Ask HN: Cost/time/obstacles for developing a modern, fast MIPS64 board?

With the recent discussion about x86 and other non-open platforms (https://news.ycombinator.com/item?id=11422531) I was wondering how feasible a (free/open) desktop-capable MIPS64-based single board computer project would be. Commercial success/viability is not a requirement.

PINE A64 raised $1.7m via kickstarter, but it's a low-power SBC and I can't tell how this amount relates to actual development/prototyping costs. Many companies have their own (not so low-power) MIPS64 hardware (e.g. http://cavium.com/OCTEON-II_CN68XX.html) and there's even an old university project that designed and locally manufactured a MIPS board (http://www.cse.unsw.edu.au/~cs9242/asyst/) so it's not such an unreasonable idea, I hope.

Can any experts please share their insights and come up with an estimate for cost/time to develop e.g. an 8/16 core MIPS64 SBC with DDR4, PCI slots and other workstation board features, up to a working prototype stage? What would be the biggest obstacles?