Bitcoin energy usage for proof of work is proportional to bitcoin price

In this blog I'm going to clear up a few disturbingly common misconceptions about bitcoin mining, and give you a good high-level idea of how bitcoin and its proof of work scheme work. Please read on if you are not aware that if bitcoin prices go up ten times again (that'd be $200000 as some already predicting) then bitcoin energy consumption will tend to go up about ten times as well, which is several percent of the world's electricity production . I'll also explain why.

More specifically, the cost of consumed electricity tends to be a major part of the bitcoin block reward, which is proportional the bitcoin price up to the havings, and depends very little on technological advances as long as proof of work is used.

Introduction

I like cryptocurrencies and admire the ingenious inventions that went into Bitcoin. BUT: as ingenious as its "proof of work" scheme was - it is solely responsible for the horrendous energy usage used up in bitcoin "mining" (as of 10/2023 somewhere in the area of 0.4% of world electricity production ) , and as Ethereum's move to "proof of stake" shows, there are other ways to reach the same goals. In this blog I'll give you an idea about how bitcoin its "proof of work" (PoW) work and discuss that the energy and resource consumption of bitcoin mining is proportional to the mining reward, which is proportional to its price (up to the relatively rare halvings). For Bitcoin that means we will get an upper limit to its price - if it goes up ten times again, politics just has to intervene because the ressource usage goes from (IMHO) entirely mad to seriously dangerous.

I somewhat suspect that many think that the humongous resources spent on bitcoin mining are somehow needed to manage the bitcoin chain. They are not. In fact, that's arguably the planets most wasteful lottery and its sole purpose is to ensure that nobody has the sole power over the bitcoin chain. And since there are alternatives without that humongous energy usage, I say proof of work just has to be replaced by one of the existing alternatives or a new system - the sooner the better.

TL;DR: My main point is that the energy and resource usage of Bitcoin is always so huge that it consumes a large fraction of the bitcoin block reward, and thus will increase to even more dangerously high values if it get's adopted more, and is not absolutely needed, too. Read the sections about the problems of PoW and the economics of PoW for that.

If you are already a bitcoin expert

My main points are the following. If you mostly agree with all of them, you can safely ignore this blog. If you disagree with any of them or don't understand what I'm talking about, I'd be happy if you continue reading and tell me your opinion.

In a steady state, the energy and resource usage of Bitcoin is always so large that it consumes a large fraction of the bitcoin block reward. Consequence: if the bitcoin price goes up 10 times, the energy usage will go up 10 times, too (up to the halving), and thus reach somewhere at 1% of world energy production.
Politics just has to intervene if that happens, and there are ways: e.g. a US + EU ban on commercial usage of Bitcoin will make the price come down again.
Changes in the energy efficiency of mining machines do hardly change the energy usage, because the difficulty will just go up. (Make a thought experiment what'd happen if a miracle ASIC with a million times the hash rate was created - you'll find no significant change in the energy usage after it gets adopted by all miners.)

The structure of the Bitcoin blockchain (simplified)

Bitcoin is basically a seriously ingenious use of public key cryptography in heavy doses. Each Bitcoin address that can hold bitcoins is a public key, and (hopefully :-) only the owner has its private key. Simplified: when spending bitcoin, you generate a transaction that basically says:

"transfer X bitcoin from address A to address B and pay transaction fee Y, 
signed by the private key of yours, truly, A's owner".

All transactions are recorded in a blockchain. Each block collects new transactions, which are distributed in a peer to peer network, where everybody can send transactions and receive all transactions that other people signed. A block also says what the previous block was, so that all blocks form a chain back to an initial block, and contains a special transaction that assigns the transaction fees of all transactions in it to the bitcoin creator, as well as a block reward of currently 6.25 bitcoin. The block reward is halved every 210.000 blocks and is the sole source of newly created bitcoins - the sum of all block rewards will eventually add up to approx. 21,000,000 bitcoins. Also are checked for validity, e.g. that nobody spends bitcoins that don't belong to him/her, and will be ignored by other miners if they are not, so the block reward would be lost.

Now that can't be everything, because everybody and his grandmother would try to create blocks as fast as possible to get those block rewards and mining fees, right? That's where "proof of work" comes into play.

How does "proof of work" work?

Traditionally, you'd have a central server that does such validity checking and approves transactions. But the very idea of Bitcoin was to create a system without a central server so that no single entity can control bitcoin, shut it down, perform fraud or exclude people from access. So, block creation should be distributed over everybody who wants to create blocks and is able to do so. Still, you want to ensure that blocks are generated at a reasonable rate - in Bitcoins case after about 10 minutes on average.

The ingenious (but now troublesome) idea that does all that was to require block creators to do additional computational work when attempting to create a valid block by introducing a random number (nonce) into the block header and impose a condition that is really really REALLY hard to meet, and can only be met by serially trying out nonces. (If you are technical: the condition is that the SHA-256 hash of the block header must be smaller than a value that depends on the current "Bitcoin difficulty" - that's the "hashes" you hear about.) That idea works smoothly because it ensures that a random one of those servers will make the block, thus distributing responsibilities over many shoulders and enable all that good properties I mentioned above. If several servers find a block at the same time, the longest ensuing block chain will win, so that's resolved after at very few blocks.

Unfortunately it's rather necessary for this kind of resolution that the power used in POW is quite huge, because somebody who could control more than 50% of the hashing power could run a number of attacs. One of the worst one might be a double spend attack, where he/she spends a large sum of bitcoins and waits a while until the transaction is accepted by the network and the receiver of the payment, and then creates a longer chain that doesn't contain that transaction, which will replace the old chain and thus make the transaction invalid (and might create havoc for a lot of other transactions, too).

It's important to know that the difficulty of that task of finding that golden random number is periodically managed so that blocks are generated on average at that one per 10 minutes rate, independently of the number and power of servers.

The problem with "proof of work"

The effect of that is that there are now probably millions of servers that sit on a almost ready made block, all of them having done the actual work of collecting and validating transactions necessary for managing the block chain, and are now desperately trying to find that one golden nonce that will make the block valid and give them to right to claim that block reward. That's called (IMHO somewhat misleadingly) called "Bitcoin mining" - it's rather like a lottery who get's the block reward, for which you pay by crunching random numbers.

Now, get that: at the time of writing the number of SHA-256 hashes that need to be computed for somebody to win that lottery is about
270 000 000 000 000 000 000 000 hashes every 10 minutes,
and a hash is in itself a not quite inexpensive operation. The energy used for this can easily serve a medium sized country, and then there are the millions of servers used for that are built and replaced once in a while, in a time where chips are scarce and sought for other things like training neural networks for artificial intelligence, and of course there is climate change.

Please notice that that hashing for proof of work takes 99.9999999999...% of the effort in creating a new block, and is not inherently needed if you find some other way to ensure distribution of block creation - and there are already quite a lot of suggested alternatives . One of them alreay used to manage the second largest crypto currency, Ethereum. Don't get me wrong - it wouldn't be a no-brainer to replace that, but I think it's important to understand the situation and I think it'd be worth a very large and very determined effort to replace PoW in Bitcoin.

The economics of "proof of work"

The important point to keep im mind here is that there is a fixed block reward, which is randomly and therefore about equally distributed among the miners in proportion to their hash rate. This means that every miner has an incentive to raise his hash rate by adding additional machines / better machines to get more of the cake. This incentive only stops when the costs get near the block reward. It is also known that bitcoin mining is a highly competitive business with low margins that can be entered quite easily and where the energy costs are known to be the largest cost factor. So the tendency is that a large part of the block reward will be spent on energy.

The consequence of that is that e.g. when the block reward is raised by a factor of 10 (by raising the bitcoin price, which is quite possible) the energy usage will tend to raise by a factor of 10, too!

Now, I'm aware that this is just a rough estimation and can easily be off by a factor of 2 or more. But I'm investigating the general tendency here, independently of short time factors like chip shortages or bitcoin price drops. The bitcoin price already went up by several orders of magnitude and could easily go up by another order of magnitude or two. At the time of speaking, the bitcoin market capitalization is 500 billion US$, which means everybody on earth owns on average $60 in bitcoin - either directly or indirectly through funds etc. (Compare: for gold, it's about $1500 per person ). And the only way that could raise is through a corresponding raise in bitcoin price as the supply doesn't change much anymore. To get a ballpark number of the energy usage of bitcoin mining has been estimated to something like 0.16% or 2021 (with higher bitcoin price) even 0.55% of the world electrical energy production. Let's compare that to our own estimation from what we have learned. Assuming that a third of the bitcoin mining reward is spent on electricity (leaving room for a ROI of up to 200% for the miners): the reward is currently 6.25 bitcoin per 10 minutes (plus transaction fees of up to 1 bitcoin per block, which we will ignore), which is about 330000 bitcoin per year, at the current price of $26000 per bitcoin this is something like 8.6 billion US$ per year. At a price of $0.1 per kWh this could buy about 29 billion kWh = 29 TWh. The global production of electric energy is somewhere around 25000 TWh , so this would be 0.12% of that. Bingo. Or, as I'd probably should better say: OUCH!

I'll leave it to you to judge whether that much energy usage is appropriate for a kind of reserve currency that isn't even in much active use, especially when there are alternatives around or in development that don't use much energy. Compare Ethereum's energy usage: when it was using PoW it was about 78 TWh/year, after moving to PoS it's about at 0.0026 TWh/year.

Quantitative proof

Since there is a bit of involved argumentation here, I'd like to turn the calculation a bit around into a kind of quantitative proof. Consider this:

Since 2011 the hash efficiency has increased by a factor of more than 1000 (compare table 2 here ).
Since Bitcoin mining has become commercialised, the electricity costs are always less than the mining reward, since that's what pays for the mining.
Today the electricity costs are at least 50% of the mining reward (see calculation below).

Thus, the ratio of electricity cost to bitcoin mining reward cannot have come down changed much, if at all. Some argue that the efficiency of mining technology can reduce the energy usage. But if that was true - then where did that huge efficiency improvement go? (That is a rhetorical question, of course - it went into the automatic difficulty adjustment).

Please notice also that this perfectly fits my claim that most of the block reward is normally spent on energy.

Here is the calculation for point 3: The Cambridge Bitcoin Electricity Consumption Index estimates the Bitcoin energy usage to an absolute minimum of 9.5GW, and the average electricity price to 0.05 USD/kWh. The bitcoin reward is 6.25 BTC per 10 minutes, that is 37.5 BTC/h = 937500 US-$ per hour. The electricity spent on bitcoin would cost 9.5GW * 0.05 USD/kWh = 475000 US-$, that is about 50% of the mining reward.

Arguments I don't agree with

Here are some arguments I often hear but that I think are not valid.

"The use of regenerative energies in Bitcoin mining solves the problem."
Well, that reduces the damage somewhat, but we are still spending the energy on something that can be done other ways and the energy and machines could be used on other things that can be done anywhere anytime, such as training AI models. So I see that as green washing. (There are some surprising energy sources used for bitcoin mining, but please notice that pretty much all of the same sources can be used for e.g. training AI models, too, which arguably produce some more lasting value.)
"Technological advances in mining reduce, such as better ASICs with better hash rate, reduce the energy usage.
As explained: the energy and resource usage depends on the block rewards, not the technology. Thought experiment: imagine a miracle ASIC with million times the hash rate of current ASICs would be created. What would happen is that those using it get a huge temporary advantage over other miners. So those would have to scramble to get the same ASICs, some would go broke. But others will join in until again the majority of the block reward is spent on energy and machines. After a while the difficulty just goes up by a factor of one million, and we are no better off than before - perhaps even worse since all that machines had to be replaced. There is even a pretty compelling quantitative argument - since 2011 the energy efficiency of mining increased by a factor than 1000, but the electricity costs are still at more than 50% of the mining reward. So that didn't change anything, if at all.
"The bitcoin mining energy is comparable to gold mining energy usage"
Right, at least for the moment. But I find that a troublesome comparison. Mined gold doesn't use up energy after it's mined / processed and the energy needed for gold mining cannot be changed all that much, but bitcoin adoption might go way up and Bitcoin's energy usage could be dropped massively by moving away from PoW, like Ethereum did.
"Politics cannot do anything - if you outlaw Bitcoin mining, it'll move to China and other places."
Well, politics cannot kill off Bitcoin completely, but it doesn't have to. Consider what comparably minor events did to the bitcoin price and consider what would happen if the EU and the US outlaw commercial usage of Bitcoin, as at least the EU already has considered. And they just will have to do that if Bitcoin isn't changed and raises even more in price - IMHO they should have at least seriously threatened that long ago.
There are various security concerns with the alternatives.
Also right. But PoW is getting very centralized and thus isn't that effective anymore, either, and the problems it causes are IMHO an extremely compelling reason for massively pushing further research and doing some compromises.

How to get out of PoW?

Unfortunately, that's a tricky question - not so much technically but organisatorically, and it doesn't seem to get much attention in the bitcoin community. In practice the bitcoin miners play a huge role in the decision what changes do get adopted for Bitcoin. And how are you ever going to convince them to approve a change that would put them out of business?

So, I see two ways forward. One is to do a mix - perhaps use PoW for some blocks and another protocol like "proof of stake" (PoS) for other blocks, so he miners still have something to do for a while. Or they could be ignored with a hard fork - switching to a new protocol that cut's out PoW completely. Either way will probably need global politics to outlaw commercial use of crypto coins lile Bitcoin that use PoW at a certain scale. That obviously wouldn't kill bitcoin as we know it, but it doesn't have to - as it's restricting it to die hard fans or a couple of countries and lower its market capitalization (and thus the energy usage of mining) to an amount where it doesn't hurt too much.

Summary

All in one: bitcoin is an incredible invention, but it's "proof of work" scheme simply has to go. It was a Albert Einstein level genial invention, but now it is arguably the planet's most wasteful lottery by far, and there are other ways to manage Bitcoin and similar crypto-currencies. Don't get me wrong - I'm not saying Bitcoin is bad or that POW is without merit. I'm arguing that if the Bitcoin community wants Bitcoin to thrive and raise in adoption and price, it has to start looking very very hard into alternatives to POW.

If you agree or disagree with me, please contact me and tell me your opinion. I'm especially interested in ideas how to make the section "the economics of PoW" more precise.

(This article is also published here).