I see the point. Basically if you imagine that Av is increasing, Vin = Vp - Vn is decreasing and it will reach zero at some point. Basically we can't look at the expression k*Av*Vin in a literal way I mean, as it looks, we have to look at it conceptually. By increasing K you are seeing what will happen to the circuit in the limit.
So you increase K, Vin decreases, you increase K again and Vin will decrease further till Vin is zero and this will oblige the opamp to have a finite voltage at the output.
I think it is this? Is this thinking correct?
Now when it comes to simple cascade amplifiers the thing are no so obvious. I will try to post a sketch and the comments from that professor.
But basically the Vgs of the dependent source you have chosen is going to be zero.