I have never heard the line "most of your companies will fail" from anyone I know of in Y Combinator. Mostly because "fail" isn't really well defined -- even a mediocre startup is usually more interesting than a boring job. Even founding the next Instagram might be a failure for someone who really cares about medicine or space.