From what I’ve seen it may be their attempts to make server-side grouping work causing extra requests (even when the full set is clientside) which is the bottle neck not the client side processing.
For example, it makes two rest calls upon every expand and dumps all data in collapsed groups from data view
That’s way too much traffic. Nothing to do with client side processing. In fact expand and contract would be snappy if they just load all the data (and keep it loaded) and use kendo’s builtin aggregation.