I often hear that improvements of mining technology (more hashes per kWh) would improve Bitcoin energy usage. Isn't that just a misunderstanding? Isn't the energy usage hardly dependent on the actual technology used by the miners, because improvements just lead to a difficulty raise?

Consider this thought experiment. Let's say there were miracle ASICs invented that have 1000 times the hash rate per kWh as current ASICs. What I think would happen is this. At first, miners changing to the miracle ASIC would be hugely more profitable than the others. So others would change to the miracle ASIC, too, and the global hash rate increases, and thus the difficulty. The miners using the old ASICs would get unprofitable and have to scramble to get the new ASICs, too. In the end all would change to the miracle ASICs, some would give up, some would come new, the difficulty would be 1000 times than before and the energy usage just about the same. (All in all the energy usage is determined by the mining reward, because there is an incentive for new miners to come in until most of the mining reward is spent on costs, and energy is the largest cost factor, so there is no coming down as long as POW is used.)

Am I missing something important there?

(BTW: I'm aware that there could be some shifts if e.g. the miracle ASIC is way more expensive than the old ones. My point in this thought experiment is that we wouldn't have to expect a change in the order of magnitude of energy usage, despite three orders of magnitude of change in mining efficiency.)

2 Answers 2


Your thought experiment gives you the correct answer: efficiency has nothing to do with energy usage. Once everyone gets the most efficient hardware, difficulty will adjust and you'll be getting the same reward per unit of energy, so they'll again put as much of it online until equilibrium with energy costs (ignoring here the fact that some may mine speculatively, at a loss, betting on an increase in price).

Ignoring speculative mining, and assuming no bottlenecks in new hardware going online, then at equilibrium you have only 2 factors affecting the total energy use:

  • If price goes up - total energy usage goes up.
  • If block reward gets halved - total energy usage gets halved.
  • If fees in native currency (BTC) go up / down - total energy usage goes up / down.

PoW blockchain networks bid for hashes, miners sell the hashes. More bidding power, more hashes. Also, miners bid for energy and hardware -- the more revenue they have from fees and block reward, the more hardware they can buy and more energy they can afford.

I object to concerns about PoW mining being called a "waste". If there was a better use for the energy, then that better use is free to outbid the PoW use. Nobody has a right to energy others have purchased. It is theirs to spend as they like, even "waste" it if they will.

But it is not wasted, the utility of expenditure is in increasing the cost to "undo" transactions, so the energy is spent on securing the blockchain by maintaining the immutability property.

  • Thanks for your answer! I admit calling it "waste" is biased. But I'm curious: do you really think > 0.3% of world electricity production is appropriate for that feature? Don't get me wrong: my point is mostly that the bitcoin community should work hard at finding an alternative. If bitcoin goes up to $200000, as some say, then it'll be more than 1-2% of world electricity, and then politics just has no choice hit Bitcoin hard once people notice that. Commented Oct 21, 2023 at 16:05
  • I reject the idea that some people should judge how much it is "appropriate" for some other to spend, so I'm not gonna judge it. Nobody forces energy producers to sell energy to miners. Free market allocates resources the best. Politics could do something more useful: allow new energy capacity projects to go online so even at $200k price it would again be 0.3%. If there's demand, capacity should increase. Kardashev 2 or bust. New capacity can come from renewables. In fact, PoW mining incentivizes creation of new renewable capacity. Commented Oct 21, 2023 at 17:00
  • Kardashev 2 is probably the right reference for Bitcoin. :-) If you want it to be 0.3% at 200k, we'd need to increase world energy production by a factor of 10, or 5 after the halving. I'd much prefer Bitcoin moving to proof of stake or something. Well, it'll have to, eventually. Commented Oct 22, 2023 at 20:26

I just realized there is also even a pretty simple but compelling quantitative argument that mining technology cannot have changed much, if anything at all. Consider this.

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

So: if better hashing efficiency by miners would help - where did that thousandfold increase in hashing energy efficiency go?

(That is, of course, a rhetorical question - it went into the automatic difficulty adjustment that by design makes sure no improvement in mining technology can ever bring the resource usage of bitcoin mining seriously down. If you read that article I cited you realize this is intentional.)

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.

(The lower bound the CBECI 9.5GW one year is 83.22 TWh per year, which is about 0.28% of the world electricity production of about 30000TWh a year - their actual estimation is 15GW. That seems reliable enough since it is in the same ballpark as with other estimations in the last few years: 0,16% and 0,55% .)

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.