Large scale is the way to make things economic, for sure. Leakage of atmosphere per unit volume is better for a start, and the problems of slosh and of asymetries of rotation causing varying centripetal force, vibration and cascading failure are much less dangerous at scale.
The only asteroid materials we're going to be refining in bulk any time soon are nickel, iron and water, but it so happens they're what is needed for this :)
Radiation environment sucks for humans, even for robots, but sufficient bulk between you and the outside will cut it down.
My only real concern with O'Neil Cylinders is how they will pay for their inputs.
Manufacturing?
Mining colonies on some moon can sell raw materials, even if the logistics chain is a decade long.
But manufacturers in free space are in competition with Earth (for the high-end stuff), and with the mining colonies (who will want to fab some least-complex stuff themselves).
