Maker of First Bitcoin Mining ASIC Acquired for $466M | Hacker News Reader

Maker of First Bitcoin Mining ASIC Acquired for $466M

coindesk.com

38 pointspmorici19 comments

This acquisition makes absolutely no economic sense to me.

I have not seen estimates that put ASICMiner over 15% of total mining, and have not seen any indications that they even have 16nm parts.

Presuming they do have 16nm parts, and can capture 100% of the GROSS revenues of mining through to 10nm part availability, (call it 2 years) they will have GROSS revs of (at today's prices): $850mm, or $425mm/yr

That's Gross. For the whole industry over the next two years. Net Income will be dramatically lower than that for the industry as a whole. And, every process jump you are at risk of being behind, and losing the sweetest bite at the apple.

I may just be jealous, of course, but I don't see how these economics pay off. In reality, you will have a hard time with more than 1/3 of mining power. If we assume ASICMiner is at 15%, that's max $75mm or so in gross revs per year. Is Bitcoin mining really a 6x Revenue /30x earnings business? Especially when 20%+ of that revs just came off the underlying's price moves in the last two weeks?

At any rate, I just don't see this as a sensible acquisition under any circumstances, and if I could short it somehow, I would short the hell out of it.

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#...

[2] https://bitcointalk.org/index.php?topic=1186100.msg12703641#...

[3] https://bitcointalk.org/index.php?topic=1186100.msg12707371#...

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.

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.

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.

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...

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. :)

Government subsidized electricity in China.

Am I the only one who thinks bitcoin is nothing less than Tulip mania https://en.wikipedia.org/wiki/Tulip_mania fueled by Chinese hardware manufacturers & US VC firms like Andreessen Horowitz?

Tulip mania's peak period lasted 6 months or so. Bitcoin has been growing for 7 YEARS. Why? Bitcoin solves a lot of real world problems. Tulips don't.

At least ~50 international financial institutions have R&D groups directly working with block chains / Bitcoin (search "R3CEV partners"). Central banks have reported positively on Bitcoin . As recently as last week a group of 90 central bankers met at the Fed to discuss the topic.[1] I can understand why it looks like a mania to an external observer's eye, but it is definitely not :)

[1] http://www.coindesk.com/central-banks-blockchain-federal-res...

Well, tell that to online drug dealers and ransomware creators.

How does HN's duplicate detection works? I had submitted this same URL 3 days ago: https://news.ycombinator.com/item?id=11866192

Will cryptocurrencies ever become large enough to compete with national currencies? I guess I haven't seen much adoption since 2011 or so, at least not compared to the hype generated in the news media.

And if Bitcoin isn't to be big, what will the acquisition of this company do for those who acquired it?

> Will cryptocurrencies ever become large enough to compete with national currencies?

it is an indirect consequence of my intended inference that mathematics will deplete the value of cryptocurrencies while simultaneously increasing the value of blockchains

> what will the acquisition of this company do for those who acquired it?

probably very little, both positive and negative

in a future where math has yet to have made nonce discovery superficial:

the company can use these chips for nonce discovery and increase their own bit wallets and move that currency around the community to convert it into 'national currency', or sell the chips and make 'national currency' from others wanting to use the chips for nonce discovery

in a future where math has made nonce discovery superficial:

the chips could probably still be useful for other applications.. look at another integrated circuit, the 555(o), whatever the original application was i am sure the developers would have been unable to predict all of the utilisations people have developed as a result of having the chip in their toolbox

(o) https://en.wikipedia.org/wiki/555_timer_IC

For some reason, this question triggered an idea. Why not create a 'gift card' system that...uses RPOW/Blockchain/Bitcoin tech, that replaces whichever currency that was used?

I know money and currency is tough to replace from the banks and nations crippling grip but...couldn't large retailers just use their sizes to shift how consumers pay for their services. IMO, with Starbucks having more than 1 Billion in their gift card network, I see this idea growing like a weed, if a few people were to experiment with it.

On a related note, I think digital currency(ies) could give all parties involved what they want (money/payment/tractability for the companies. goods/services/privacy for the consumer. tractability/data for govts). I hope this industry blossoms, I'm sure it will be only time before someone turns this industry on it's head.

I toyed with an idea a bit like that a few years back. The tokens would be denominated in your national currency (so no exchange rate volatility), could be bought or sold (in batches) for free (depending on mechanism), and were for online purchases or transfers only (so no interfacing with physical point of sale presences required). The purchases could be made for free via a bank transfer to a specified account with your account number in the reference field (these are normally free and near instant in the UK via the Faster Payments Service) and if you wanted to cash out your tokens could could similarly arrange a bank transfer back. It was to be completely centralised, so nothing like Bitcoin. It was actually meant for another idea which required an escrow service. There were no specific advantages to the "senders" of the tokens other than that they could use the other service, but the big benefit to the "receivers" of the tokens was that they didn't have to pay the high credit card processing fees or PayPal fees. Anyway, lots of challenges to setting up a service such as this, such as needing a business account to send/receive the transfers (business accounts in the UK attract usage fees unlike normal personal accounts), and of course all the payments regulation, which would have meant it would have been difficult to operate for free (although the idea had been that the other service would subsidise it).

There's no real reason for companies to do what you're suggesting. Why would retailers move away from their local currency? And even if they decided to do that why would they switch to something like the blockchain instead of just a database to track holdings?

It would need to be a concerted effort, imo. Not just one or two actors. But if enough market share made the move, the benefits? Numerous. As I already outlined, in my OP. In addition, to what I said, the (new) market would cut out all middle men (Visa, etc...Chase/BoA, etc...maybe even others). I think some downsides would be more or bigger insurers and maybe an aggregate of wealth but even our current system doesn't tackle these issues.

No system will be perfect in the first increment but I think this would be a step in a better direction, for all involved.

> I guess I haven't seen much adoption since 2011 or so

Here are some huge companies that have started accepting it in recent years:

  - WordPress
  - Microsoft
  - Expedia
  - Overstock
  - Dell

Also worth noting that Braintree (a PayPal-owned payment processor) accepts it, and there have been rumors for a while that PayPal itself is gearing up to do so.

Reply on news.ycombinator.com