I personally wouldn't waste my time developing multiplayer game only, there is little chance you make money at all. Look into top 100, or probably even top 250 and try to find at least one game which is multi only, will be very hard. iOS games are for quick fix when you have few minutes to spare. So no matter how hard is to make AI I would concentrate on that first and multi player later.
This is what we decided to do with our game. Initially I designed it to be only a multiplayer game, then we decided to put AI into the mix so players could play against AI, then we put a Story/Campaign Mode in as well to add more content for those that just wanted a quick fix. In the end, we now have a game that will hopefully have a mode for everybody! We are releasing the game without the online multiplayer initially and will develop that in the following weeks. Once we release the multiplayer, we will likely be releasing a free version of the game with just a limited online multiplayer and a tutorial in it. That way we will be populating the game with a constant stream of players so nobody has to scrounge around for opponents.
Unless the online play is asynchronous then definitely get AI in there from day 1, and add the online multiplayer in an update. I made the mistake of not having any AI in v1.0 of my turn-based war-game as I thought Game Center head-to-head would be more popular than it is so got the online multiplayer working first. In v1.1 I added the much needed AI but in hindsight I should have definitely done it the other way round. The vast majority of players want single player games - I had local multiplayer pass-n-play from day 1 but only a small percentage actually want this, especially on iPhone. AI is essential, especially with a board game (unless as I say you have your own servers hosting asynchronous games).
