Quote from: iCEBREAKER on May 25, 2016, 08:42:16 PM

That's what the Cryptonote protocol's cryptonight PoW does.

https://en.wikipedia.org/wiki/CryptoNote#Egalitarian_proof_of_work

It uses the AES-NI acceleration built into modern CPUs

Er, that's actually a fairly minor part of what makes CryptoNight (the CryptoNote PoW) unique.

CryptoNight is actually one of the most sophisticated memory-hard algorithms out there, for reasons I think few people (Claymore excepted) appreciate at this stage. Every memory-hard algorithm needs at least a small compute-hard kernel (e.g. Ethash's use of FNV), and it was smart for CryptoNight to choose AES-NI, but that's really just a clever twist compared to the overall design of the PoW. If you've tried writing an ATI/AMD GPU miner for CryptoNight you'll begin to see why it's so unusual.

In particular,

Quote from: iCEBREAKER on May 25, 2016, 08:42:16 PM

Unless you have better fabs than Intel and AMD, you can't (economically) built an ASIC with superior specs.

This is just plain not true. Even a fairly ancient FPGA can do at least hundreds of times as many AES operations per second or per watt as the best Intel CPU, since the Intel CPUs have just one AES circuit per core (and you can't power down the rest of the chip and use only that circuit). It's also not important to CryptoNight -- the work of CryptoNight is in the memory accesses; the AES operation is just there to "daisy-chain" the accesses together into an artifact proving that you burned some memory bandwidth.