I wrote a paper like 8 years ago on how FPGAs would likely always be too expensive to find practical use in consumer electronics as a reconfigurable cpu/system in pretty much any application. So, to be proven so wrong, I think it's super interesting.
If I were starting now I would probably wait a little longer for whatever Kevtris has with the Zimba 3000 and get that and then not have to deal with RGB modding an NES and now a PC Engine, getting everdrives, rgb cables, tracking down and fixing a sega cd, having a bin full of controllers and extension cables, lucking into a good price on GC component cables, etc etc etc. But I'm already there for everything except maybe GameGear, and certainly you're going to find some minor issues with less popular FPGA implementations which may or may not ever get fixed, so real hardware is going to still be better if you want to do the work.
Then also I don't think FPGA systems are going to be able to do PS1, Saturn, N64, and on anytime soon due to size of FPGA needed plus complexity of implementation goes way up (prove me wrong!).