My family has had a cattle ranch in Colombia for over 25 years. We experimented with intensive rotational grazing and found that it works but relies heavily on precise rotations. Those rotations can eventually be automated with computer vision and robotics.
If you can build this autonomous system such that it is quick to install and uninstall, then you can take advantage of small, empty plots of land. The supply set of land is then much larger and you can both minimize land cost per animal as well as optimize land/animal combinations.
With AVs and decreasing transportation costs, you can optimize an owned animal population over a large network of small rented plots. My hypothesis is that this can be scaled up to replace feedlot operations at lower costs to the consumer.