Chip design is a complicated thing. One would assume people would realize this, but I see the process over-simplified repeatedly. There are far more variables involved than just "How many NM is it?"
FC found this out the hard way, the same as BFL and many others. They taped out chips that ended up sucking up too much juice and running way too hot. There are a variety of solutions, but I like the path FC is taking. His 40nm chips will likely outperform the 28nm chips from other companies by a fair margin. And they will cost a fraction of what the 28nm chips will to manufacture. The problem though will be massive heat generation. Solution? Immersive cooling.
I mean really, there is an obvious strategy here. Lets look at the advantages:
Chips at less than 1/10 the cost of the competition
Higher performance chips (as measured in Gh/s)
Plenty of available foundries
Higher density cards and racks
Lower electricity costs for cooling
Deliverable turnkey farms (in storage containers)
Lower facilities costs ( don't need a traditional datacenter)
This really is the future of ASIC mining by inventing it, just like the tag line says. Friedcat is addressing issues most of the competition hasn't even contemplated yet. He is going industrial.
When mining goes "industrial" the community is dead. The capabilities to deploy a liquid cooled behemoth fall to the very few, so hopefully the future has an answer to the block halving issue because the return on something that expansive will be questionable from the get-go, let alone remaining viable long-term. Removing the concept of the block reward halving, the specs on a box would have to be absurdly impressive considering the higher-end that folks like KNC are starting to shoot for. If a Neptune becomes the new norm for the hobbyist-turned-pro miner, we're at the verge of 2014 seeing an even more absurd growth curve than 2013 as far as hashrate distribution.
Not to mention the ever-present X-factor that you could get a BFL-type company or even a KNC who says screw it, we're making our retirement, who cares how much hardware we push out and what impact it has on "the network". Standard industry practice among the more greedy VC types is to bleed an industry of all the cash it has before moving on to the next industry, leaving both companies and customers in the dust. BTC isn't immune to that, look how little cash it takes to tip markets and drive prices up and down the "to-the-moon" scale people like to use.
I don't blame FC for thinking big, I blame the ASIC manu's across the board for thinking scaling up to industrial requirements is the future of mining. The only direction this is going is centralization.
