There are already a lot of projects around here focused on improving the enemy AI, and it might be possible that these improvements can be applied to multiplayer.
The problem is that the OP2 AI is not really an AI at all, its basically a set of instructions the enemy follows exactly the same each time. For that reason it wouldn't make a good opponent in multiplayer.
The other problem is that because it is not a real AI, you can't just give it buildings, plop it down in any map, and expect it to function. The OP2 opponents don't actually possess the ability to analyze and react to their surroundings in any significant way. In other words, if you played the AI over multiplayer, you would have to first manually program the behavior of the AI on that map, right down to what buildings it builds where, what units it sends at you when, and exactly how it defends itself. You would have to do this differently for every map you wanted to face it in, and for every type of multiplayer game (Last one Standing, Resource Race, Space Race, etc.).
In other words, it would take a lot of effort to get an enemy that wouldn't be all that great anyway. For that reason, most of the community here focuses on using AI in campaign and colony games where its deterministic behavior is more acceptable.
However, there are attempts being made around here to create an AI that can analyze and respond to general situations rather than following specific instructions. I don't know how far the projects have progressed, but it is possible that you could someday see a multiplayer-worthy AI for OP2.
If you have some programming knowledge, you might even be able to help make it happen.
