This is one of my principal reasons for being a singularity-skeptic. While runaway technological innovation might be possible, I think it's likely that economics and politics will stop it.
Physics won't stop Moore's law. Industry consolidation, slowing innovation, and a lack of demand will. I would not be terribly surprised if we never see computers more than 2X as fast/powerful as today's, since without corresponding software and networking innovation to drive demand why would they be built? That's a tad shy if what would be required for strong AI, "mind uploading," or any of that stuff.
We're already seeing that effect in chips, somewhat. The market has gotten to the point where relatively cheap chips are good enough for most people's purposes -- looking at webpages and viewing streaming video on a mobile device. You don't need a top of the line Intel chip to do that; a cheap ARM clone is fine.
There is still a higher end driver in the server space, though, as virtualization progresses... How many virtual machines can you fit into one box?
Physics won't stop Moore's law. Industry consolidation, slowing innovation, and a lack of demand will. I would not be terribly surprised if we never see computers more than 2X as fast/powerful as today's, since without corresponding software and networking innovation to drive demand why would they be built? That's a tad shy if what would be required for strong AI, "mind uploading," or any of that stuff.