[jaggs]

Two reasons?

1. Management 2. Scalability

Running your own local AI takes time, expertise and commitment. Right now the ROI is probably not strong enough to warrant the effort.

Couple this to the fact that it's not clear how much local compute power you need, and it's easy to see why companies are hesitating?

Interestingly enough, there are definitely a number of sectors using local AI with gusto. The financial sector comes to mind.

[briansun]

Well put. Management overhead + unclear capacity planning kills many pilots.