"Good luck finding:
a.People very well-versed in laying out bleeding-edge IC design.
b.Said people also knowing how to route by-hand the SHA256 logic cores inside the chips. Foundry-supplied IP blocks will be useless aside from memory and coms.
c.Then convincing TSMC, GloFo or Samsung to sell you time to make some test wafers. IF they will even talk/work with you Foundry cost per test run will be several 100k$ up-front.
d. rinse and repeat ‘c’ several times until the chips works reasonably ‘close to specs’. Redesign costs/time for each test run - massive.
e. And this one is a doozy - gathering KNOWLEDGABLE investors willing to put up several 10’s of millions USD to do all of this. Given the horrific past history of companies ‘making’ ASIC-based miners using their own chips and then for various reasons - mostly pure and simple nefarious greed followed closely by them frankly totally blowing the chip and power design - all failing miserably, good luck with that…
f. Convincing your Foundry to allot production time for you to produce enough chips per-run to make all this worth while. That may be as hard as ‘e’…"
I think I should build my own GPU mining rig first, hehe.