Topic: Is mining gear efficiency a strong reason for difficulty increment? (Read 473 times)

Well if you leapfrog over the 5 and on to the 3nm   lets do some math.

s17+ = 40watts a th 7nm   x 70% =  28 watts a th  x 50% = 14 watts a th  for 3nm gear

cost of double up grade is say 400 million + 1.5 billion = 2 billion  for sake of argument. Lets think 2022   not 2021   what do we have  in 2022  2 years before the 6.125 to 3.0625 ½ ing    so  365 x 2 x 144 = 105,120 x 6.125 = 643,860 reward coins.

from 2022 to 2024  that the 3nm can grab   lets pretend they figure good fees out.  Say 0.875 coins  are fees  this means 735,840 coins

so if a coin is 10,000 we have  6.438 billion to 7.358 billion worth of coins  that  does not justify building the gear. So the question is what happens does tech pressure drive coins up to  30,000 or so   that makes 19 to 22 billion of coins up for grabs.

I like the  info  since efficiency could drop from 40 to 14 watts a th.

 So the s21  could use 3000 watts and mine 210th that  would be a nice piece of gear.

Good stuff. Thanks for sharing. Appreciate it.

This article has a bunch of interesting information.

https://semiengineering.com/5nm-vs-3nm/

They give a design cost range for 7nm ASICs to be from $120 million to $420 million. Their cost to actually manufacture the miners is probably pretty small, paying off that design investment is probably the biggest factor.

Estimates from Samsung and TSMC about efficiency improvements with the upcoming 5nm node are 20% to 30% while design cost will be a lot higher.  They are saying there will be a bigger efficiency jump for 3nm, 50% better than 5nm, but that won't hit the market until 2021, and design price will be huge, $500 million to $1.5 billion.

I think BTC price is going to have to jump up a lot to justify a billion-dollar investment in a single miner. Good news for us miners... makes it a little easier to justify buying some 7nm miners.

Yes, I was fortunate to acquire S17's 56 th and T17's 38 th in decent volume before the price runnup. I'm with you. I'm waiting now. No way I pay these prices. It'll come down. I can wait. I'm willing to buy gear at a 12 month payback window at today's metrics but i won't go beyond that. Truth be told, I got my S17's way cheaper. I've been tempted several times to sell them at 2 and 3 x what I paid for them. but alas, I am a miner. Not a "re-seller".

Low power =  biggest key most of the time.
low setup/building/floor cost = somewhat important

best efficient gear vs best priced gear vs best durable gear is the complicated part.

If bitmain called me on the phone and said they would send me 1000 s17 pros I would take them over any other shipment of gear offered to me.  But  they are not doing that. And at 3000 they are not worth buying to me.  My power deal of 50-50 coin spilt means   I get 26.5 th of each 53th unit.

so 26.5 x 0.00001920 = 0.0005088 x 8200 = $4.17   a day or 3000/4.17 =  720 days  far too long to break even.

Bitmain sold s17 pros under 2000 back in april of 2019   which means they made money at 1600 a unit.
I suspect they can sell them as low as 1200 and turn a good profit.  So buying the most efficient gear  at a really high price does not work well.

Of course Being in the USA  adds in the import tax issues  27.6% is really big   but that is another angle.

The cold hard truth is that in "mining"  just like every other business on the planet now, It's "go big or go home'. The Commercial mines are in a race to the bottom. You have to run way more gear to achieve less margin. Just like Amazon et al. Large Corporations are gobbling up all the profit throughout almost every business on earth. So, cheap power and the most efficient gear possible will be the ONLY thing left. All others will die along the way. Keep in mind that even commercial miners that do not manage their business costs well, are susceptible to extinction. This process will not be a straight line. Factor in the fluctuations in difficulty as it ebbs and flows and there's no way to know how long this process will take, but rest assured it is real and it is happening. I started 3 years ago with 2 L3+ and have worked feverishly to expand knowing this trend is firmly in place. I'm now at 20 PH and expanding as fast as I can. There is no middle ground. It's go big ( and cheap ) or go home. I hate that it's this way but I'm already committed.

Lastly, as a commercial miner I do NOT believe running S9's just because they are cheap and have short pay offs is a good strategy. I am 75-80% certain that these units will be obsolete post Halving. Think L3+. Hard to believe I know but for all intents and purposes, if you want to be a miner, you need the lowest power, lowest floor costs, best gear and lots of it. IMHO.

The talk of mining lightbulbs and all that was always a joke. Miners work on economies of scale, even at an individual machine level. The higher percentage of device cost is ASICs, the more viable it's going to be. The support circuitry to handle a single chip isn't much more than what's required to handle ten chips, so it makes no economic sense to drop single chips with full support circuitry into a dozen different gadgets instead of keeping them in one box at half the cost.

for the home miner the beta a1041 from Canaan is good if it has the beta psu and is set to 30th speed.  Does around 60-65 watts a th and is quieter then an s9.  All the big 3000 plus watt miners are crazy loud.  I guess the s17 pro 53th is okay set to lowest speed. I get about 40 th and use 1440 watts I would say it is better then the s9 for sound.

Hotmine. They are an Ukrainian company.

the russian/Ukrainians are making a 60-65 watt 8th heater. 1000 usd. This is the only true home unit over 5th and under 70 watts a th.  Please forgive me if it is only Russian or only Ukraine builders and not both. Cell phone is hard to look at all the site.

One small point. I think the whole small/home is rapidly evaporating. The only current (e,g, efficient) mining hardware comes in the form of a hot, loud, higher voltage (i.e. not 120V) box. With the exception of sidehack's efforts, there are NO highly efficient miners that exist at less than 25 TH, are there?

Remember when there was talk of "mining light bulbs", "mining routers", or mining "appliances"? Almost sounds funny doesn't it? As Phil said all Bitcoin mining is "industrial" in nature, and I don't see that trend changing in any significant way going forward.

Both mikeywith and wndsnb have good points.

USA fed laws
State laws
City laws

All vary.

New Jersey and Massachusetts are very good.

Both can use mining gear a leverage tool to create a larger solar array.

You can do a mixed gear setup we did this. We now earn about .6 btc and 7 ltc a month.

We will add a s17 pro every once in a while from profits. Which is how we can justify buying them.

I understand the value of getting cheap power with cheap old gear. But sliding in a good piece here and there is okay.  If you are not fully free or have limited free power. Say 10 s9s are using the max power you can do for free.  Eventually you need to get an s17 from the s9 profits. When you do it lower the s9s to a slower speed and add in the hash from the s17.  You increase your hash rate. Wait for more time and add in  another s17.  Do feed all the profits to new s17s hold some of the coin cash some of the coin.

A ratio of

new gear
Held coins
Cashed coins.  Maybe .33+.33+.34=1.0

In the end, making an investment in mining is like any other investment. You are making decisions based on incomplete information about future events.

Charts lie all the time. You could look at that profitability chart for Jan 2017 to Jan 2018, profitability increased by 2.5X. The 1st 6 months of 2019 the profitability doubled. Of course, the chart will have an overall negative bias because gear has been getting more and more efficient.

After the S9 was released in 2016, profitability went down by half, but then was flat for 9 months before increasing all through 2017.

If you are bullish on BTC price, think diff might stall because S9s are getting shut down, and think asic manufacturers might not be able to keep up with demand because foundries are tied up pumping out new 5G components, maybe you could justify the cost.

I hear you phill, but this can only be profitable in a very large scale as far as I can see it, also , is there any reason why this solar array plan wouldn't work with less efficient gears given the fact the mining profit is not key here? is NJ the only state that has such laws?


The whole point of this topic is to prove that mining gear's efficiency is not the main reason for difficulty increment , and the numbers I presented should be  good enough to support that claim.

It's true that charts can't tell you what the profitability per TH is going to be next month, but they sure as hell can tell you that next year is going to be worse than the previous one, charts are not random drawings, they are merely real numbers extracted from real data, it's clear as day that profitability will keep declining despite any other future events.

That is because you are looking at the mining as a mining setup.
It is not it is a tool to create a 1300000 million solar array. In NJ , USA I get to use 300,000 in mining gear to create a solar array worth 4x the price of the mining gear. Which is rolled into the price of the array and gets tax credits because it was what created the power demand in that site.

I know that what it earns in two years is really not anywhere as important as the solar array which would not be put into play without the mining gear.

I used 448k as earnings for sake of simplicity.

Also this is a very specific example of how over paying for gear creates a good deal (large solar array)

A lot of companies are doing clever moves with gear and making money outside the box of using the gear to mine with.

I understand the numbers, but there are two major problems in your analysis

The first one is that this only applies to large scale mining only, what about the rest of the small / mid size miners?

The second issues is that, the text quoted below is pure speculation which is unlikely to come true


By looking again at the profitably chart, there is 0% chance you are going to net  448,000$ in two years, below is a chart showing the past 2 years profitability in a simple moving average of 30 days



chart > https://bitinfocharts.com/comparison/mining_profitability-btc-sma30.html#log&2y

when you place the arrow at any point, the number you see the average of the previous 30 days ( to avoid day to day variances)

1st-Sep 2017 - 30th-Sep 2018:
average (take from 1st day of each month) : 1.1$

Sep 2018 - Sep 2019:
average (take from 1st day of each month) : 0.22$

Sep 2019- Sep 2020 :
average (take from 1st day of each month) : ??

Sep 2020- Sep 2021 :
average (take from 1st day of each month) : ??

is it safe to assume that we are going to lose 50% every year?

so


becomes this ?

4ph  224,256  112,128$ first year, second year 56,064$  in 2 years  :  448,000 = 168,192$

However during that period , bitcoin price went from 4000$ to 8000$ , so 300,000$ invested in bitcoin now might very well be 600,000.

of course all these numbers are pure speculation, but that charts don't lie, profitability has way more chances of going down than up.

Here goes  Lakewood City in NJ has special economic incentives.  So sales taxes = 3% not 7%  300,000 in inprovements  costs 309,000 not 321,000
NJ has special tax incentives for solar investment  1,600,000 investment in solar  will get back 800,000 in incentives.  2019 is last year for this plan
USA feds has 30% tax credit so 1,600,000  will get back   480,000

so  800,000 +   480,000 = 1,280,000 in credits.  so 1,600,000- 1,280,000  = 320,000 real cost for setup
An empty warehouse  using little or no power  but has a 3 acre roof  has room for 1500 x 400 watt panels  that is 600,000 watts  cost is  1,300,000 but power company does not allow you to use it  since you are not pulling the power.   Buy 100 s17pros now for 300,000

it creates a 100 x 2225 = 222,500 power demand
roll that into the  1,300,000 in panels  = 1,600,000   oct  nov dec power bills allow you around 36,000 in power cost
1,636,000 is cost of solar plus gear  incentives bring you to 356,000  real cost.  and you have roughly   600,000/5 =  120,000 watts 24/7/365

run the s17 pros on low power  144000 watts 24/7/365
this means  you are getting  100 x x 40t =4ph for 24kwatts a day 24/7/365     for a net cost of 16,588 in yearly power past what you generate.
and 356,000 real cost out of pocket.

4ph earns 614 usd a day.  or 224,256 a year.   in 2 years  you have  448,000 back.
and a fully paid off 600,000 watt solar array which is much more valuable then the miners.

14 lines.  of what you need to make buying the s17s at 3000 right now worth it.

20 lines will suffice i hope   , i am eager to see how actually losing money is a good business model in some countries/states, till then , enjoy the party.

There are so many varying taxes laws country to country state to state city to city  that  I could write 20 pages on justifying why  a guy should buy  100000 in s17s at 3000 each.  I am running to a party in 10 minutes I will get back to this and give and example of how laws in Lakewood NJ USA would work for a 3,000,000 build using 100,000 in s17's.

ok so now we agree that efficiency is not the most important factor as far as difficulty is concerned,  after all 100% improvement over the course of 4 years and millions of dollars spend on research is not impressive to say the least.


That is the main point, in fact as time passes the pressure gets down to those  with cheaper power, at one point only those with 0-2 cent might really actually make money , I am not talking about making profit on daily bases, I am talking about making more profit that simply hodling btc or keeping your money at bank for 0.25% interest rate or whatever that is where you live.

people have this misconception of profitably not changing a lot because price moves difficulty, thus any investment at any given time will yield the same profit in average, which is not the case at all.

let's zoom-out and have a look at these 3 charts that represent data of  nearly 9 years starting July 2010 and ending today.

1- Price



2-Difficulty



3-Mining Profitability



Theoretically,Mining Profitability should stay flat, or at worst case scenario, should decrease slowly given that the cost of buying 1TH is decreasing as well.

in fact here is a funny story, mining profitability now is nearly exactly the same as it was back in Jan 2019 when price was hanging at about 3k , with most efficient gears are due to ship later this year , it is safe to assume that the majority of extra 50E addition in hashrate since then, did not come only from the new efficient gears, so what happened?

My take on this that people with cheap power (2 cents or so )as phill explained are still filling in those mining farms with old gears such as S9 or even T9, in fact a T9 that goes for as cheap as a 100$ makes almost 30$ a month at 2 cents kw so there is really no reason why they wouldn't be adding more of those.

So where do the 4-7 cents per kw miners stand in all this mess? a T9 at 4 cents makes only 7.5$ a month and loses at 5 cents , S9 does not look much better, so ultimately the 4-7 cents people have no option but to pay premium for a gear that is only double efficient.

now let's examine the situation and see how it goes for buying say S17 pro today for 3500$ ( the most efficient gear), and based on the profitability chart it's pretty obvious that we are only going down, despite the spikes here and there the average trend is down, we are at 18 cents per th today, the numbers say we are going much lower, but let's assume we stay flat for the sake of it.

S17 pro makes 5.54$ per day at 5 cents  that is 166$ per month , so you are looking at 21 months to ROI and hey , that is probably the BEST scenario you are going to have, and you are actually paying 0.45 bitcoin for that gear , it's almost nearly impossible to mine that much before either the miner dies or becomes unprofitable and you sell it for 200-300$ to those guys who have 2 cents per kwh.

I am kinda mixing a few topics at once, but it's really all related, since people who pay more for power count totally on efficiency, efficiency does not move nearly 1/10 as fast as difficulty does, I just don't see the logic here, can somebody who paid a fortune for these gears and has 5-7 cent per kwh rate explain the logic behind their move?

Think a 80,000 foot building wired and hooked up to run 12 megawatts  all done back in 2017.  before sept of 2017  and this guy made serious coin on the sept-dec runup.

let him have  3 cent power.

Last spring  he jumped on s17's which were  well under 1800   usd he put them in  filled the 12 megawatts  for only  1800 x 5000 = 9,000,000

he now has 1800 x 53 = 95,400 th or  95.4ph  running since april or may.  I do not know hwo this guy is but I am certain  there are setups like this.

They are far ahead of the curve.

That is a good point. Back in 2016 there were not nearly as many huge mines already running. Now we've got 50ET of S9s already running and the only cost for those industrial miners to upgrade is new gear price and a bit of labor.

Worldwide mining is more and more industrial.

Worldwide power is about 2-5 cents a kwatts for good industrial rates.

So base  big farms  on 2-5 cent numbers.

Add in building
add in wires
add in internet.
add in  cooling
add in cost of gear.

say  that is 2 cents for big guys.

so  4-7 cents  is what you need to make on  your gear per kwatt.

there is always pressure to the 4-7 cent earning range. 

And of course  on the other side  speculators traders ya da ya da ya da.

plus  less coins enter the market.

In  dec 2017  diff was 1.4 and coins were 20000    best gear = 100 watts a th
in sept 2019  diff is 12.7 and coins are 8000         best gear = 37 watts a th

If I am an established big miner

I have 1 new cost that would be new gear.

my building = done
my wires = done

so if I get 3 cent power on   the s17 pro  it cost   2.225 x 24 = 54 kwatts x 3 cent = 1.62 a day  it makes 53 x  0.00001920 = 0.0010176 btc or 8.24 -1.62 =

6.62 a day  this guy  drags  the profit down for anyone that can't match him.

his building is in place
his cooling is in place
his wires are setup.

If my small farm in clifton was all s17 pro  I would have 2.5ph vs 1.0ph.  I make money with  my 1.0ph  but I would make 2.5x  what I make now.

These whale farms drag price down.  At least my guess is that.

I think it's more related to popularity and the perceived risk of investing in mining. Back in 2016, I think it was a much harder sell than today to get investors to put big $ into large industrial farms. Then the run up to 20K happened in 2017, fees were out of control, and miners were making a killing. It got a lot of news coverage, and more legitimacy.

The fact that S9s are still profitable for 5 cent power after 3 years also would help convince investors to spend on new gear even with longer ROI time.

As for the price of gear right now, asic manufacturers currently can't get as much chip fab capacity as they'd like so it is limiting the amount of new hardware available. Limited supply = higher prices. If they had all the chip manufacturing capacity they wanted, prices would be lower, and diff would be rising even faster.

Sorry can't give a great reply but the s9 at the time was still like a 100w/th miner or worse I believe. Didn't it only get down to these levels after a firmware release in the last year or so. This is also an interesting factor, imagine where Diff would have been had Bitmain not hoarded it to themselves for so long.

Personally i don't think it's the efficiency alone.  It's more what the efficiency and experience have allowed them to do. My largest most compact miner is 65TH. That used to take 4 machines to achieve; if they all fired 100%For a non commercial purchaser. If you then add in the adoption and interest in mining since then production levels have skyrocketed which has more impact than the other 2 combined.

I was checking mining difficulty and saw it hanging at about 12.75t , which is a 100% increment from the 6.5t mark last May, Bitcoin price was trading at 4600$ then and now it's at 8200$ so that is also nearly double as well, although technically , price moves way faster the difficulty (in both directions), by going through the different subjects on the forum and else where,and reading what people have to say about it , i think that most people think that new gen miners are the main cause for this non-stop difficulty spikes.

That makes sense, but I have a huge problem with the way the numbers match, if we go back in time and check this very date May 31, 2016, which was the release date for Antminer S9.

At that time we had :

Bitcoin price             = 436$
Difficulty                  = 0.2t
Best Gear efficiency  = 85w/th
Price per TH             =150$
Profit per TH            =1.2$ / day
1000$ investment     =280$ return per month

Currently we have :

Bitcoin price             = 8200$      >  18x UP
Difficulty                  = 12.74t      >  63x UP
Best Gear efficiency  = 45w/th     >  2x   UP
Price per TH             =55$           > 3x   Down
Profit per TH            =0.18$ /day > 6x Down
1000$ investment    =90$ return per month

The strange thing about these numbers that since May 2016 (4 years) mining efficiency has only doubled from S9 to the most effiecnt gear of today S17e/S17pro etc., while diff is 63 times higher,  I don't see how mining efficiency matters here ! anyone cares to explain ??

in total ,mining now is 3 times less profitable than it was 4 years ago despite the fact that efficiency has only doubled once, this indicates that people are willing to earn peanuts from mining, and with these numbers in hand it is safe to assume that probably the majority of miners will end up losing money in the long run if they kept buying gears at high prices.

What do you think the minimum profit per th are we ever going to get to?