The problem will always be the amount of data to analyze I feel. I would love to see how it handles with rather large datasets in production.
Regardless, I think it's hard to move away from the ad-hoc nature of using R with R Services. Correct me if I'm wrong, but R Services is designed to sort of enterprise R scripts that can be repeatable. Most R scripts are ad-hoc in nature for one-off analysis projects, not enterprise. If you want to enterprise it, you go C++.