Maybe not, but they're closely connected. Kurzweil's methodology, from what I vaguely remember of reading his books years ago, is to take his extrapolation of Moore's law, and apply it to non-CPU things. This is sane, because to get a Singularity, you need those non-CPU things - you can't build a Singularity just because you suddenly have a bunch of exaflops; we just don't have the software.
If computing power keeps growing, but we aren't getting actual new benefits out of it, then that does drastically affect the accuracy of the 'overall' prediction.
(Even if we had literally infinite computing power, I don't know whether we could build an AI quickly; the theoretical models I've heard of like AIXI exist only as theorems, and though we would be able to brute-force the space of all possible programs, how would we know which one is the AI we want? This shows you that Moore's law likely is not the restraint on the Singularity.)
This is sane, because to get a Singularity, you need those non-CPU things
No one knows including Kurzweil. Maybe it really does take x number of flops, maybe we need better software. Maybe we need quantum computers. Anyone that pretends to know is lying.
If computing power keeps growing, but we aren't getting actual new benefits out of it, then that does drastically affect the accuracy of the 'overall' prediction.
(Even if we had literally infinite computing power, I don't know whether we could build an AI quickly; the theoretical models I've heard of like AIXI exist only as theorems, and though we would be able to brute-force the space of all possible programs, how would we know which one is the AI we want? This shows you that Moore's law likely is not the restraint on the Singularity.)