This would make sense as it's based on Google Earth which is 3D.
Even WebGL web version of Google Maps is already 3D since some time ago. See e.g. Colosseum here:
https://www.google.com/maps/place/Colosseum/@41.8950247,12.4...
From what I saw in demo, it looks like they are generating simple 3D models from the photos. Which is pretty damn cool!
Here is an interesting quora on where Google Earth sources their models from: https://www.quora.com/How-are-the-3D-models-of-buildings-gen...
This would make sense as it's based on Google Earth which is 3D.
Even WebGL web version of Google Maps is already 3D since some time ago. See e.g. Colosseum here:
https://www.google.com/maps/place/Colosseum/@41.8950247,12.4...