Intellect in the machine

Here's a random thought. People have theorized that as the complexity of computers increases, it is likely that some form of machine intelligence and self-consciousness will appear. At what point, however, can we say that this is unlikely, and that it is more logical to assume that intelligence and consciousness cannot be created through technology? When a brain-like network has as many connections as the human brain, but no self-consciousness? 10x? 100x? I don't have an answer, I'm just wondering.