While both 3D stacking and 2D shrinking provide benefits, both require investment to develop. In particular, if a single EUV lithography exposure tool and a complete TSV fab line both cost ~$100M, then won't limited resources force most fabs to chose one or the other?
TSV fab costs are much lower than $100M. The cost of TSVs is not the issue today. Its the design tools and more importantly, who will do the 3D assembly.
Unless something utterly revolutionary occurs, it seems almost certain
that outsourced test assembly and packaging (OTAP) companies will be
who will do the 3D assembly (except for microprocessors). At the 2009 IMAPS Global Business Council meeting (I wasn't able to attend, but I know that you presented there), Lee Smith of Amkor said that they're ready to assemble 3D stacks using TSV whenever customers demand it.. so is demand now just limited by the lack of proven EDA tools?