market may primarily shift to tablets; I'm not convinced by that and I think there's still a problem with the use case for content creators on tablets which includes an awful lot of school children. But the market I absolutely can't see abandoning PCs as we see them now and moving to handheld touchscreens is the office market and that's far from insignificant.
Tablets cost considerably less than desktop computers. There is nothing preventing you from being able to plug a second monitor and a mouse into a tablet and use it exactly the same way that you use a laptop and a desktop today.
I don't dispute it could happen, but I'd be surprised.
the server market, for all of my career, really, has been following the consumer market. The stuff you see in a server is basically upmarket desktop kit; the primary difference being the addition of ecc in various places.
If ARM beats X86 on the desktop, within a short number of years, they will beat x86 on the server, too.
Now, arm is a /long way/ from winning the desktop market, and I'm not at all sure they will. I'm just saying... /if/ x86 loses the consumer market, they will also lose the cost-sensitive server market.
x86 only owns the server market because the r&d and economies of scale paid for by the consumer market.