World Canvases. One with many panels at different world coordinates, or many canvases?

Which is more efficient? I’ve recently loaded up one of Unity’s Sample projects and saw that they had a single Canvas with world space rendering, but many different panels at different world coordinates.

For dynamic spawning, I’ve typically made the prefab at the canvas level.
This style of, perhaps optimization, would suggest that the panels should be made into prefabs and to exert world transform manipulation while adding it into a singleton canvas.

Is there any known guidance on what would be the preferred approach as far as optimization?

If the canvases have to register themselves with a render pass individually, that could reduce the amount of draw calls necessary to handle the UI.

So just conceptually, I think panels would be more performant.