Not much of a difference between a so-called PC processor and a server CPU. Mostly marketing speak.
You clearly don't work on server or desktop CPUs.
Increased IO bandwidth, hardware automatic fail over, more cores, support for data-center relevant interfaces, engineered full load, full temperature, 100% duty cycle operation over the lifetime of the device, parity and error correction everywhere inside and out. These are the things that make the server CPUs worthwhile.
I will give you increased bandwidth, it's quite expensive to increase bandwidth and the server parts like Xeon E5 and beyond do have increased bandwidth. As do the extreme edition consumer parts which are essentially Xeon E5s.
Core count is really a profit optimization thing, larger chips are more expensive due to taking up more space. Considering the profits and manufacturing technology Intel has, they could easily make consumer products at current prices with higher core counts and make a profit. How do you like weak AMD?
"Parity and error correction inside and out" ... I'll say you refer to ECC, I really doubt there is any difference in this inside the CPU (Aka, registers, cache and whatever else they need memory bits internally) But I'll give you that the Xeons do support ECC DRAM, but realistically this is a market segmentation issue, supporting ECC on consumer products would be trivial.
The lifetime, temperature and duty cycle stuff sounds like it's from some wank marketing handbook. These are primarily affected by manufacturing (defects) and the QA can at best only guarantee the operation of the chip on delivery. I really expect that the only difference between sever parts and consumer parts in this segment is that the server parts have a larger margin on operation tolerances.