This is a common misrepresentation of how quantum computers work. Because we can start in one state (e.g. |0^n>, the all-zero n-qubit state), and apply the Hadamard gate to each qubit. And now we are in the universal superposition state (measuring has an equal probability for each n-bit bitstring). Oh no! what happened?
If you are under the belief that quantum computation is 'like classical computation, but in parallel', going from one single state to a superposition is impossible. In other words, it doesn't accurately capture what quantum computation does.