As a refference for Harry:
http://en.wikipedia.org/wiki/Die_shrink
4 nodes better then 40nm corresponds to a FINFET 11nm.
But as I mentioned previously, I am sure this POET statetment of "4 nodes better in speed" is very conservative; they previously mentioned clock speeds of 40GHz (Si-CMOS is sticking at about 4GHz), and recently also a 5x speed advantage! Well, the future will tell us! First quarter of 2015?
Half-shrink
Main ITRS node | Stop gap half-node
| 250 nm |
220 nm |
| 180 nm |
150 nm |
| 130 nm |
110 nm |
| 90 nm |
80 nm |
| 65 nm |
55 nm |
| 45 nm |
40 nm |
| 32 nm |
28 nm |
| 22 nm |
20 nm |
| 14 nm |
16 nm |
| 10 nm |
11 nm |