One doesn't have anything to do with the other. This is all about presentation layer and where the data is rasterized. Google's database isn't made up of millions of bitmaps, that's just what's served to the presentation layer so the app/browser doesn't have to perform the rendering.

Just in case: I'm talking about maps and map data used to display the streets and calculate routing, not pretty photographs of trees and buildings, including, but not limited to satellite imagery.

The point I was making is that having the source data on the device and the ability to rasterize within the app gives you the ability to zoom and obtain more detail without hitting the net. A potential sacrifice in CPU/GPU performance for the rendering, but a savings in network access. Should come out ahead since bitmaps take up a lot of bandwidth and the platforms usually have CPU/GPU to spare.