Some further comments:
Recently, NASA Tech Briefs (June 2010) published an article on a fully printed phased array antenna which used FETS based on carbon nanotubes (CNT's). These operated at 5.2GHz, but were capable of ~100GHz operation at high power. Switching voltage was a low 1.8V while previous flexible FET's required ~50V. The article contained a wealth of performance data. You can find the link here:
Kind of like the "Replicator" of Star Trek fame.
Punch in a demand for a particular part, and wait for it to appear. In this case, the raw materials can be pretty much limited while getting a wide variety of functions out. While the size is currently large, we can expect reductions over time. I would like to see some performance data published on what can be made now.
Way-back-when, I remember getting my first schematic capture program from a company called DataNet. It ran on my IBM AT, and was the greatest thing since sliced bread.
A few years later, I was able to create a state machine in schematic form and receive a simulation, and a file suitable for programming something like an 18CV8. Total magic: stepper motor controller in two (software) steps.
Now it sounds as though we are closing in on the following work flow: define the design via VHDL, schematic, tables; simulate; send it to your desktop printer to be realized in a few minutes in hardware.