I've heard this before. I think it's definitely possible that Spotify's revshare works out in favor of big artists, but I haven't heard an argument for that that didn't involve making broad, unsupported assumptions about users' listening habits.