Yep, that's a biggie, I know bitcoin ASICs in general have a bit of a way to go before they're fully bleeding edge state of the art, but this seems like an exponential jump that's larger than might be expected. As if we were all using 8088s at 6mhz (PC/XT class) and then all of a sudden there's a 600Mhz Pentium III (Or if you like 486sx33s to 3.3Ghz  i5s)

Then experience in the community thus far has led us to believe that fabless developers and layout houses have been extremely over optimistic about device performance, and we've seen products or samples hit 1/6 to 1/3 of expected process maximum clock rate, and round about a half of initially projected hash.

I don't think Yifu or friedcat ever told us what they were aiming for on their processes, but initial CPUs on 130nm were running 1.5Gh before they were tuned up. I think 500Mhz on their devices is considered "highly overclocked" so they're below 1/3, nominal clock being under a quarter.

I have some suspicion that the massive parallelism of multiple SHA cores throws the developers for a loop as regards power distribution and heat removal, when other devices they have made have maybe been more sequential, i.e. less of the chip actually turned "on" per clock tick, leading to false assumptions.

Anyway, if Uniquify has spun Hashfast a line of 28nm being good for 4Ghz... then past experience seems to indicate a SHA ASIC will have an absolute maximum of about 1.33Ghz on that, and may need to be dropped to 1Ghz or below to meet power targets. 100 engine chip at 4Ghz, too good to be true. 400 engine chip at 1Ghz, too big to be true.

So from extrapolating from the previous efforts in the field, it certainly seems too good to be true, and sideshow shenanigans with 419-like megabuck solicitations and mysterious shills don't paint a pretty picture.