It is bad design if their upgrade programs brick their devices against all recovery methods. I've worked for several companies as a firmware engineer and our products were not brickable. The bootloader is not customer flashable, everything else is. Several of my products were RAID controllers in competition with LSI. I would suggest shouting far and wide that a consumer upgrade procedure was capable of bricking their device. Even the small computers which use uboot and like-bootloaders can be field recovered using an inexpensive (under $50) bring-up device if the manufacturers so choose. Designs like graphics controllers should have a recovery method using a PC app which placed the executable into RAM and then the user forces the graphics card into a special boot mode which looks for the executable at a pre-defined address.
I've never worked on a PC motherboard main-CPU design but I would think that a good BIOS design should have an A-image and a B-image where you can not bootload the A-image from itself and vice-versa.