>I expect the same for 7nm. Both 12nm and 7nm will be much more affordable in 2020. Right now the uncertainty is 5nm and 3nm. Both are technically achievable, the problem is cost. At the moment 5nm+ cost structure looks like exclusive to certain industry and clients. [1]
That was my oldest comment I could find on HN made in 2017. Although my projection was made on Anandtech some time around 2016.
I was originally projecting a slow down to 3 years cadence post 5nm at 3 / 2nm. But at the time there were no "Trillion Dollar" company and AI wasn't a thing. I was also at the time ignorant of HyperScaler's volume.
Even if all of that will allow us to go further, 2nm in 2025, 1.4nm in 2027 and 1nm in 2029 / 2030. Surely the hypotheses remains, that at some point the market will not be able to justify the cost of new node. And will have to amortized over a longer period of time.
At 1nm, if it were to cost $30K per Wafer, a 100mm die would cost $50 at 100% yield. That is excluding the potential $2 - $3B design cost.
[1] https://news.ycombinator.com/item?id=14835609