I found it surprising that we don't specify the value category here but say:

has type void

http://eel.is/c++draft/basic.lval#1.2 generally makes any expression of type void a prvalue, so this change is editorial only.

I believe it is a slight improvement because it reduces the reader's dependency on that general rule and makes the wording more conventional. We have the tendency to say "prvalue of type void" almost anywhere else, such as in https://eel.is/c++draft/expr.type.conv#2. For better or for worse, that's why I didn't even know about the blanket rule for void up until recently.

Perhaps in the long run, since almost any place specifies that void expressions are pvalues explicitly, we could even turn it into a note:

Note: Expressions of type cv void are always prvalues.