This is not totally crazy. A problem would be you then need a lift to carry everything to the top floor and an expensive structure to support it. You also have

This is not totally crazy. A problem would be you then need a lift to carry everything to the top floor and an expensive structure to support it. You also have local wind and weather conditions to deal with at altitude. Alternatively, you could use a ground-based slingshot to vertically propel gliders, an approach that is already used in the military and for some long distance UAV testing. However, the wing-based UAV vs. rotor-based UAV physics are very different. Landing an elevated glider with a clear path to its target is one thing, maneuvering between trees and power lines in a complex environment without rotors is quite another. Then you need to look at a 'convertible' UAV with both features... undoubtedly this raises the cost, reduces the carrying capacity, and is already being pursued. Finally, you could zoom out and re-evaluate the business case. Why not use a network of local dispatch points (handover points to ground-based UAVs or vans) to reduce the distance to the consumer and with the smaller geographic scope negotiate private airspace access on locally agreeable terms. I believe this is the future for general consumption (food, regular deliveries within small package weight/dimension limits).