This isn't a particularly good introduction to quantum computing. The fact that we can't scale down transistors into the quantum realm doesn't have much to do with the real motivation behind making quantum computers. The real motivation for making a quantum computer is that it allows us to perform algorithms that exploit the quantum behaviour of qubits.

To give a concrete example of this, there is a basic quantum logic gate called a Hadamard gate. This takes a qubit in the state |0> and transforms it into a qubit in the state (|0> + |1>)/sqrt(2). The resultant state is what is known as a superposition state - namely a superposition of both the |0> and |1> states. You can think of the |0> and |1> states as being analogous to the 0 and 1 states of a traditional bit. What this means is that when the qubit is measured (read queried) it will return a value of 0 or 1 with equal probability.

The point I'm making is that this is radically different to traditional computing and isn't simply just the next step along in 'valve transistor -> semiconductor -> ?'