Understanding Bitcoin’s energy use

Why does Bitcoin consume electricity?
Does energy usage increase with the number of transactions?
Can we compare Bitcoin’s energy usage to traditional payments technologies like credit cards?
Are Bitcoin miners polluters?
Will Bitcoin miners pollute more or less in the future?
Will cryptocurrencies become more efficient over time?

Bitcoin consumes a sizable amount of electricity. As of June 2021, estimates suggest something around 110 terawatt hours (TWh) per year, which, for scale, is close to the electricity consumption of the Netherlands (111 TWh) but a bit less than the global ‘phantom’ electricity consumption from electronics that are left plugged in while in standby mode (124 TWh). The mere fact that Bitcoin uses this much electricity is not itself a problem, however. From a policy perspective we should be more interested in whether Bitcoin’s electricity usage is derived from dirty and finite or clean and renewable energy sources, and whether that energy usage is justified by the capabilities and potential of the technology to improve the world. This backgrounder will investigate those and other fundamental questions and direct readers to the most up-to-date research.

Why does Bitcoin consume electricity?

People who don’t understand the core computer science breakthrough inherent in Bitcoin naturally assume that Bitcoin miners only do one thing: burn electricity to enrich themselves with new bitcoins. These critics are mistaken, but it’s a reasonable mistake to make because “mining,” the term itself, is a bit misleading.

Search the original Bitcoin white paper from 2008 and you won’t find the word “mining.” The word first popped up in 2010 on an internet forum called BitcoinTalk and, for a while, it was used interchangeably with the term “minting.” When technologies don’t have singular corporations or inventors working on them and branding them, terms to describe them emerge organically from the community—it comes with the territory.

“Mining” is a misleading term because it only tells half the story. Real world miners leverage expensive capital to dig impressive holes into the earth, to discover and extract marketable metals, minerals, and gems. They burn resources to get valuables. A bitcoin miner does this too, no doubt, but they also do something else that is fundamental to Bitcoin’s core computer science breakthrough: they validate data on a computer network running an open consensus mechanism.

We have an in-depth explanation of what mining accomplished and how it works, but let’s talk here for a minute about open consensus, which is an inherently useful technology that has never been possible before.

Bitcoin, aside from being an asset, is also a network of connected computers on the internet that, together, keep a record of all bitcoin transactions between the participants. You join the network by running freely available software on an internet-connected computer, and when you join, your computer will let you receive and send payments in bitcoin to and from other computers, and it will also help store and update a continually growing list of everyone’s transactions, called the blockchain. People trust Bitcoin as a store of value and medium of exchange in part because everyone can see this blockchain and see all the historical transactions (including their own) going back to the launch of the network in 2009.

Now what about this “open consensus mechanism” mentioned earlier? Consensus in this context just means agreement; we’re just trying to get all the computers connected to the network to agree. Mechanisms for generating consensus between several computers have been around since the 1980s. Those old mechanisms would allow, for example, six data centers owned by IBM to stay in sync with each other, storing and updating some data that IBM cares about and wants redundantly stored on multiple machines.

Bitcoin’s network is also redundantly storing data (the transactions) across a bunch of machines (all connected nodes), and it needs these computers to stay in consensus. Bitcoin’s core computer science innovation was the invention of the world’s first open consensus mechanism. Our IBM consensus example is closed; the consensus rules only allow a set number of computers (e.g. six data centers) to participate at a time, and only computers that IBM authorizes can join (kind of like an intranet). Bitcoin’s consensus mechanism allows an unlimited number of computers to participate and anyone can join (more like the internet)! That’s what is meant by an open consensus mechanism. This is also why people say that Bitcoin is “decentralized,” and why it’s accurate to call Bitcoin peer-to-peer digital cash, as compared with the centralized digital money created by a company, like PayPal or Venmo, that secures the payments data and decides who can add new payments data.

Older, closed consensus mechanisms stay in sync because identified participants take turns adding new data to the record, and they are secure because only identified participants are allowed to add data. If you have an unlimited number of unknown participants, how do you have them take turns and how do you know they aren’t committing fraud? This is where Bitcoin mining comes in. We have a longer post on the subject but here are the highlights in bullet form:

  • Miners are picked by lottery,
  • A winner is picked every 10 minutes by algorithm,
  • To be picked you have to perform some costly and verifiable computing work (lottery tickets must have a non-trivial cost).
  • Miners who try to put invalid transactions into the blockchain will not be picked.

This mechanism makes fraud non-viable because miners suffer a cost to even be eligible in the lottery and they lose their eligibility if they try to submit invalid transactions. Your attack won’t work and you’ll lose money in the process.

To keep the lottery fair, the price of a ticket rises as people buy more of them; in other words miners have to compete. So if a lot of people are willing to spend computing effort to join the consensus, then the costs of participation will rise as the computing work you need to perform becomes more and more difficult. More computations means more electricity and that, after some adieu, is why Bitcoin’s electricity usage has been going up.

You might think this is all a bit of a rube goldberg machine. If you do, you’re starting to understand Bitcoin! It is, indeed, complicated, but as of right now it is the most reliable mechanism in existence that allows a large set of unidentified computers to agree over shared data and, therefore, the best way to have peer-to-peer electronic cash. This energy usage is anything but useless, it is securing data about transactions worth hundreds of billions of dollars. Unlike the energy used by a gold miner, it goes directly to providing a public good: online peer-to-peer payments infrastructure that anyone on Earth with a smartphone and internet connection can use.

Does energy usage increase with the number of transactions?

It’s a common misconception that energy usage scales up with the number of bitcoin transactions and that if Bitcoin ever became widely adopted for payments, then the energy usage would be enough to boil the oceans. This is incorrect. As we’ve learned above, miner energy usage moves up or down with the amount of competition between miners, not the number of transactions being validated. Digital signature validation uses a trivial amount of computing power. Your three-year-old laptop can verify a signature in a matter of milliseconds, and you’d be hard-pressed to find even a scrap of evidence of that work in your electrical bill.

Why is there so much competition driving so much energy usage? Economics. Bitcoins are expensive right now, and every 10 minutes one miner will get 6.25 shiny new ones. This competition is healthy because it means that the effort spent securing the network scales automatically with the value of the transaction data on the blockchain—not the number of transactions. So the more value there is riding on the Bitcoin network (because individuals value it more as reflected in the price), the more resources will be devoted to its security. That makes for a noteworthy contrast with data secured by, say, Equifax or any other big data company where effort spent securing data scales somewhat arbitrarily with a corporate management team’s estimation of cybersecurity risks and fear of data breach liability.

One final thing about that competition. It may get less fierce as time goes on. The reward of new bitcoins halves every four years until it goes effectively to zero. Miners will keep working because they can also collect fees that users of the network add to their transaction messages, but the total take-home payment for a winning miner will probably be less in the future than it is today even if the price of a Bitcoin continues to rise. Smaller rewards will mean less computing power dedicated to winning and less electricity consumed.

Can we compare Bitcoin’s energy usage to traditional payments technologies like credit cards?

This is comparing apples and oranges, but investigating the question will still help us better understand the energy usage landscape. Unlike a Bitcoin transaction, a credit card transaction is not a settled payment. It is a payment authorization, merely the beginning of a convoluted dance between no fewer than five players (the cardholder, the card issuing bank, the card network, the merchant acquiring bank, and the merchant). Eventually (think days or even weeks) the authorized payment will be pulled through that network of players (it ends when you pay your credit card bill), but it certainly doesn’t happen with a mere swipe or tap.

In contrast, a Bitcoin transaction is settled the moment it hits the blockchain. And it need not travel through any clunky analog institutions along the way. The miner who put it in a block probably used a bunch of electricity, it’s true, but that amount is substantially less than the energy it costs to keep the several banks and financial companies involved in a traditional payments settlement running at full steam for a few days.

It’s difficult to estimate the energy usage of the traditional financial system because of the sheer number of institutions and governments involved. In the US alone there are over 4,000 banks, 75,000 branches, and over 470,000 ATMs. Each physical branch and ATM has a carbon footprint, as do the computer systems employed to reconcile accounts between banks, the fintech and credit card companies that facilitate consumer payments ultimately settled by banks, and the state and federal governments that regulate and ultimately enforce contracts for settlement. Even a generally conservative energy estimate for the traditional financial system suggests that it uses as much as five times the energy usage of Bitcoin mining.

It’s somewhat more straightforward to estimate the energy usage of the gold industry. Comparing bitcoin with gold is, again, not a like-for-like comparison but reasonable estimates suggest that bitcoin uses about half the energy of gold mining and recycling.

Are Bitcoin miners polluters?

Energy use is not bad in and of itself. It is greenhouse gases that are bad, and Bitcoin miners already use more clean and renewable sources like geothermal and hydroelectric as a percentage of their total electricity consumption than the American household average. This is another hard thing to quantify, but reasonable estimates start at 28% and go as high as 74% renewable. The reality is likely somewhere in between but that’s significantly better than the total US renewable energy share at 11.6%.

Miners are mobile while electricity is not. Miners can easily locate wherever they have an internet connection and electricity. If power is already being generated in a certain location and the local population doesn’t need it, there’s no way to transport that electricity to where it would be more useful. Even though most power is generated locally we still lose a tremendous amount to transmission inefficiency (e.g. nine percent average losses in California and as much as 30% losses in India). Many types of power plants are also simply impossible to shut down without losses; the only way to stop a hydroelectric dam, for example, is to release excess water downstream and give up all that stored potential energy. Because of transmission inefficiencies, electricity that is produced beyond the needs of the local economy is often referred to as ‘trapped energy’ and it’s a real problem from an energy policy standpoint. Many Bitcoin miners choose to base their operations in parts of the world where significant green electricity is trapped because it makes mining more profitable: Iceland where there is geothermal power available, the Columbia River Valley in the US, and Sichuan province in China where there is abundant hydroelectric power. While China has lately been attempting to ban or limit mining activities in their borders, they’ve allowed mining to continue in Sichuan because of the amount of otherwise trapped and lost hydroelectric power in that province.

Some miners are even using power generated by natural gas flaring. Whenever oil companies drill, some amount of gas will build up on site and need to be released. Companies can simply vent this into the atmosphere or they can burn it to produce electricity. Venting is much worse for the environment than burning because methane is, pound-for-pound, 25 times more powerful as a greenhouse gas than carbon dioxide. Safely burning excess gas, however, requires costly infrastructure. Some Bitcoin miners are happy to foot the cost of safely burning flare gas because they can use the resultant power, on site, to mine bitcoins.

The mobility of mining is a double edged sword, however. If electricity is cheap in a region where coal-fired power plants predominate, Bitcoin miners will likely locate there as well. While mining in China’s Sichuan province predominantly uses hydro power, China’s other prominent Bitcoin mining region, Xinjiang, has cheap electricity because of government subsidized coal mining. While some may wish to blame Bitcoin miners for that pollution, the fact is that Xinjiang electricity is cheap and pollutes because the Chinese government is paying people to burn coal. Bitcoin doesn’t make that any better or worse, but Bitcoin is also not the problem in this case; the problem in China’s energy policy.

Will Bitcoin miners pollute more or less in the future?

If Bitcoin mining becomes the dominant driver of energy consumption on the planet, then that could actually be a good thing for the environment. Just as the consumer electronics revolution drove the massive computing efficiencies known as Moore’s law; the Bitcoin revolution could drive a similar explosion of innovation in clean, efficient energy.

Aluminum smelting consumes about three percent of the entire global supply of energy, yet we don’t read articles raising the alarm on unibody MacBook Pros like we see about Bitcoin. Smelting isn’t often thought of as a problem because heavy industry drives electricity efficiency. Why? Because heavy industry is a big consumer, they’re always looking for the cheapest possible source of electricity. Like Bitcoin miners, aluminum smelters chose to locate in places like Iceland so that they can take advantage of trapped green energy.

Heavy industry can generally be based anywhere, and electrical costs tend to be a large percentage of their total costs. Electricity is 40-45% of costs to chemicals manufacturing (like chlorine production) and a whopping 30-50% of costs to steel and aluminum smelting. That means that heavy industry will base itself where costs are lower, and that will tend to be wherever electricity is affordable because its production is more efficient. Demand drives supply and thus rewards those who develop cheaper modes of electricity generation. Lately that has roundly been a green affair. The cheapest electricity on the planet is now wind and solar energy. Geothermal and hydroelectric are also top contenders and don’t have to deal with storage issues.

However, electricity costs may not always be top of mind for typical heavy industry. They may put up with expensive, dirty energy if other costs drive their decision-making. Industries also like to be where their customers are, where it is cheap to ship material inputs like metal, and where governments grant them subsidies in order to encourage industrial growth.

But electricity costs matter even more to a Bitcoin miner than typical heavy industry. Electricity costs can be 30-70% of their total costs of operation. Also, Bitcoin miners don’t need to worry about the geography of their customers or materials shipping routes. Bitcoins are digital, they have only two inputs (electricity and hardware) and network latency is trivial as compared with a truck full of steel.

If Bitcoin mining really does begin to consume vast quantities of the global electricity supply it will, it follows, spur massive growth in efficient electricity production—i.e. the green energy revolution. Moore’s Law was partially a story about incredible advances in materials science, but it was also a story about incredible demand for computing that drove those advances and made semiconductor research and development profitable. If you want to see a Moore’s-Law-like revolution in green energy, then you should be rooting for, and not against, Bitcoin. The fact is that the Bitcoin network, right now, is providing a ±$200,000 bounty every 10 minutes (the mining reward) to the person who can find the cheapest energy on the planet. Got cheap green power? Bitcoin could make building more of it well worth your time.

Will cryptocurrencies become more efficient over time?

We already learned that energy use will not increase alongside the number of Bitcoin transactions. As it turns out, however, there’s a good chance the energy use stays constant or even decreases while the number of transactions skyrockets. Several efforts are underway to develop second-layer networks, or new open consensus mechanisms, that could allow for thousands and even millions of transactions per second. These scaling solutions take various approaches. The Lightning Network and similar payment channel infrastructure aims to do batch settlement of several transactions by writing only two transactions to the blockchain itself, and it has automatic controls in place so you won’t have to trust the honesty of the person doing the batch settlement on your behalf.

Other developers are experimenting with new open consensus mechanisms that may enable massive scaling while simultaneously reducing energy usage. Recall that Bitcoin’s consensus mechanism has a built-in lottery that miners are competing to win. Recall also that every lottery ticket has a cost measured in verifiable computations made. These computations are called a “proof-of-work” and we often refer to Bitcoin’s consensus mechanism as proof-of-work based.

Ethereum developers, to take one example, are now experimenting with a “proof-of-stake” consensus mechanism called Casper, and other proof-of-stake cryptocurrencies have recently launched, like Tezos and Cosmos. In a proof-of-stake consensus mechanism miners (or validators) still compete with each other to win a lottery, but tickets are provably costly because the participants prove that they own or have a stake in some amount of the cryptocurrency secured by the blockchain. It might be more appropriate to call these consensus participants stakers than miners but who knows which name will stick or if proof-of-stake will even be as secure and viable as proof-of-work. The one thing we know for sure is that there is no shortage of innovative thinking out there, and the cryptocurrency revolution may turn out much greener than you’d think.