Anecdotally-speaking, for a general-purpose Dell/HP/IBM server in a consistently-cooled/powered environment, you'll either have something fail right away within the first day/week like RAM (usually right away), PSU(s), or hard drives perhaps within a few months, followed more rarely by RAID controller/motherboards several years after deployment (blown caps; was an issue several years ago with some boards manufactured by one particular Taiwanese firm IIRC).
Again, I'm grossly generalizing here, but the stuff that goes wrong during manufacture/assembly will usually crop up shortly after deployment. For example, harddrives, well, I haven't had a drive from a Dell server fail on me in over two years (over a dozen servers of various size); had bad RAM shipped twice in that time, but the instability was picked up right away and threw an error on the LCD diagnostics. Had a bad PSU once that I can remember, but that server closet room was an embarrassment with faulty grounds and old dead UPSes.
Having said that, I have many many old servers (Pentium II/III vintage) with SCSI drives that still POST and run Windows 2000/Linux stable.