It isn't possible to clone a repo more efficiently with only an archive of a single commit.
In order to find out the data that both sides have, the client and server negotiate a set of common commits and tags. This may happen several times. Once the server knows what the client has and what it wants, it sends the rest. If the client says that it has revision A, for example, then the server knows, unless the client has said that it is shallow or partial, that the client must have all items preceding A in the history.
In your case, you don't have any commit objects, and you can't create a commit object that the server knows about, since you don't have the preceding objects to do so. Even if you did create a new repository with a single commit, the server wouldn't know about it and couldn't send you less data as a result.
So that's why you really cannot clone a repository having had only a zip file of a single revision. You'll just have to do a fresh clone.