This is the other main flaw in Kurzweil's argument, that Moore's law somehow translates to software. Hardware has been following Moore's law, but software hasn't. We are just getting to the point of basic usability of voice recognition and computer vision, this does not translate into reverse engineering the entire brain in 10 years. In fact, I don't think we've sufficiently proven that it's even possible to replace the brain's wetware with hardware based on transistors. Also, the brain's ability to develop depends on its connection to the human body and the sensory organs; having a human-like brain that exists without a body is a postulation, not something that will just happen given enough time.
What the large print giveth, the small print taketh away.