Pages:
Author

Topic: 0.19 J/GH - CoinBau looks for investors in German mining technology - page 3. (Read 22380 times)

member
Activity: 94
Merit: 16
Quote
We will operate this hardware in an own mine. Our target is holding 15% of the global hash rate in 2016 and keeping this percentage long-term.  To finance the setup of this large-scale mining operation the ASICrising GmbH will become CoinBau AG, a German incorporated company, selling shares.

Teach us how to cool them and we will run rigs with your hardware.
newbie
Activity: 17
Merit: 0
CoinBau has reached the press. There have been a couple of articles:

Wall Street Journal (ENG): blogs. wsj.com/digits/2014/08/15/german-startup-says-its-new-chip-halves-bitcoin-mining-energy/

golem.de (GER): http://www.golem.de/news/virtuelle-waehrungen-sparsame-asics-fuer-das-bitcoin-mining-1408-108584.html

heise.de (GER): http://www.heise.de/newsticker/meldung/Dresdener-Startup-verspricht-effizienten-Bitcoin-Chip-2293679.html

newbie
Activity: 1
Merit: 0
hi guys,

maybe someone of the CoinBau team wants to visit our monthly Leipzig barrel-roll (https://bitcointalksearch.org/topic/m.8378170), since there is no bitcoin meet up at Dresden afaik?

if not I could offer to visit CoinBau in Dresden early September and get pictures/videos and ask specific questions for you guys if intersted?

I'm no miner and not the most (bitcoin) tech-sawy guy, but I'd be happy to help gather more info on this project.



We are glad to visit you in Leipzig on Thursday.

Best,
Sebastian - CoinBau
legendary
Activity: 874
Merit: 1000
monero
hi guys,

maybe someone of the CoinBau team wants to visit our monthly Leipzig barrel-roll (https://bitcointalksearch.org/topic/m.8378170), since there is no bitcoin meet up at Dresden afaik?

if not I could offer to visit CoinBau in Dresden early September and get pictures/videos and ask specific questions for you guys if intersted?

I'm no miner and not the most (bitcoin) tech-sawy guy, but I'd be happy to help gather more info on this project.

legendary
Activity: 1176
Merit: 1001
It's the same concept. Your return is in a really really good case scenario 160% of what you invested, after one year. Even if they had 12M to invest, you would have 12 + 12*1.6 $M worth of hardware up after one year. As you said that's roughly 50PH at the end of 2015. Not one order of magnitude off, but a lot less of the goal.

Yes, I don't have any problem predicting an exahash or bigger network before the end of 2015.
And at that point you have to tapeout your 14nm version. With what money?

I'm guessing it here, but I think that if you used some other manufacturer hardware and started hashing in one month instead of 3/4 (with a proven technology) you would mine during those 2/3 months more than the difference of the cost of electricity of running the two types of hardware.
full member
Activity: 129
Merit: 100
Come on, I followed your loved Hashfast thread. You are not naïve, you know what really matters when it comes to mining!
Fine.

You think that investing 15M thanks to this innovative design you will be able to produce enough starting miners (10PH of them?) so that you will be able to have a competitive edge against the rest of the market. This advantage will be so huge that it will let you reinvest the spent capital on new hardware exponentially, until when you reach 15% of the global hashrate.

Let's test it:
10PH, starting December (I'm being optimistic), at the cost of $5M (since that you will execute everything perfectly), with a power usage of 3MW and a cost/kWh of $0.07 (I'm, again, being optimistic. Iceland is not synonymous for free electricity. It's only a little bit cheaper. Asia is where the cheap eletricity is, and you won't be able to compete with the chinese, that have both cheaper production costs and cheaper electricity so that they can be competitive with a slighter worse J/GH). With the standard 31%/month difficulty increase (the one elaborated from the previous months), you will not recover your investment:
https://tradeblock.com/mining/a/8cca8d1f97

With 25%/month, you would make $200k.
With 15%/month (not gonna happen), you will make $6M before the end of 2015.

Now, let's suppose the last scenario to be the correct one. You make $6M before the end of the year. That's a one year only roi of 160%. That's great, however it's still one order of magnitude off from what you need to catch up with the network and maintain 15% of it.

Easy, isn't it?

Yes, easy.
But why are you assuming that they spent only $5M of the remaining $12.5M (after tape-out costs) for hardware.
If they spent all, they would start with 25 PH and would make $15M with mining in one year (2.5x $6M according to your calculations), which they could constantly re-invest in additional 30 PH, resulting in 55 PH online after one year.

So what is your prediction for the global hashrate end of 2015? More than 1000 PH?
Even in this case they would already have 5.5% after one year, this is no order of magnitude away from 15%.

Sounds crazy, of course, but not impossible.

legendary
Activity: 1176
Merit: 1001
Come on, I followed your loved Hashfast thread. You are not naïve, you know what really matters when it comes to mining!
Fine.

You think that investing 15M thanks to this innovative design you will be able to produce enough starting miners (10PH of them?) so that you will be able to have a competitive edge against the rest of the market. This advantage will be so huge that it will let you reinvest the spent capital on new hardware exponentially, until when you reach 15% of the global hashrate.

Let's test it:
10PH, starting December (I'm being optimistic), at the cost of $5M (since that you will execute everything perfectly), with a power usage of 3MW and a cost/kWh of $0.07 (I'm, again, being optimistic. Iceland is not synonymous for free electricity. It's only a little bit cheaper. Asia is where the cheap eletricity is, and you won't be able to compete with the chinese, that have both cheaper production costs and cheaper electricity so that they can be competitive with a slighter worse J/GH). With the standard 31%/month difficulty increase (the one elaborated from the previous months), you will not recover your investment:
https://tradeblock.com/mining/a/8cca8d1f97

With 25%/month, you would make $200k.
With 15%/month (not gonna happen), you will make $6M before the end of 2015.

Now, let's suppose the last scenario to be the correct one. You make $6M before the end of the year. That's a one year only roi of 160%. That's great, however it's still one order of magnitude off from what you need to catch up with the network and maintain 15% of it.

Easy, isn't it?
full member
Activity: 129
Merit: 100
Instead the mining profit will be constantly reinvested in order to build up the mine to 15% of global rate.
That's nonsense, and you know it. It's a really competitive market now and you won't have enough margin to expand later on. You will be lucky if you recovered the costs of your miners.

That’s the real nonsense!

If it would be impossible to reach break-even and make profit (e.g. to invest in additional/new hardware) with 0.19 J/GH (0.27 J/GH at wall) produced for $0.5/GH or less running in Island, nobody would make profit anymore and in this case nobody would mine and invest in new hardware.

Come on, I followed your loved Hashfast thread. You are not naïve, you know what really matters when it comes to mining!
legendary
Activity: 1176
Merit: 1001
Instead the mining profit will be constantly reinvested in order to build up the mine to 15% of global rate.
That's nonsense, and you know it. It's a really competitive market now and you won't have enough margin to expand later on. You will be lucky if you recovered the costs of your miners.
newbie
Activity: 17
Merit: 0
Hello all,

seems there have been plenty of discussions since Friday. Maybe I can clarify a couple of questions:

1. @rocks: you are completely right. A successful mine does not only require low-energy hardware but also cheap energy. Unfortunately, Germany and most parts of EU are too expensive. But Iceland is a quite efficient spot. Our Gen1 miners are running their and I can tell you: Sebastian saw a number of CoinTerra, KNC and Bitmine miners there, too. There are a few other spots with cheap energy in the world. However, you have to look worldwide.

2. The $15 million dollar can only pay for the very first miners which will not be 15% of the global hashrate. Instead the mining profit will be constantly reinvested in order to build up the mine to 15% of global rate. That's the reason for no dividend payout before year 3. We want to use the money in the build-up phase. We have detailed liquidity plans which we can discuss under NDA.

3. The time horizon. I already said it and I gladly repeat it: we believe in a long-term success of digital hardware and the Bitcoin in special. We want to take part in this process and we are sure there are other persons like us who do not only think of their short-term profit but have faith in the Bitcoin like us and want to be a long-term part of this revolutionary technology. We see it not as something to play and make money with but believe in the future success of this high-end technology.

4. In business you often find pioneers and late followers. The pioneers open doors and make their profit in the extremely volatile ramp-up phase. The late followers often have the better technology and consolidate the market. Yahoo for instance was much earlier in the world than Google.

5. brontosaurus explained the MPW very well. Typically, you get 50 to 100 chips from an MPW run. But the fab produces significantly more wafers than required for backup reasons and in different corners which we bought. Additionally, you can make special arrangements with a little bit more chips per wafer. In fact, we did not have enough chips for 15 fully assembled Gen1 cases. That's why we have 7 PCBs with 28 chips per board and a bit less than 100 PCBs with 20 chips per PCB. Each PCB is individually speed graded haveing chips running between 570 and 800 MHz. So, in all we had about 2000 chips and not more.

Best regards,
Markus
full member
Activity: 129
Merit: 100


In an MPW run you have only a few chips per wafer and, thus, the costs per chip are pretty high and dominate everything. Since the production of a wafer takes 2 months or more, in March or April we wouldn't have been able to get a return of investment with this first gen hardware. We would have needed a full mask tapeout but did not want to waste the money on an old technology when our 2nd generation was nearly final.

Best regards,
Markus

How many gen-1 chip used in the 1.8TH/s machine?
How many 1.8TH/s machines you have made?
What is the die size of gen-1 chip?
How many chips have you get back from your MPW run?
You tape out in August and you have the chip in hands in the early of this year, why not make a full mask? It will get the investment returned if you tape out full mask in the Jan or Feb. Are OP's team smart enough to make the chips but stupid enough cannot calculate an ROI?

Those 1.8TH/s miners built in rack have to be out of a full mask production. However, OP don't have the access to the full mask.

I repeat, this project is a scam.

Dude youre retarded, I urge everyone to ignore this shitposter, he/she is just another armchair expert with a obvious chip (no pun intended) on shoulder.

To clarify their August Gen1 chips have nothing to do with masks, speed, power as advertised in this thread. These were obviously stated as 28nm first gen chips (which is amusing since another company claims to be the first with 28nm implementation), horserider is clearly austistic.

FUCKER FUKT,

Every investment decision should be based on the team you are going to invest. This team is lying about the Gen1 chips. They are lying.



Quite nice troll style for somebody with your forum status! Wink

I think you should do your investigations more carefully, before you call somebody a liar.

Their design service partner is RacyICs, with offices less than 5 km away from the GF 28nm fab in Dresden. According to the RacyICs website they are a GF channel partner and they are offering access to 28nm MPWs to anybody. If you are interested just send them an RFQ about 10 mm² and 3000 samples, then you will see if they make you a quote for this.
Furthermore they have a former GF manager in the Coinbau team. So I have no doubts that they got such an MPW deal for their gen 1.

In general, why should they try to lie? They are looking for serious investors and I guess they prefer big investors bringing more than $250k (unfortunately not my league Sad ). I’m sure that such investors prove any claim they make including their gen 1 story before they sign a contract.
full member
Activity: 129
Merit: 100
Here's the FACTS on MPW.

You buy a 'site' on the wafer and it's cost depends on your chip size. You get charged for a standard sized block, for 28nm it's about $9k per square mm, but you need to buy a minimum size usually in multiples of about 10 mm2. This gets you 100 die.

If your chip is 30 mm2 you need to buy 3 sites. You could then buy a further 3 sites and double up the design, ie get 200 die per wafer. It's expensive to do this for production, but not for a trial run: a full mask set will set you back just under $2m depending on how hard you can negotiate. If you need 1000 chips at 30mm2 then your NRE is roughly 3 x 10 x $9k, $270k for a single site and then you can buy 9 additional wafers, each with 100 die for about $15k each. Your total cost is then $405k.

People do get deals though. Depends on how heavily loaded the MPW runs are.

This aside, I have severe doubts about the viability of their business plan, but that does not mean they should be called scammers.

I also have severe doubts that the sheep who follow the Asicminer banner (and their desire to 'distribute the hashrate' by piling thousands of Petahashes into the  network) are living in the same reality as others and while AM is clearly not a scam, their business proposition is much. much worse than what Coinbau is offering.

Fine, get back 200 die is still a small investment. The real question is, how many chips are been assembled into that rack of machines?  That isa bout three thousands. MPW run will not get back three thousands samples but only tens of it.

200 dies?
The calculation above was already for 3000 dies, which would cost you $405k with an MPW. The shown rack contains 10 machines, with 196 ASICs per machine (7 cards with 28 ASICs each). So they need only 1960 dies for a complete rack.
Any more questions?
donator
Activity: 1120
Merit: 1001
Here's the FACTS on MPW.

You buy a 'site' on the wafer and it's cost depends on your chip size. You get charged for a standard sized block, for 28nm it's about $9k per square mm, but you need to buy a minimum size usually in multiples of about 10 mm2. This gets you 100 die.

If your chip is 30 mm2 you need to buy 3 sites. You could then buy a further 3 sites and double up the design, ie get 200 die per wafer. It's expensive to do this for production, but not for a trial run: a full mask set will set you back just under $2m depending on how hard you can negotiate. If you need 1000 chips at 30mm2 then your NRE is roughly 3 x 10 x $9k, $270k for a single site and then you can buy 9 additional wafers, each with 100 die for about $15k each. Your total cost is then $405k.

People do get deals though. Depends on how heavily loaded the MPW runs are.

This aside, I have severe doubts about the viability of their business plan, but that does not mean they should be called scammers.

I also have severe doubts that the sheep who follow the Asicminer banner (and their desire to 'distribute the hashrate' by piling thousands of Petahashes into the  network) are living in the same reality as others and while AM is clearly not a scam, their business proposition is much. much worse than what Coinbau is offering.

Fine, get back 200 die is still a small investment. The real question is, how many chips are been assembled into that rack of machines?  That isa bout three thousands. MPW run will not get back three thousands samples but only tens of it.
donator
Activity: 1120
Merit: 1001


In an MPW run you have only a few chips per wafer and, thus, the costs per chip are pretty high and dominate everything. Since the production of a wafer takes 2 months or more, in March or April we wouldn't have been able to get a return of investment with this first gen hardware. We would have needed a full mask tapeout but did not want to waste the money on an old technology when our 2nd generation was nearly final.

Best regards,
Markus

How many gen-1 chip used in the 1.8TH/s machine?
How many 1.8TH/s machines you have made?
What is the die size of gen-1 chip?
How many chips have you get back from your MPW run?
You tape out in August and you have the chip in hands in the early of this year, why not make a full mask? It will get the investment returned if you tape out full mask in the Jan or Feb. Are OP's team smart enough to make the chips but stupid enough cannot calculate an ROI?

Those 1.8TH/s miners built in rack have to be out of a full mask production. However, OP don't have the access to the full mask.

I repeat, this project is a scam.

Dude youre retarded, I urge everyone to ignore this shitposter, he/she is just another armchair expert with a obvious chip (no pun intended) on shoulder.

To clarify their August Gen1 chips have nothing to do with masks, speed, power as advertised in this thread. These were obviously stated as 28nm first gen chips (which is amusing since another company claims to be the first with 28nm implementation), horserider is clearly austistic.

FUCKER FUKT,

Every investment decision should be based on the team you are going to invest. This team is lying about the Gen1 chips. They are lying.

hero member
Activity: 714
Merit: 500
I guess nobody would pre-order another miner in the current market, especially if there is no working prototype or previously established trust in the company.

So sorry if you are indeed trying to build a reputable business, but you are most likely too late to the game.
hero member
Activity: 770
Merit: 509
I also have severe doubts that the sheep who follow the Asicminer banner (and their desire to 'distribute the hashrate' by piling thousands of Petahashes into the  network) are living in the same reality as others and while AM is clearly not a scam, their business proposition is much. much worse than what Coinbau is offering.

As a sheep following the asicminer banner I can assure you that the goal is not to "distribute the hashrate". The goal is to dominate the network and crush the competition.

I'm not sure how asicminer is offering a bad business proposition. They already completed their IPO almost 2 years ago and paid it back ~6 times over. I don't think there are any bitcoin investments that even compare.

Coinbau is offering basically the same thing as asicminer besides having a much more halfassed business proposition with a bunch of promises and hardware which has only been seen by the scammers at MAT.

Also asicminer asked for ~$200,000. Coinbau is asking for $15 million...
sr. member
Activity: 441
Merit: 250
Here's the FACTS on MPW.

You buy a 'site' on the wafer and it's cost depends on your chip size. You get charged for a standard sized block, for 28nm it's about $9k per square mm, but you need to buy a minimum size usually in multiples of about 10 mm2. This gets you 100 die.

If your chip is 30 mm2 you need to buy 3 sites. You could then buy a further 3 sites and double up the design, ie get 200 die per wafer. It's expensive to do this for production, but not for a trial run: a full mask set will set you back just under $2m depending on how hard you can negotiate. If you need 1000 chips at 30mm2 then your NRE is roughly 3 x 10 x $9k, $270k for a single site and then you can buy 9 additional wafers, each with 100 die for about $15k each. Your total cost is then $405k.

People do get deals though. Depends on how heavily loaded the MPW runs are.

This aside, I have severe doubts about the viability of their business plan, but that does not mean they should be called scammers.

I also have severe doubts that the sheep who follow the Asicminer banner (and their desire to 'distribute the hashrate' by piling thousands of Petahashes into the  network) are living in the same reality as others and while AM is clearly not a scam, their business proposition is much. much worse than what Coinbau is offering.
full member
Activity: 172
Merit: 100
First payouts after 3 years?
I'm not sure if it's legit or not and if legit then competent enough to deliver but 3 years sounds to me like an epoch.
I have merely scant idea (if any at all) what bitcoin ecosystem will look like in three years from now. So, plain and simple, just stopped to analyse. Good luck.

full member
Activity: 129
Merit: 100


In an MPW run you have only a few chips per wafer and, thus, the costs per chip are pretty high and dominate everything. Since the production of a wafer takes 2 months or more, in March or April we wouldn't have been able to get a return of investment with this first gen hardware. We would have needed a full mask tapeout but did not want to waste the money on an old technology when our 2nd generation was nearly final.

Best regards,
Markus

How many gen-1 chip used in the 1.8TH/s machine?
How many 1.8TH/s machines you have made?
What is the die size of gen-1 chip?
How many chips have you get back from your MPW run?
You tape out in August and you have the chip in hands in the early of this year, why not make a full mask? It will get the investment returned if you tape out full mask in the Jan or Feb. Are OP's team smart enough to make the chips but stupid enough cannot calculate an ROI?

Those 1.8TH/s miners built in rack have to be out of a full mask production. However, OP don't have the access to the full mask.

I repeat, this project is a scam.

Sorry, I didn’t get your point here. What are you trying to say?
They are lying and already had a full mask for their gen1 but were not competent enough to make money with it? Because otherwise they would not need to search for investors to build up >20 PH large scale mine?
Or do you mean that the gen1 chips are not from them?

An ASICrising gen 1 machine contains about 196 gen 1 ASICs. This is public for anybody, just have a look on their website.
For e.g. 15 prototypes they would need 3000 samples. The die size of an 10 GH/s 28nm ASIC should not be larger than 9mm2, otherwise it would be a very bad design. I guess you know that.

I’m not an expert for 28nm prices, but I know from older technologies, that a small MPW contribution including 3000 samples is much cheaper than a full mask set.

Anyway, wasn’t there recently another hardware manufacturer who raised more than $20M to build up a large scale mine? Why did they do that although their gen 1 hardware was very successful?
Pages:
Jump to: