ASICMINER: Entering the Future of ASIC Mining by Inventing It

[chriswilmer]

Thought experiment:

ASICMiner is valued at 30 million bitcoins and you decide you want to buy the whole company (all of the shares). The board agrees to this and all shareholders are forced to sell their shares at the corresponding price.
(30 million bitcoins divided by 400,000 shares).

How do you actually pay for this since it is impossible to have 30 million bitcoins?

You would need to pay in installments. Perhaps you had 5 million bitcoins in your possession already, and based off of ASICMiner's revenue (which would obviously need to be a lot for such an incredible valuation), you would pay off an additional 5 million bitcoins every 6 months or so, possibly with interest.

At least, that is one way that I could see this working.

[dree12]

ASICMiner products probably exceed 30% of the hashrate share already. But you have to realize this:

ASICMiner valuation = (ASICMiner hashrate share + ASICMiner customer hashrate share) * mining share of bitcoin valuation * bitcoin valuation

I highly doubt all mining makes up 30% of Bitcoin's industry, let alone a part of mining.

I highly doubt money supply = GDP.

They have different units, so I fail to see how they can even be compared.

You were the one that compared it to "Bitcoin's industry".

I mean "industry" as in "tourist industry". Making up 50% of industry does not mean making up 50% of GDP.

What I'm comparing is money supply to stock supply. If ASICMiner is valued at greater than all BTC combined is valued, then it follows that ASICMiner must have revenues separate from Bitcoin. This is not true, so ASICMiner stock will be capped.

No, it doesn't follow.  Both incorporate future expectations into their valuation, so it's not so simple as you imply.  They also have vastly different velocities (turnover of shares/coins).

Why would the future value of ASICMiner exceed that of Bitcoin? Unless ASICMiner is expected to make forays into mining diamonds, this doesn't make much sense.

[notme]

ASICMiner products probably exceed 30% of the hashrate share already. But you have to realize this:

ASICMiner valuation = (ASICMiner hashrate share + ASICMiner customer hashrate share) * mining share of bitcoin valuation * bitcoin valuation

I highly doubt all mining makes up 30% of Bitcoin's industry, let alone a part of mining.

I highly doubt money supply = GDP.

They have different units, so I fail to see how they can even be compared.

You were the one that compared it to "Bitcoin's industry".

I mean "industry" as in "tourist industry". Making up 50% of industry does not mean making up 50% of GDP.

What I'm comparing is money supply to stock supply. If ASICMiner is valued at greater than all BTC combined is valued, then it follows that ASICMiner must have revenues separate from Bitcoin. This is not true, so ASICMiner stock will be capped.

No, it doesn't follow.  Both incorporate future expectations into their valuation, so it's not so simple as you imply.  They also have vastly different velocities (turnover of shares/coins).

Why would the future value of ASICMiner exceed that of Bitcoin? Unless ASICMiner is expected to make forays into mining diamonds, this doesn't make much sense.

Do you really think everybody who buys mining hardware will turn a profit in bitcoin terms?

[🏰 TradeFortress 🏰]

ASICMINER could develop hardware to mine alt coins (eg Litecoin) in the future, which would cost more resources but it is theoretically possible for ASICMINER to exceed the valuation of Bitcoin.

[freedomno1]

ASICMINER could develop hardware to mine alt coins (eg Litecoin) in the future, which would cost more resources but it is theoretically possible for ASICMINER to exceed the valuation of Bitcoin.

Agree a company is dynamic not static and developing into the alt-coin market would work
However I recall friedcat saying that any such risk would not be the burden of AM investors but would be spun off into a new entity instead of added to the AM portfolio.

Edit in: Thinking about it we do review this a lot edited in after Transaction Fees

[tehelsper]

AM market cap and the total number of bitcoins aren't as related as everyone seems to assume. The market determines the market cap based on what people are willing to pay for it to gain a certain return.

Lets say after all 21 million bitcoins are generated, TX fees have somehow reached 25 BTC per block and AM holds 30% of the total hashrate. Each year, a single AM share would pay out .9985 BTC in dividends. If the market was willing to get a 1% return on AM, it cost 98.55 BTC per share and and the total market cap would be nearly 40 million BTC. I know this unreasonable, but I think it shows the market determines the market cap and there is no 21 million BTC hard limit.

A better example may be 20% of the total hash rate, 10 BTC per block, and an accepted dividend yield of 5%. AM market cap would be about 2.1 million BTC. So, if there were 10 companies the size of AsicMiner at that time, then no one could buy all 10 companies at once because the combined market caps would be over 21 million.

I've created a spreadsheet here that people can play with. It has two knobs for turning (AM hash rate, and Tx fees per block)
https://docs.google.com/spreadsheet/ccc?key=0ApG_UykyjHF0dEU2b1B5ZVZTeW5OY3NLTGRYTW9xM3c&usp=sharing

Edit: Before people start screaming, I know some of these situations are VERY unrealistic. They are just meant to point out an example.

[BitAddict]

AM market cap and the total number of bitcoins aren't as related as everyone seems to assume. The market determines the market cap based on what people are willing to pay for it to gain a certain return.

Lets say after all 21 million bitcoins are generated, TX fees have somehow reached 25 BTC per block and AM holds 30% of the total hashrate. Each year, a single AM share would pay out .9985 BTC in dividends. If the market was willing to get a 1% return on AM, it cost 98.55 BTC per share and and the total market cap would be nearly 40 million BTC. I know this unreasonable, but I think it shows the market determines the market cap and there is no 21 million BTC hard limit.

A better example may be 20% of the total hash rate, 10 BTC per block, and an accepted dividend yield of 5%. AM market cap would be about 2.1 million BTC. So, if there were 10 companies the size of AsicMiner at that time, then no one could buy all 10 companies at once because the combined market caps would be over 21 million.

I've created a spreadsheet here that people can play with. It has two knobs for turning (AM hash rate, and Tx fees per block)
https://docs.google.com/spreadsheet/ccc?key=0ApG_UykyjHF0dEU2b1B5ZVZTeW5OY3NLTGRYTW9xM3c&usp=sharing

Edit: Before people start screaming, I know some of these situations are VERY unrealistic. They are just meant to point out an example.

It is impossible that 10 companies have 20% of the total hasrate at the same time
So, second example is not good. But I get your point.

[samson]

I wonder if the Intel CPU SHA extensions will have any effect on ASIC mining ?

On chip SHA2 extensions in the Intel CPU : http://software.intel.com/en-us/articles/intel-sha-extensions

Could be big news

[dree12]

ASICMiner products probably exceed 30% of the hashrate share already. But you have to realize this:

ASICMiner valuation = (ASICMiner hashrate share + ASICMiner customer hashrate share) * mining share of bitcoin valuation * bitcoin valuation

I highly doubt all mining makes up 30% of Bitcoin's industry, let alone a part of mining.

I highly doubt money supply = GDP.

They have different units, so I fail to see how they can even be compared.

You were the one that compared it to "Bitcoin's industry".

I mean "industry" as in "tourist industry". Making up 50% of industry does not mean making up 50% of GDP.

What I'm comparing is money supply to stock supply. If ASICMiner is valued at greater than all BTC combined is valued, then it follows that ASICMiner must have revenues separate from Bitcoin. This is not true, so ASICMiner stock will be capped.

No, it doesn't follow.  Both incorporate future expectations into their valuation, so it's not so simple as you imply.  They also have vastly different velocities (turnover of shares/coins).

Why would the future value of ASICMiner exceed that of Bitcoin? Unless ASICMiner is expected to make forays into mining diamonds, this doesn't make much sense.

Do you really think everybody who buys mining hardware will turn a profit in bitcoin terms?

Point taken. I still find double digits unlikely, but I retract my statement that it would be impossible.

[stripykitteh]

I wonder if the Intel CPU SHA extensions will have any effect on ASIC mining ?

On chip SHA2 extensions in the Intel CPU : http://software.intel.com/en-us/articles/intel-sha-extensions

Could be big news

Could be. It's not clear from the whitepaper what specific architecture changes Intel are making to accommodate these instructions (i.e., it might be just a single hashing core).

Having said that I cannot believe that there aren't a few skunkworks ASIC chip designs floating around in Intel at the moment, given the number of chip designers they have.

[gog1]

I wonder if the Intel CPU SHA extensions will have any effect on ASIC mining ?

On chip SHA2 extensions in the Intel CPU : http://software.intel.com/en-us/articles/intel-sha-extensions

Could be big news

Could be. It's not clear from the whitepaper what specific architecture changes Intel are making to accommodate these instructions (i.e., it might be just a single hashing core).

Having said that I cannot believe that there aren't a few skunkworks ASIC chip designs floating around in Intel at the moment, given the number of chip designers they have.

How much did AES-NI speeds things up?  I suppose we maybe looking at similar factor.  Even with a speed up of 25-50x, CPU is still no match to ASIC.

[stripykitteh]

I wonder if the Intel CPU SHA extensions will have any effect on ASIC mining ?

On chip SHA2 extensions in the Intel CPU : http://software.intel.com/en-us/articles/intel-sha-extensions

Could be big news

Could be. It's not clear from the whitepaper what specific architecture changes Intel are making to accommodate these instructions (i.e., it might be just a single hashing core).

Having said that I cannot believe that there aren't a few skunkworks ASIC chip designs floating around in Intel at the moment, given the number of chip designers they have.

How much did AES-NI speeds things up?  I suppose we maybe looking at similar factor.  Even with a speed up of 25-50x, CPU is still no match to ASIC.

Yeah, I don't think we need to panic just yet. It's more likely to be used as a speedup for things like SSL which have to be decrypted in the CPU at the moment. Given that Intel is a conservative company with long lead times, it could be 12 months before you can even get your hands on hardware with these instructions, and even then it wouldn't be much chop as a miner. But it's still pretty cool.

[tehelsper]

AM market cap and the total number of bitcoins aren't as related as everyone seems to assume. The market determines the market cap based on what people are willing to pay for it to gain a certain return.

Lets say after all 21 million bitcoins are generated, TX fees have somehow reached 25 BTC per block and AM holds 30% of the total hashrate. Each year, a single AM share would pay out .9985 BTC in dividends. If the market was willing to get a 1% return on AM, it cost 98.55 BTC per share and and the total market cap would be nearly 40 million BTC. I know this unreasonable, but I think it shows the market determines the market cap and there is no 21 million BTC hard limit.

A better example may be 20% of the total hash rate, 10 BTC per block, and an accepted dividend yield of 5%. AM market cap would be about 2.1 million BTC. So, if there were 10 companies the size of AsicMiner at that time, then no one could buy all 10 companies at once because the combined market caps would be over 21 million.

I've created a spreadsheet here that people can play with. It has two knobs for turning (AM hash rate, and Tx fees per block)
https://docs.google.com/spreadsheet/ccc?key=0ApG_UykyjHF0dEU2b1B5ZVZTeW5OY3NLTGRYTW9xM3c&usp=sharing

Edit: Before people start screaming, I know some of these situations are VERY unrealistic. They are just meant to point out an example.

It is impossible that 10 companies have 20% of the total hasrate at the same time
So, second example is not good. But I get your point.

Yeah, I meant size in terms of market cap, not hashrate. 5 could all be mining companies and the others could be completely unrelated to bitcoin, just denominated in it.

[gog1]

actually just did some math, Core i7 3930k does 66.6 MH/s; a speed up by 50x would be 3.3 GH

I can imagine people using their machines at work to mine or mine at the library.

[dhenson]

actually just did some math, Core i7 3930k does 66.6 MH/s; a speed up by 50x would be 3.3 GH

I can imagine people using their machines at work to mine or mine at the library.

The biggest benefactor of the new instruction set will be the bot nets.

[Strange Vlad]

Two important questions about these Intel extensions:
1) When will the CPUs supporting them become available to buy?
2) What will be the hashrate? The hashrate per dollar cost of CPU?

LOL certainly I don't believe in 33GH per CPU. It would require hashing 10 cores per CPU at 3.3 GHz even if the calculation takes only 1 tick (which it certainly won't, 3-5 ticks is more realistic), and there's no reason to have as much as 10 cores for any non-mining purposes.
EDIT: disregard that, I didn't notice the point in 3.3

[mem]

actually just did some math, Core i7 3930k does 66.6 MH/s; a speed up by 50x would be 3.3 GH

I can imagine people using their machines at work to mine or mine at the library.

The biggest benefactor of the new instruction set will be the bot nets.

sad but true.

[joris]

ASICMiner products probably exceed 30% of the hashrate share already. But you have to realize this:

ASICMiner valuation = (ASICMiner hashrate share + ASICMiner customer hashrate share) * mining share of bitcoin valuation * bitcoin valuation

I highly doubt all mining makes up 30% of Bitcoin's industry, let alone a part of mining.

I highly doubt money supply = GDP.

They have different units, so I fail to see how they can even be compared.

What I'm comparing is money supply to stock supply. If ASICMiner is valued at greater than all BTC combined is valued, then it follows that ASICMiner must have revenues separate from Bitcoin. This is not true, so ASICMiner stock will be capped.

Perceived value can't be capped. One share can change hands for price X, ten shares for price Y and all shares for price Z. X probably ain't Z.

The valuation formula doesn't make sense to me. "Bitcoin valuation" seems strange for "ASICMiner valuation", in BTC at least. It's should be only about cashflow. Let's say 33% of hashrate ~ 1200 BTC per day + BTC from hardware sales (at any price) is the cash coming in. After cash expenses, the net cash flow is worth something to retain or acquire, the price of which to be determined by the irrational brains of current and future shareholders. No absolute reasons in there to "have revenues separate from Bitcoin".

How to pay? With debt, gold, property or whatever. No restriction to the amount of BTC in the economy.

Of course not only cashflow influences the perceived value, but also their current control of the mining sector. They seem to be able to choose between selling their hardware or using it themselves. If f.e. 100th comes online, it wouldn't surprise me if ASICMINER turns on 45 TH overnight to keep up their share. While if hashrate from other sources grows to slow, ASICMINER goes helicopter Ben with well-priced hardware.

Cashflow in from mining is more or less constant, cash out for mining increases with difficulty, while hardware will be the other way around. Less cash in with constant costsfor manufacturing. It will take a while, but the extreme profitability of ASICMINER will come to an end when margins become 'normal'. But I don't dare to say anything about the share price that the market will see fit for these stages.

[CMMPro]

I wonder if the Intel CPU SHA extensions will have any effect on ASIC mining ?

On chip SHA2 extensions in the Intel CPU : http://software.intel.com/en-us/articles/intel-sha-extensions

Could be big news

Could be. It's not clear from the whitepaper what specific architecture changes Intel are making to accommodate these instructions (i.e., it might be just a single hashing core).

Having said that I cannot believe that there aren't a few skunkworks ASIC chip designs floating around in Intel at the moment, given the number of chip designers they have.

How much did AES-NI speeds things up?  I suppose we maybe looking at similar factor.  Even with a speed up of 25-50x, CPU is still no match to ASIC.

Yeah, I don't think we need to panic just yet. It's more likely to be used as a speedup for things like SSL which have to be decrypted in the CPU at the moment. Given that Intel is a conservative company with long lead times, it could be 12 months before you can even get your hands on hardware with these instructions, and even then it wouldn't be much chop as a miner. But it's still pretty cool.

Exactly...not sure why people are even wasting brain cycles thinking about this.

Intel isn't interested in developing a mining chip or even to do more than a few hashes per second. It would be entirely dedicated to security applications for web commerce, website security etc.

[ElideN]

how can i be a part of those shares?