There was a whole Byte magazine dedicated to it back in the day. It was thought (at the time) that the only way we'd ever break the 1 GHz barrier was to use GaAs, which obviously turned out to be wrong.