Blockchain Energy Use: A Media Package

10 772
Avatar for AndrewStone
2 years ago

Recently Bitcoin’s energy use has risen to new heights, causing significant media coverage. This coverage is universally terrible – full of incorrect facts, misinformation, and (dare I say it) lies. It is important to be accurate with information, both for your journalistic integrity and for your employer since incorrect information could have an adverse effect on millions of people’s investments.

I am the founder and lead developer of Bitcoin Unlimited, an organization that has produced Bitcoin mining (“full node”) software and currently produces the same for Bitcoin Cash. I also teach graduate and senior level courses on Bitcoin and blockchains at Umass Amherst.

The following statements are entirely true but may surprise or even shock you. If you don’t understand them, you need to read this entire article before printing misinformation about Bitcoin’s and other cryptocurrencies’ energy use:

  1. If Bitcoin’s transaction rate increased by 10 times, or even 100x, or 100000x, its energy use would not materially increase.

  2. If more efficient Bitcoin mining devices are created, Bitcoin’s energy use will stay the same.

  3. If Bitcoin’s price doubled, its energy use would slowly rise until it has doubled.

  4. If Bitcoin’s price halved, its energy use would rapidly halve.

  5. Given a stable price, Bitcoin’s energy use will halve every 4 years.

And here’s a classic piece of misinformation:

  1. Although Bitcoin mining uses a lot of renewables, that doesn’t matter because it means coal and oil is being burnt elsewhere to power other stuff.

Bitcoin Energy Use is Defined by Price, not Production

This is the single most important idea to understand. Whenever Bitcoin commits transactions to the blockchain (by solving a block, which is what the miners do and where all the energy goes), the miner gains transaction fees (negligible) and some free Bitcoins (inflation). Currently the inflation is about 900 BTC per day, or 6.25 BTC per block. At current prices, this is 312,500 USD per block! It costs a few thousand dollars to buy a mining machine. So holy shit right?! Buy mining machines and run them! And that’s what many people did. So now you are competing with everyone else who are running mining machines to be the lucky person to solve a block, which happens about every 10 minutes no matter how many machines are running.

All those machines cost almost nothing for upkeep compared to the energy they use. So based on natural economic laws, people will keep turning on mining machines until the aggregate energy cost (energy used by every mining machine in the network) is about 312,500 USD per 10 minute period or 1.875M USD per hour, at today’s price of about 50000 USD per BTC. But at that point mining for bitcoins will cost more in energy than it produces in BTC, so new machines will not be added. If the price of a bitcoin goes up, more machines are turned on. If the price of a bitcoin goes down, machines are turned off.

Now I think you understand my statements 3 and 4, with the added observation that it takes time to bring new machines online but they can be turned off pretty much instantly. And a bit of thinking will allow you to understand 2: basically, if machines became twice as efficient, people will run more machines, until again the energy consumed basically equals the money received.

Statement 5 simply follows from one additional fact: Bitcoin’s inflation rate halves every 4 years. So 4 years from now miners will receive 3.125 BTC per block, or only 156K USD if the price of BTC doesn’t change. This means they’ll have to shut off machines to save energy, because you can’t spend 312K USD to produce 156K USD for very long. Eventually, 100+ years from now, this inflation will dwindle to nothing. And somewhere between now and then the transaction fees will start to matter, but that is a topic for another time.

Now Understand What Will Be Known As “The World’s Dumbest Decision”

As I described the situation above, I only talked about blocks, not transactions. That’s because creating blocks with 10, 1000, 1000000, or 10000000 transactions in them is irrelevant with respect to mining. All those transactions get boiled down into a single 80 byte chunk of data before being passed to the miners using an efficient algorithm that can take place on a normal computer (so negligible energy cost), but you’ll have to take my course to understand the details.

So taking the Bitcoin energy use and dividing it by the number of transactions is like taking the entire wind energy in the Atlantic Ocean, dividing it by the number of sailboats on the water, and thereby concluding that sailboats consume lots of energy.

But there’s a catch. In “The World’s Dumbest Decision”, powerful members of the Bitcoin community chose to limit the block size to about 1MB every 10 minutes (note that the average web page is > 2MB today and load in 10ths of a second), and therefore the number of transactions (sailboats) to about 2000 per block. This split the community almost in half (you might realize at this point what side of that split I was on), and created tremendous opportunities for profit-creating sidechains and nascent alternative cryptocurrencies, which some developers and miners had significant financial interest in.

So due to the “law of unintended consequences”, or what we could also call the “law of either being too stupid to think things through or too selfish to do the right thing”, the energy per transaction division actually does makes sense for Bitcoin right now.

But does not make sense for the industry as a whole, and can be remedied in Bitcoin as soon as its politics change. So it ought to be presented as such – a Bitcoin-only and possibly temporary problem.

From this description, I hope you understand #1 now. If Bitcoin ever increases its transaction volume, it will do so without increasing its energy use. And if it does not increase its transaction volume, other cryptocurrencies such as Bitcoin Cash, have already done so and so will take a larger and larger slice of Bitcoin’s transaction volume until Bitcoin becomes irrelevant.

Bitcoin’s Interaction With Renewable Energy

Its important to understand that energy supply and therefore price varies dramatically from location to location, and from one moment to another. This is because transporting energy is expensive, and can only occur over existing wires so has inflexible capacity. If you doubt this, you only need to read about the significant power outage in Texas last week (Feb, 2021), and the huge energy prices that resulted, and also about the success of the Hornsdale Power Reserve (

So statement 6 is simply not true, irrespective of Bitcoin. Consumption of renewable energy at some location absolutely does not imply production of non-renewable energy elsewhere. You simply can’t get that energy from here to there.

And if you look at the location of Bitcoin mining hardware, you will find that much of it is near renewable sources, typically hydro. This should be your first hint that there is something much more interesting going on than the knee-jerk idea that Bitcoin is evil because it “wastes” power.

What is actually happening is that hydro plants often produce a huge surplus of energy since the energy they produce is proportional to the water flowing through the dam, not the demand for that energy. This energy can’t be transported far because the grid can’t handle the capacity, and no one wants it anyway. So these hydro plants offer this energy very inexpensively so long as consumption is right near the dam. Its pretty hard to move an entire factory with all its workers 100s or 1000s of miles away up into some remote mountainous location. Its pretty easy to move bitcoin miners.

But yes, there’s an ugly truth that if the price of bitcoin increases too much too quickly, then it becomes (at least for a while) profitable to run the miners on more expensive, non-surplus, non-renewable energy. This is what we see during Bitcoin’s short “boom” cycles which ironically is the only time you reporters come running over here to write articles.

Forward Looking Ideas About Energy

Dismissing Bitcoin as a dirty technology may be throwing out amazing techniques to actually aid renewable use, so I’d like to look into my crystal ball a bit and propose some “crazy” ideas.

First, Bitcoin mining could allow greater use of nuclear power, where new designs are generally agreed to be cleaner than coal/gas and seem to be much safer (but I’m not here to advocate for nuclear, I’m just here to point out a possible interaction with Bitcoin mining). The problem with nuclear is that it provides a steady “base load” but can’t ramp up and down quickly. And you know, you can’t over produce energy without bringing down the grid. You have to meet demand exactly (sometimes bulk energy prices actually go negative – providers literally will PAY YOU to consume energy). So you have to run quick-to-power-up but dirty non-renewables to handle any load variation. But what if there was a “magic black box” that could soak up extra supply and turn it directly into money? In that case, nuclear baseline load could be higher.

Second, Bitcoin mining could do the same thing for renewables. The problem with renewables is their erratic production. But what if you could profitably over-build renewable sites, so you are producing (say) 150% of power demand on a sunny or windy day? And then on a cloudy or light wind day you drop down to 100% demand. Why is this good? Well, if you built out to produce 100% only on the brightest or windiest days, then on cloudy or just breezy days you have to fire up the coal or gas generator to meet demand.

We cannot overbuild and use that energy for much else. For example, could we build a factory that uses that extra-but-variable 50% to make aluminum? No, because it and its workers would sit idle too often, need to be fired up at strange times of the night, and work continuous multi-day and night shifts. Worse, sometimes we need to start 1% of that factory’s production, other times 100%. But Bitcoin miners have none of these problems. So we may be able to deploy Bitcoin miners to convert extra renewable electricity production into money that essentially pays for that deployment of extra renewable capacity, averting the need to fire up dirty non-renewable production.

Finally, a Bitcoin miner is a great device for “load shedding” (this is when the power company automatically turns off power to stop the entire grid from failing).

You could even imagine it behaving as a lossless energy “teleporter”. Imagine a future where miners are more or less evenly distributed across a geographic region, producing money from solar power. Does one region need more energy and one region have too much? Rather than pass the power through lossy, expensive-to-maintain, and “nimby” (Not In My Backyard) high voltage power lines, turn off miners in the region that needs energy, and turn them on in the region that has too much.

Co-locating bitcoin mines near surplus hydro is the first step in the future I’ve outlined here. And its already happened. But there are a few reasons these more advanced ideas haven’t been fully realized. The first is simply time. The second is that the boom cycles of Bitcoin make it profitable to mine (during the boom) with expensive energy. But as Bitcoin grows, the boom cycles will stabilize (look at prior boom cycles, they were bigger on a percentage basis).

The third is that mining hardware used to be getting iteratively but significantly more efficient on a yearly basis, obsoleting old hardware. So miners had to run their hardware 24x7 so that it would pay back its investment before becoming obsolete. However, just like how computer CPU performance increases are much smaller now than they were in the 1980’s and 1990’s, the mining hardware’s chips are now approaching state-of-the-art in terms of transistor density. This happened because its a lot cheaper to produce chips using older technology, so the first mining chips were produced using old and relatively cheap technology. The next chips used less old and more expensive technology, and so on. Mining chips literally passed through much the same technologies and therefore transistor density increases as CPUs, but did so in 7 years rather than 40.

So we can expect that the usable life of a bitcoin miner today or in the near future will be long enough to allow them to be powered on intermittently rather than continuously.

Fourth, in some areas Bitcoin mining has gotten a bad reputation for encouraging the building out of power connectivity to a site during the boom and then cancelling the contract during the bust . Well, honestly that should be on whoever signed that contract on the power side, although I can't really blame them for not understanding this emerging market. Make the miner pay for that build out, and you have a free data center built and ready to rent if the miner moves out.

Of course, you are thinking that we could just push surplus power into batteries. For sure, but how much battery capacity is ultimately deployed depends on the continued development of battery technology, their usable life, and the ability to transport that power. It seems unlikely that a battery will ever be developed that can store a rainy season’s worth of excess hydro power into the rest of the year’s worth of demand. Certainly, such a device would look nothing like today’s batteries. It is more likely to efficiently convert electricity into some relatively stable chemical, like gasoline.

Or perhaps that “storage technology” will be a little black box that converts electricity directly into a new form of money in exchange for the valuable service of helping to secure the world's financial network. This new money is stored easily, cheaply, does not age, and can be used to purchase power during the dry season from someone’s solar surplus. Conversely that solar producer purchases your power during the rainy season.

The future of renewable energy production and storage is by no means figured out. But sitting here today, I think that cryptocurrency mining may play a very interesting role in it.

$ 24.18
$ 10.23 from @TheRandomRewarder
$ 10.00 from @bitcoincashautist
$ 1.00 from @emergent_reasons
+ 9
Avatar for AndrewStone
2 years ago


Overall, great article.

I'm looking forward to more growth so that the more energy efficient alternatives end up getting more market share.

$ 0.00
2 years ago

Very great article keep it up dear friend 😘

$ 0.00
2 years ago

Dams store energy by letting the water get higher. It's not as flexible as gas peaking units, but a heck of a lot better than solar (without expensive batteries).

Solar producing too much during the day or nuclear producing too much at night does not enable bitcoin to benefit from the variation or for the world to benefit from bitcoin. Equipment used intermittently is repurposeable which means it could be used in an attack during its off time. The article seems to put an emphasis on coin generation as if generating coins at this late stage adds value to the coin or to the world (other than financing more equipment which adds value via increased security, not directly "more coin"). In the initial stages coin generation was required for fair and decentralized distribution of coin (assuming adopting_early + keeping_faith = filthy_rich is "fair").

The importance of mining now is security which means the equipment should be evenly distributed across the world and constantly in use. Being able to turn the equipment on & off and move it to where electricity is cheapest (which might be a "Bitcoin-unfriendly" country like China) are both detrimental to security. I see no value added to Bitcoin or the world by "energy arbitrage"). Security depends exclusively on the CAPEX of the equipment (OPEX like electricity and Moore's law decreases it). This is well-established in research and there are no opposing viewpoints. If electricity use cuts in half due to halving or price cuts in half (all other things being equal) then security is theoretically ruined due to all the idle equipment that can be used for very profitable short-term attacks (or enable the attacker to pick which transactions are valid).

It's more important to point out that Bitcoin's security is aligned with zero energy use (other than creating the equipment). Creating the equipment is also very "wasteful" of society's resources (including a lot of energy). "Wasteful" means value is "burnt" and the resulting equipment can't be used for anything other than Bitcoin. By the definition of security in POW, it can't be used for anything else. POW is more correctly called proof of waste (I'm serious). But this subtlety will be missed by the populace.

Concerning your item 2, if price were trusted to be stable or believed to only increase and if Moore's law were dead and if it's surviving on fees, then miners could concentrate on spending a lot more on efficient equipment to eliminate electrical costs. If the equipment will remain competitive forever, there's almost no limit to how much CAPEX they might spend to make OPEX near zero.

CAPEX is the only source of security. OPEX is the ideal way to generate and distribute coin. There's good and bad to BTC combining the two. Proof: if there is only OPEX, coin generation can be done by anyone paying electricity (or rented if some location has cheaper electricity) so it's fair distribution. but consensus security on transactions is broken because you only need electricity to attack 10 blocks and get maybe 10x more profit from double spends. In the CAPEX-only case, there's no incremental cost for each coin generated (which breaks the whole idea) and there might be some company that makes highly efficient equipment that it does not share with anyone. Papers show CAPEX is the sole source of consensus security on transactions.

$ 0.10
2 years ago

hi, noob here. can you point me in the direction of some of this CAPEX/OPEX research? i do not know where to begin and would like to take this discussion seriously. thanks.

$ 0.00
2 years ago

I don't agree with everything he wrote but he's not wrong re. CAPEX being the guarantee of security. PoS is really not a hard stake, you can dump it at will and exit the business at the expense of whatever bagholders you sell to. With PoW you'd dump your hardware at a loss and the expense would be yours. Miners are committed to continue to mine especially with ASIC hardware which can't be repurposed -- they have no choice but to do it to return the capital investment. And unlike PoS those who already own PoW hardware can't control the supply, anyone can make/buy their own hardware, others already owning the hardware have no say in it. With PoS you can't get in unless someone's willing to sell you their stake.

$ 0.55
2 years ago

POS could use POW to initially create the coin. It is the only known way to decentrally & fairly distribute coin. I do not see that POS is different from POW in regards to selling. To mix and match your two comparisons: You could sell equipment to new bagholders. The stake can't be repurposed. The root of the problem with POS is that the stake is on the chain instead of off-chain like the equipment. This makes randomly selecting a leader (block creator) a lot harder for some reason. If you add VDF or VRF to POS, you start with a pseudo-random value (like POW's nonce) and end with creating the block from it. In POW, the process is reversed.

$ 0.05
2 years ago

Hardware and stake are different because hardware naturally depreciates in value whereas your stake could go either way. If you exit stake "mining" by selling with a profit you're being rewarded for quitting the job. You could also stop mining and simply hodl the coin and speculate on price appreciation, and your "equipment" wouldn't depreciate so there's only missed revenue but no capex depreciation. Incentives don't feel right with staking.

$ 0.00
2 years ago

If my last paragraph wasn't enough, then this long and boring paper goes into every aspect. In short, the heavy cost of the equipment (CAPEX) marries the miner to the coin. If he hurts the coin, he hurts his future profits spread out over many many blocks. But if all of his cost is immediate like electricity or renting the equipment, then attacker needs to spend only a little on electricity (or renting) on a few blocks to get larger rewards. The value in transactions (his potential for double spends) must typically be much much greater than the fees+reward miners get for the electricity. For some reason, when equipment is not repurposable, there's not a significant rental market.

$ 0.20
2 years ago

thanks for the paper. now to hurt my brain!

$ 0.00
2 years ago

Given a stable price, Bitcoin’s energy use will halve every 4 years

This one isn't necessarily correct, right? Assuming bitcoin manage to attract fees to cover, or outperform, the reduction in block subsidy you won't necessarily get a halving in energy, even if the price is stable.

$ 0.00
2 years ago