Oddly enough I was writing about ASICs yesterday :-) http://hashingit.com/9-where-next-for-bitcoin-mining-asics. There's unlikely to be anything smaller than 20 nm this year and the value of Bitcoin ASICs are unlikely to be anything like large enough to get the fabs interested until they have a lot of capacity. 10nm is at least another 2 years after that (70% shrinks take 2 years). Even when they do happen though there's not going to be anything like the same jumps in performance that the shift to 28 nm devices has seen. ASIC designers are going to have to get much more creative and that will increase their design times and costs.