As reality capture technologies have advanced in AEC, users have been able to generate increasingly detailed and accurate surveys, each image formed from millions of data points. However, this depth of information also raises some information challenges, particularly if site-based personnel want to access and interact with the data. In this Viewpoint article, Steve Salmon of Pointfuse discusses the importance of intelligent meshing technology in a much more efficient representation of point clouds, benefiting the construction as well as subsequent operation and maintenance (O&M) of buildings and infrastructure.
AECbytes Blog 
Well, yeah. Nobody can do much with a raw point cloud. At best, it is something to guide the production of something else. If polygons can serve as that basis more nimbly, I am very interested to learn more.
It is already possible to take classified binary Lidar point cloud files straight into Revit topography. The level of detail of such terrain is really amazing, it is pretty much like a land survey. But Lidar data does not take days like a land survey, it is available in many cases for free in the US and European countries. This is really a must to work with precise topography early in the project. We can just anticipate that in the future, better scanner and processing possibilities will pave the way for bringing even more elements and detail into 3D models.