4 ms·

It's not ASICMiner that is being acquired, it's Canaan Creative. You seem confused because you seem to think Canaan Creative does the mining. They don't. They d

by mrb 10y ago

It's not ASICMiner that is being acquired, it's Canaan Creative. You seem confused because you seem to think Canaan Creative does the mining. They don't. They design chips and mining gear. That's it. They are a traditional hardware company, not a mining company. I don't think anybody knows how much market share Canaan Creative has. It may very well be higher than 15%. As to 16nm: their CEO NG Zhang said they are already working on one, Avalon 7.[1] He also claims it is not that crucial to have 16nm right now (but it will be in 6-12 months). 16nm may cut power consumption by up to 50% compared to 28nm according to him,[2] however he rightly points out even professional miners like BitFury who have developped 16nm gear still run plenty of 28nm gear since mining is still highly profitable right now.[3] Finally, Canaan Creative is the Bitcoin ASIC designer with the most experience. They have done 5 generations of chips: 110nm, 55nm, 40nm, and two 28nm chips. They all shipped and were very successful. So I am not too worried their 16nm chips will come when the time is right. [1] https://bitcointalk.org/index.php?topic=1453737.msg14843375#msg14843375 https://bitcointalk.org/index.php?topic=1453737.msg14843375#... [2] https://bitcointalk.org/index.php?topic=1186100.msg12703641#msg12703641 https://bitcointalk.org/index.php?topic=1186100.msg12703641#... [3] https://bitcointalk.org/index.php?topic=1186100.msg12707371#msg12707371 https://bitcointalk.org/index.php?topic=1186100.msg12707371#...

vessenes 10y ago
This is helpful context, and I like the idea that the most experienced Bitcoin ASIC team out there is worth something. But that's really a lot of something on top of what their mining skills seem to be; there are a number of bankrupt mining company chip designers out there I know personally who aren't being snapped up by the market -- and this would seem to indicate that bitcoin chip design skills don't have quite the value Canaan is getting. 28nm gear is not profitable right now as far as I know, or it just got reprofitable a day or two ago. 28nm parts give us something like .51 W / GH = 12 kwh per day per TH. That power quote may be at the wall, but probably it's a bit more than that. 1 TH = $1.74/day right now, pre-halving, soon to be $0.85 Fully loaded costs of energy and datacenter: $.06 - $.12 / kwh = costs $.75 - $1.44 day. So, at $400 coins, those rigs were a loss unless you were in very cheap energy and labor markets, and then just barely making some cash. At $670, they are making a bit of money again, if you kept them around and didn't decomm them in favor of 16nm parts. I think the end of the road has been not just in sight but essentially here for 28nm parts for the last couple of months. But, I did oversee a multi-megawatt mining business that mined unprofitably for a number of months to finish out our datacenter lease, so that is also a possibility here, and might explain why 28nm miners won't go away right this second. That said, if you know of ways to get fully loaded costs down into the $.02 -$.04/kwh, I would really like to learn more about that - I'd LOVE to be wrong about this.
- pmorici 10y ago
  Government subsidized electricity in China.
- mrb 10y ago
  Canaan's 28nm chip (Avalon A3218 in the Avalon 6) does a lot better than what you quoted: 0.3 J/GH at the wall. As to electricity costs, I was personally running a small 10 kW GPU farm at $.02/kWh back in 2011 in Douglas County, Washington where most Public Utility Districts are basically non-profits due to historical reasons in the region. Most smart large-scale miners are paying $0.05/kWh or less. If not, they are dumb and go bankrupt like KnCMiner did 2 weeks ago. So 1 TH/s of Avalon 6 gear costs $0.36/day, however at $0.05/kWh it mines $1.74/day right now (or $1.00/day 2 weeks ago when BTC was around $400). Certainly profitable.
  - vessenes 10y ago
    Thanks for this context also. There are some capex issues with scaling up that farm, and I bet you aren't costing your own time in there. But, I hear you that it's possible, and it's interesting to hear that $.05 fully loaded is possible outside of government subsidy situations.
    - mrb 10y ago
      No, I did not count my labor. It is true that capex and opex of mining gear is well-known and can be calculated so as to be profitable in and of itself. However knowing how to manage all the other costs (labor + capex on the DC itself) is what distinguish good mining companies from bad ones. IMHO these areas is where KnCMiner made the biggest mistakes. Just compare a picture of a KnCMiner DC vs. a Chinese DC and you will see KnCMiner did not try hard to build cheap DCs...
      - vessenes 10y ago
        We ran two different ones; one with engineered fire suppression and custom airflow, and one with a bunch of fans from the equivalent of Walmart. The walmart guys ended up with a lot more of our business. :)

by mrb 10y ago

vessenes 10y ago

This is helpful context, and I like the idea that the most experienced Bitcoin ASIC team out there is worth something. But that's really a lot of something on top of what their mining skills seem to be; there are a number of bankrupt mining company chip designers out there I know personally who aren't being snapped up by the market -- and this would seem to indicate that bitcoin chip design skills don't have quite the value Canaan is getting. 28nm gear is not profitable right now as far as I know, or it just got reprofitable a day or two ago. 28nm parts give us something like .51 W / GH = 12 kwh per day per TH. That power quote may be at the wall, but probably it's a bit more than that. 1 TH = $1.74/day right now, pre-halving, soon to be $0.85 Fully loaded costs of energy and datacenter: $.06 - $.12 / kwh = costs $.75 - $1.44 day. So, at $400 coins, those rigs were a loss unless you were in very cheap energy and labor markets, and then just barely making some cash. At $670, they are making a bit of money again, if you kept them around and didn't decomm them in favor of 16nm parts. I think the end of the road has been not just in sight but essentially here for 28nm parts for the last couple of months. But, I did oversee a multi-megawatt mining business that mined unprofitably for a number of months to finish out our datacenter lease, so that is also a possibility here, and might explain why 28nm miners won't go away right this second. That said, if you know of ways to get fully loaded costs down into the $.02 -$.04/kwh, I would really like to learn more about that - I'd LOVE to be wrong about this.

pmorici 10y ago
Government subsidized electricity in China.
mrb 10y ago
Canaan's 28nm chip (Avalon A3218 in the Avalon 6) does a lot better than what you quoted: 0.3 J/GH at the wall. As to electricity costs, I was personally running a small 10 kW GPU farm at $.02/kWh back in 2011 in Douglas County, Washington where most Public Utility Districts are basically non-profits due to historical reasons in the region. Most smart large-scale miners are paying $0.05/kWh or less. If not, they are dumb and go bankrupt like KnCMiner did 2 weeks ago. So 1 TH/s of Avalon 6 gear costs $0.36/day, however at $0.05/kWh it mines $1.74/day right now (or $1.00/day 2 weeks ago when BTC was around $400). Certainly profitable.
- vessenes 10y ago
  Thanks for this context also. There are some capex issues with scaling up that farm, and I bet you aren't costing your own time in there. But, I hear you that it's possible, and it's interesting to hear that $.05 fully loaded is possible outside of government subsidy situations.
  - mrb 10y ago
    No, I did not count my labor. It is true that capex and opex of mining gear is well-known and can be calculated so as to be profitable in and of itself. However knowing how to manage all the other costs (labor + capex on the DC itself) is what distinguish good mining companies from bad ones. IMHO these areas is where KnCMiner made the biggest mistakes. Just compare a picture of a KnCMiner DC vs. a Chinese DC and you will see KnCMiner did not try hard to build cheap DCs...
    - vessenes 10y ago
      We ran two different ones; one with engineered fire suppression and custom airflow, and one with a bunch of fans from the equivalent of Walmart. The walmart guys ended up with a lot more of our business. :)