IMO, uploading will require a large increase in computing capacity
just to run the upload, plus major advances in fields like
neurophysiology. Many of the uploading methods we have discussed
here depend on nanotech to derive the state of the meat brain for
flash uploading or to provide the interconnection for gradual
uploading. If you think uploading will precede nanotech, you will
have to develop a non-nanotech solution to one of these problems.
By contrast, nanotech is a much easier problem. In "Nanosystems",
in 1992, K. Eric Drexler described diamondoid nanotechnology
rigirously, and sketched out an implementation path via
wet chemistry less rigorously. The number of problems to solve
along this path is miniscule by comparison to uploading problems.
Furthermore, the current rapid pace of research in nanotubes gives
us what appears to be a simpler path to diamondoid nanotech.
Therefore I would vote for nanotech first and uploading second,
except for one little thing:
There is the potential for "spontaneous order" and other emergent
behaviour of complex systems to create a superintelligence based
on the current internet, together with some unpredictable additional
factor. This might occur precipitously, as a "phase change"
rather than as a build-up. If so, the resulting SI may complete
the development of nanotech and use it for uploads at what
appears to humanity to be the same time.