The PDP-11 is (mostly) little endian, and pre-dates all of Intel's processors. Intel's endianness comes from the Datapoint 2200, which was bit serial, and the carry circuitry is simpler if you do things little-endian.
Honestly, I think the real answer is that big endian is easier to diagram if you come from a culture with a left-to-right writing system and a habit of putting zero on the left on graph axis.
(I like little endian systems)