Here's Mike Roco, senior adviser for nanotechnology to the US National Science Foundation, on "Nanotechnology's Future":
Today nanotechnology is still in a formative phase -- not unlike the condition of computer science in the 1960s or biotechnology in the 1980s... Over the next couple of decades, nanotech will evolve through four overlapping stages of industrial prototyping and early commercialization...
Below is a graphic that illustrates those four stages, or generations:
Notice how Roco describes the significant differences that will emerge during the fourth generation:
After 2015-2020, the field will expand to include molecular nanosystems -- heterogeneous networks in which molecules and supramolecular structures serve as distinct devices. The proteins inside cells work together this way, but whereas biological systems are water-based and markedly temperature-sensitive, these molecular nanosystems will be able to operate in a far wider range of environments and should be much faster. Computers and robots could be reduced to extraordinarily small sizes. Medical applications might be as ambitious as new types of genetic therapies and antiaging treatments. New interfaces linking people directly to electronics could change telecommunications.
I want to emphasize what he's saying here: "whereas biological systems are water-based and markedly temperature-sensitive," by contrast, fourth-generation molecular nanosystems "will be able to operate in a far wider range of environments and should be much faster."
And guess what -- those "extraordinarily small" computers and robots that Roco foresees not only will provide smaller, faster, better medical applications and communications interfaces, but also will form the internal structure of a nanofactory.