What dies is the motherboard, not the CPU. When the motherboard dies, the CPU is so outdated it doesn't even make sense to keep it.
This is aimed at companies like Facebook, Google, Amazon, and the like. When managing thousands of servers, any number of components will die on a fairly regular basis. Some will die withing a few weeks of them going online. When you have 200k servers and a component with a 1% infant mortality rate, having the ability to quickly and easily change the component is a blessing. You can do this with just about all of the components in a server accept the processor (relatively speaking).
As for why would you want the ability to switch between architectures as the drop of a hat. It is actually really simple. What is the cheapest that day? If the processor in a server goes and you have to find the exact processor that is compatible with that motherboard, you cannot use the cheapest component. Most likely you will have to go with something that is more expensive. With this design, you can simply go to the bin, grab any processor card, and slap it into the server. You might have to change a flag in your server management software saying that node should now network boot the ARM image rather than the x86 image, but that is about it.
There is very little future in being right when your boss is wrong.