Right at the moment the current hashrate is 723 Phash/s, so BitFury hashing at 200 Phash/s rate would get ~28% [1] of the newly generated bitcoin in average, i.e. 0.277 * 144 [block/day] * 25 [BTC/block] ~= 996 [BTC/day]. (Assumes that the difficulty approximately keeps up with the hashrate, which seems to be fine to me.) That translates to more than 400,000 USD per day if BitFury manages to sell them at the current price (highly unlikely) without disrupting the market (impossible at all), but yeah, that gives the rough order of magnitude.
[1] This also means that putting 650 Phash/s gigs into the network is not as profitable as it seems, since it will double (55%) the ratio while three times more expensive (if it linearly scales).
Since the amount of Bitcoins introduced into world daily is near constant (except for the scheduled halving of the rewards every few years), SOMEBODY is selling most of the freshly mined coins without disrupting the market.
It doesn't really matter how high the difficulty is or who the miner is the coins are mined and most of them sold.
There might be a small fraction of coins which are still mined in pools using outdated miners and those people don't bother selling them coins(fractions of coins that is).
It has been accepted wisdom that most mining operations sell most/all of their mined coins.
It's a race to the bottom. It will end with the party that manages to just win that race by making the smallest profit possible at the highest efficiency of Joules per hash computed. Bitcoin is interesting for many reasons, the real-world effects of a couple of configuration settings and some cleverly picked auto-scaling parameters are immense. The number of orders of magnitude that the protocol has survived with minor tweaks to date is very impressive.
Anywhere in the USA who has to pay USA power prices is going to be at a disadvantage by that alone then. It would seem that locating the hashing equipment to the place with the cheapest power possible would be a logical step when things start to get cut that fine. Unless that could be overcome with clever usage of naturally generated local power (such as solar power).
You'd think Canada would have more major data centers. They've got a lot of extremely cheap hydro power. You could surely lay ultra-high-speed fiber along the same right-of-ways that the transmission lines use. Also, it's cold as hell more often than not, so cooling is less of an issue than putting them out in the desert somewhere.
Photovoltaic power is actually less efficient where it's warmer. The ideal for photovoltaic generation is somewhere cold but sunny. And I'm not aware of too many solar thermal plants that are online and profitable versus other generating methods.
USA (Washington state) has the cheapest power in the world (at volume -- places like Iceland are cheaper/nearly free, but with too limited capacity for large scale mining).
At that price point, yes. Maybe the problem has been not enough power required... I don't know. I just know that most bitcoin companies in Iceland have moved or are moving to get cheaper power elsewhere.
There have already been articles about people doing that a couple years ago. I believe they were located in a cooler climate as well, made cooling the datacenter a bit cheaper so more power was going to computing than maintaining the computing equipment.
I'll look for that article now.
EDIT: Not the article I was looking for, but it covers it and was from 2013. This guy was using the cheap Icelandic power and the cold Icelandic air to his advantage.
What you describe is price-based competition. It’s probably the most documented situation in economics, and it present in many markets. It is a race to the bottom, but rarely ever ends; when it does, it is generally by disruption (something people on HackerNews know a lot about). In the mean time, it drives technical innovation after technical innovation—but I doubt, after 40 years of Moore’s law that it will suddenly end.
[1] This also means that putting 650 Phash/s gigs into the network is not as profitable as it seems, since it will double (55%) the ratio while three times more expensive (if it linearly scales).