I'm not so sure about that, when we'll have enough horse power and RAM to completely emulate the complexity of a brain, and we will have working models of neuron activity in the brain, then I think we'll be able to simulate a human brain.
I don't think we'll need to really understand the whole working of the brain or do any breakthrough progress in AI to do this. We would only need to create something similar to the inner brain working, and I think similar capabilities will emerge, even if we don't know how they work.
That is exactly my thought. Simulation of brain processes, even without truly understanding them, is probably our best chance of developing 'REAL' AI. It has much more potential than hoping that with more lines of code a program will somehow become 'intelligent'.