Sum of costs taken from sum of income is profit. Costs are easily controllable by not going with bleeding edge technology. Income is not easily controllable, but is dependant on the market, and the market wants hashrate regardless of the type of technology. If you can make a device that hashes faster NOW cheaper or at the same price as a "better technology" device 2 months from now, you will gain market share and therefore more profit. The key here is that the hash rate per chip is not as important as the cost of manufacturing that same hash rate with the cheapest method possible.
I don't think you're thinking clearly. Better technology is better
because it's
cheaper to manufacture (Per GH/s). Why else would people even use it?
Again, I don't think you see my main point - and I'm sorry if I'm not making it more clear, I'm not really sure how to put it much differently. You're arguing the merits of feature size (the "better technology" that everyone keeps talking about) as an automatic higher output for cost/watt, as opposed to total all-in cost of production. My point is that the all-in cost per hash is what matters, no matter what the technology is. I was only half kidding when I said they could use vacuum tubes if it made financial sense - if that's the cheapest way to get the highest hash rate commercially, then guess what the market will buy? Vacuum tubes.
The difference is the up-front cost. It might cost few hundred thousand dollars to design a 130nm chip, while costing about $2m to produce a 28nm chip.
But, the chips will cost (about?) the same per mm2 to manufacture.
You can't say all that matters is manufacturing cost, and then say that feature size doesn't matter, because manufacturing cost depends on feature size. The smaller the feature size, the cheaper the chips per hash.
Actually, no - the smaller the feature size, the more transistors per wafer - but again, if the engineering layout of the chip is better even at a larger size (which is what LC contends with their specific design), then it's absolutely possible to have a better hash rate at larger sizes for less absolute dollars (or yuan or BTC or whatever.)
Now, the one catch with that may be the demand for 28nm fabs. AMD and Nvidia and lots of other companies are all going to be wanting access to a limited number of wafer runs. So the cost for 28nm might be slightly higher then other levels. So 40 or 65nm chips may be cost competitive.
And, of course, there's turn around time. Bitcoin mining, unlike pretty much everything else, time to market is absolutely critical. One month of delay can mean the difference between success and failure. That's where 130nm tech is superior at the moment. Labcoin may pay for all of it's 130nm chips, plus profit in the month or two before 28nm chips flood the market.
I think you just accidentally made my point for me after all. And again, Labcoin is not hanging their hat on 130nm - first comes proof of concept and getting the assembly process worked out so they can ramp up on the next best cost combination of size, time to manufacture, and design. It's not an engineering race, it's a business race.
And furthermore, the MFG price of the chips isn't that important, at this point. Next year, though it will start to matter more.
But it is absolutely important - for investors with coins ready to drop on some company, all-in cost per hash (remember, they're not selling technology specifics to end users, they're selling hash rate) is very, very important to the success of the company. It matters now for people investing now.