The ASICs themselves don't need more than 64-128 KB of SRAM per circuit to be able to hash much more efficiently than a GPU for any N.
Sure, so would FPGA's but at what cost? I mean really at what cost?
Gen1 Grid seed and Alpha T will suck at VTC mining, Yeah they might do it at a better Hash/KWh out of the gate but there is much more to consider. Such as re-sale value (in case of crypto-crash) and the premium needed to aquire the ASIC's to start off with.
So, fast forward to Gen 2, or more preferably Gen 3, GridSeed. Its got more ram, Its faster, its 24nm or whatever. Its pretty sweet for mining VTC by todays standards. But what about AMD/Nvidia/Intels next chips, The ones that are still just multi-million dollar computer models at the moment. Those are getting faster too.
Unless scrypt ASIC rolls out way different than BTC asic in some way. I don't forecast a point I can forsee where I would rather have some $7,000K Propriety black box to get 30Mhash Scrypt (15Mhash VTC) When I can buy AMD R11 980X Ultra's for $499 MSRP. For all we know these future serious cards might hash at 2Mhs scrypt.
I'm betting my money that GPU's will need to run, One way or another for a *long* time. Regardless of grid-seed or whomever is doing behind closed doors at the moment.
You don't need more RAM; at the current level of RAM it is possible to mine VertCoin with approximately the same efficiency as LTC. This is true for the upcoming adjustments in N values, too. You just need to add a few circuits to regenerate some values on the fly when necessary, for whatever N value.
Which means that when Joe the ASIC wafer designer who makes ASICs for Litecoin sees that VertCoin has some value, he can add these circuits for barely any design overhead to his next generation of ASICs and then we're left with very early ASIC adoption for VertCoin, which is exactly what you guys sought to avoid.
