How GeometryReader3D acquires depth

I had a question while researching the layout of spatial views in visionOS. As I understand it, in a 2D plan layout GeometryReader can take the suggested dimensions provided by the parent view. In 3D spatial Layout, GeometryReader3D can obtain additional depth value. However, in layout's SizeThatFits function, ProposedViewSize passed by parent view to child view does not contain depth information. How does GeometryReader3D access this depth information?