Energy consumption will become an issue if bitcoin really breaks through

[brenzi]

Let's say that Bitcoin takes the place of the USD within 10 years and see what the energy consumption of the bitcoin network will be.

facts:
 - USD "M2" money supply in 2009: 8E12 USD (http://money.howstuffworks.com/how-much-money-is-in-the-world.htm)
 - in 2023 there will be 19E6 BTC. Mining will be honoured with 6.25 BTC/10min => 329'000 BTC/y (https://en.bitcoin.it/wiki/Controlled_supply)
 
assumptions:
 - BTC takes the place of USD => one BTC will be worth 421'053$ (of course valuation of BTC will devaluate USD...but let's leave that aside.)
 - let's say 1kWh costs 10cts USD
 - all miners use the same recent technology. Or at least with equal efficiency in MH/J
 - people are not paying significant transaction fees by 2023 (the more people will pay fees, the higher the resulting profitable global energy consumption)

(edit) In a free market the following formula approximates energy consumption of any PoW based cryptocurrency:
GlobalEnergyConsumptionForMining ~= (MiningReward + TransactionFees) * bitcoinValueInUSD / EnergyCostInUSDperkWh

So mining would be profitable somewhere below an energy consumption of the bitcoin network of:

break even price: 421'053$/BTC*329'000BTC/y = 139E9$/y
equivalent energy: 1.39E12 kWh/y

Today's global electricity consumption is around 20E12 kWh/y (http://en.wikipedia.org/wiki/Electric_energy_consumption)

This means that the bitcoin network by itself would raise global electricity consumption by up to 7%

Please show me where I'm wrong!

We could of course put this less dramatic:
  - in 2033 it would be 174E9 kWh/y
  - in 2100 it would be 2.65 GWh/y

today we would be at 5MWh if all miners were using newest ASIC's (https://en.bitcoin.it/wiki/Mining_hardware_comparison)

(edit)
CONCLUSIONS:
 

• according to economical basics, energy consumption of the bitcoin network does not depend on mining gear efficiency!
• the problem described applies to every proof-of-work based currency (including Litecoin) because proof-of-work equals proof-of-energy-consumption at a market equilibrium
• bitcoin value is NOT directly backed by energy consumption. But mining rewards and transaction fees are.
• the slower the bitcoin value rises and the lower the transaction fees, the lower the energy consumption of the bitcoin network on the long run

[naphto]

This means that the bitcoin network by itself would raise global electricity consumption by up to 7%

Please show me where I'm wrong!

You are wrongly assuming that bitcoins will still be used then.
People will switch to LTC, or any other altcoins easier to mine ...

[RenegadeMind]

Your numbers are way off.

Energy consumption is massively impacted by the size of the semiconductor die, e.g. 110 nm vs. 50 nm vs. 20 nm. Currently, most ASIC rigs are using 110+ nm chips, but the currently available semiconductor techology available to OEMs is in the 2x-nm class. The 1x-nm class isn't far behind now.

Have a read around here:

http://www.samsung.com/global/business/semiconductor/minisite/Greenmemory/

That will give you an idea about how die size affect energy consumption.

With ever smaller classes of chips, energy consumption will decrease. So, any prediction of future energy consumption must take into account the energy efficiency of semiconductor technologies available in the future.

For example, would you calculate it in cords of wood or tonnes of coal for a steam engine to produce your electricity? Of course not, because that technology is pretty much obsolete now (outside of a few specialty cases).

[brenzi]

Energy consumption is massively impacted by the size of the semiconductor die, e.g. 110 nm vs. 50 nm vs. 20 nm. Currently, most ASIC rigs are using 110+ nm chips, but the currently available semiconductor techology available to OEMs is in the 2x-nm class. The 1x-nm class isn't far behind now.

I'm an electrical engineer, so I know about the technological constraints you are talking about. But you are wrongly assuming that the efficiency in MH/J affects overall energy consumption. It does not (at least to my understanding of a free market). More efficient HW will just make the Hashrate go up what in turn will make difficulty level go up.
As long as additional gear can be run profitably, this will be done.

[odolvlobo]

Your numbers are way off.

Energy consumption is massively impacted by the size of the semiconductor die, e.g. 110 nm vs. 50 nm vs. 20 nm. Currently, most ASIC rigs are using 110+ nm chips, but the currently available semiconductor techology available to OEMs is in the 2x-nm class. The 1x-nm class isn't far behind now.

Have a read around here:

http://www.samsung.com/global/business/semiconductor/minisite/Greenmemory/

That will give you an idea about how die size affect energy consumption.

With ever smaller classes of chips, energy consumption will decrease. So, any prediction of future energy consumption must take into account the energy efficiency of semiconductor technologies available in the future.

For example, would you calculate it in cords of wood or tonnes of coal for a steam engine to produce your electricity? Of course not, because that technology is pretty much obsolete now (outside of a few specialty cases).

The power efficiency of the mining equipment is not as relevant as you think. Mining is a nearly perfectly competitive market. Miners will increase the amount of mining (and therefore the amount of energy consumed) until their profits are reduced to some tolerable level (below that they stop mining).

The relevant numbers are:
1. The number of bitcoins mined (from the reward and the transaction fee) and their value.
2. The cost of the equipment.
3. The cost of the power.
4. The operating margin of the miners.

The cost of mining will rise to the value of the mined bitcoins minus the margin. The cost of mining is the cost of the equipment plus the cost of the energy. So, the amount of energy spent on mining is the value of the mined bitcoins minus the margin and minus the cost of the equipment.

Also, as the cost of the equipment drops (assuming that it does), energy cost will become the dominant limit, regardless of the power efficiency of the equipment.

[FTWbitcoinFTW]

Gavin got some interesting idea for future.

New bitcoins are generated, or "mined," when computers succeed at solving increasingly complex equations. Bloomberg recently described this mining process as an "environmental disaster" because of the energy required to power the machines working on the problems.

The bitcoin mining process incentivizes people to be as efficient as possible and use as little power as possible to create bitcoins and to validate the transactions. The more efficient you are, less you spend on electricity and the more profitable you’ll be. In the future, I expect to see bitcoin mining in places where electricity is free or cheap. You could put solar array in the Arizona desert attached to bitcoin miners and instead of trying to ship that electricity all over world, you could ship Bitcoin all over the world. The output of bitcoin mining is heat. You’ll see bitcoin mining happening in places where people need heat anyway. I could imagine bitcoin heaters that, in addition to generating heat, generate bitcoin.]

http://www.huffingtonpost.com/2013/04/16/gavin-andresen-bitcoin_n_3093316.html

I want a bitcoin heater !

[johnyj]

It is possible that in the future, a handful of ASIC companies developed extremly power efficient mining rigs with leading <10nm technology to mine and raise the difficulty, thus majority of miners would mine at a loss and quit the mining, so the total power consumption will be less

The worst case is everyone in the world will have same technology and mine, the electricity usage will be huge

But then there is a limit on circuit breaks in each household

[wobber]

Internet power consumption is how much % of total world? 10%? Or 15%?

Please consider that many phone carriers, TV, mobile also use internet for data transport.

[RenegadeMind]

Energy consumption is massively impacted by the size of the semiconductor die, e.g. 110 nm vs. 50 nm vs. 20 nm. Currently, most ASIC rigs are using 110+ nm chips, but the currently available semiconductor techology available to OEMs is in the 2x-nm class. The 1x-nm class isn't far behind now.

I'm an electrical engineer, so I know about the technological constraints you are talking about. But you are wrongly assuming that the efficiency in MH/J affects overall energy consumption. It does not (at least to my understanding of a free market). More efficient HW will just make the Hashrate go up what in turn will make difficulty level go up.
As long as additional gear can be run profitably, this will be done.

Good point.

However:

We could of course put this less dramatic:
  - in 2033 it would be 174E9 kWh/y
  - in 2100 it would be 2.65 GWh/y

today we would be at 5MWh if all miners were using newest ASIC's (https://en.bitcoin.it/wiki/Mining_hardware_comparison)

It then seems that X years down the road, mining will only be done by massive supercomputers. Or that seems to be the implication to me anyways. Could be wrong, but if the hashrate an difficulty go up like that, we should see ASICs pushed out as being unprofitable the same way that CPU mining has been pushed out, and the way that GPUs older than <insert # of months/years here> are not profitable now.

The question then seems to be about the initial cost of hardware? i.e. Are we assuming that the limiting factor will be electricity when the limiting factor in the future may be the cost of hardware?

[brenzi]

It then seems that X years down the road, mining will only be done by massive supercomputers. Or that seems to be the implication to me anyways. Could be wrong, but if the hashrate an difficulty go up like that, we should see ASICs pushed out as being unprofitable the same way that CPU mining has been pushed out, and the way that GPUs older than <insert # of months/years here> are not profitable now.

The question then seems to be about the initial cost of hardware? i.e. Are we assuming that the limiting factor will be electricity when the limiting factor in the future may be the cost of hardware?

Because of the scaling factor I don't think that HW cost will dominate. There will be such a huge market for mining gear that prices will drop. However: If a producer of top-notch mining gear can earn more money by mining than by selling his gear, the game could change. The bitcoin network would then become dominated by semiconductor-fabs. And knowing that such fabs are incredibly expensive to run this unfortunately means that bitcoin will be dominated by a handful of players (The end of all that's good about bitcoin).

One more thing: We didn't so far talk about the grey energy that needs to be spent before mining gear even starts to mine. That might at one point become important as life-cycles may be expected to be quite short for mining gear.

[mestar]

The question then seems to be about the initial cost of hardware? i.e. Are we assuming that the limiting factor will be electricity when the limiting factor in the future may be the cost of hardware?

Cost of hardware influences your decision do you buy new mining hardware or not.  

Once you have the hardware, the decision when you turn it off depends only on running costs.  So, yes, the electricity will always be the limiting factor. There will always be some miners that run at the edge of profitability.

[johnyj]

And actually this is good, it pushes more people to research for the more efficient energy resource

[Lethn]

I'd agree with johnyj and the great thing is we can fund all sorts of crazy ideas with Bitcoin it's not like our governments could freeze our accounts anymore like in the old days trolololollol But seriously guys, I think bio-engineered carbon dioxide eating bacteria is the answer.

[benjamindees]

You could put solar array in the Arizona desert attached to bitcoin miners and instead of trying to ship that electricity all over world, you could ship Bitcoin all over the world. The output of bitcoin mining is heat. You’ll see bitcoin mining happening in places where people need heat anyway. I could imagine bitcoin heaters that, in addition to generating heat, generate bitcoin.]

http://www.huffingtonpost.com/2013/04/16/gavin-andresen-bitcoin_n_3093316.html

I want a bitcoin heater !

If you look closely, this is already happening.  Prices on BTC-E (Russian exchange) are consistently lower than elsewhere.  There are likely more miners, making use of obsolete mining equipment for supplemental heat, in colder areas.

In the next few months, there will be incentive for mining to shift to the southern hemisphere.  (We could use an exchange in Australia or South America to take advantage of this.)

But one thing about this concept, is that it is still constrained by fiat currencies.  In order to take advantage of these arbitrage opportunities, to bring cheap Bitcoins produced in colder areas to the market, we must have easy access to either a common, global exchange or another currency that can be easily transferred and used to purchase them.

[Birdy]

You also need to substract the amount of energy, fuel and other ressources that are used for our current currncy system,

[brenzi]

And actually this is good, it pushes more people to research for the more efficient energy resource

Unfortunately, this won't help at all. Let's say that we find a supercheap possibility to produce electricity a 1 cts $ /kWh. This only multiplies the profitable energy consumption of the bitcoin network. It's the proof-of-energy-consumption principle that is the problem in the first place.

We could even convert the whole solar irradiation on earth into electricity with an efficiency of 100% (we never will..of course). The bitcoin network would still burn a significant part of it, given that its market cap is significant on global scale.

[brenzi]

Internet power consumption is how much % of total world? 10%? Or 15%?

Please consider that many phone carriers, TV, mobile also use internet for data transport.

As the bitcoin revolution would be historically comparable to the internet revolution your point is valid. But it then comes back to the benefit for society that we expect from bitcoin. I'm not yet convinced, that the majority of the world's polulation will have a better life thanks to bitcoin breaking through. But that's a different topic.

[Birdy]

And actually this is good, it pushes more people to research for the more efficient energy resource

Unfortunately, this won't help at all. Let's say that we find a supercheap possibility to produce electricity a 1 cts $ /kWh. This only multiplies the profitable energy consumption of the bitcoin network. It's the proof-of-energy-consumption principle that is the problem in the first place.

Yes, brenzi is right. If energy is cheaper, more will be used.
It happened in a lot of other technology,too.

[benjamindees]

Let's say that we find a supercheap possibility to produce electricity a 1 cts $ /kWh. This only multiplies the profitable energy consumption of the bitcoin network. It's the proof-of-energy-consumption principle that is the problem in the first place.

We could even convert the whole solar irradiation on earth into electricity with an efficiency of 100% (we never will..of course). The bitcoin network would still burn a significant part of it, given that its market cap is significant on global scale.

You're right.  And I haven't checked your figures.  But if you think that 7% is "significant", then we'll just have to agree to disagree.  As pointed out, this is well within the range of energy already used for heat in colder areas.

[brenzi]

You're right.  And I haven't checked your figures.  But if you think that 7% is "significant", then we'll just have to agree to disagree.  As pointed out, this is well within the range of energy already used for heat in colder areas.

Heating with electricity is a stupid option in the environmental sense because each Joule of electrically produced heat causes even more primary energy consumption at the power plant because of low efficiency. Of course, a geothermal heat pump is a different case (But then you won't be mining at the same time anyway). So Gavin's "bitcoin heater" to me just makes things a little less worse.

And how to use your mining heat in summer anyway?