Currently I've been flying in X-plane 11 with Ortho4xp and downloaded data can easily run into the hundreds of GBs and even multi terabytes.
We took some open city point-cloud data last week and created 3d tiles from that using Cesium, and I was amazed at the result. Nothing like what you're seeing here, but I imagine an artist or possibly AI could do enough correction to make the cities look amazing and realistic enough.
Because you aren't looking at the city like a map, they don't need Google Earth level of accuracy.
Though another possibility is that Google Earth is licensing their data for games, so possibly Microsoft has licensed that?
One of Bing's particular focuses contrasting them to Google Earth/Maps photography has been a particular interest in "Birds Eye" isometric-feeling photography taken from helicopters at particular altitudes and compass angles, I would imagine is particularly applicable to trying to generate rich models for Flight Simulator use at similar altitudes, more so than a lot of traditional "flat" overhead aerial photography used by map providers.
(It's gotten harder to find Bird's Eye in Bing Maps redesigns sadly. It's "hidden" in the right-click menu now. Still, it is a useful and interesting type of aerial photographer that should be celebrated as useful and interesting.)
However, when I now see Birds Eye, it isn't available in Sydney, and many other locations in the world. Where I did see it (San Francisco & Paris) it is just the photographs, no photogrametry, no rotation, very little zoom, etc. etc.
Am I missing something? Can you let me know what you're seeing?