Not all Bitcoin ASICs are the same.
You can't compare friedcat's first gen ASIC and bitfury's second gen ASIC to HashFast's 3rd gen ASIC.
28nm is nothing like the larger feature sizes, 1 billion transistors pumping out 400GH/s is nothing like tiny previous designs.
Cointerra's credentials are most certainly an argument against them.
Their expertise is throwing the IP Samsung licensed from ARM onto a SOC. Low power, low gate activity ratio SOCs, not huge hot high-performance 3rd gen Bitcoin ASICs.
Neither Friedcat nor (afaik) bitfury ever even designed an asic. Any asic. friedcat did get help from some people who did with experience in... embedded soc's.
This "high power" "400 billion transistor" FUDis just that: FUD. Copy pasting a single hashing core a few hundred times isnt exactly the hardest thing on the planet, especially not when compared to laying out a modern SoC with half a dozen clockdomains. And if anything, designing for low power is what really takes skill and experience, its the hardest thing there is. High power is a pretty simple problem by comparison. Lastly, designing for 28nm also isnt meaningfully harder than for other process nodes. The fabbing is a different story, but thats done by tsmc/GF/whatever. Being ex samsung, they are likely the only bitcoin asic designers with 28nm experience.
Anyway, its clear you have an agenda here, and Im getting really tired of this mentality of rooting for one company and fudding all the others (wtf is up with that?), so your ignore button is about to get a bit more yellow, spare yourself the trouble of answering.