Over the past few weeks Bitcoin’s insane energy consumption took center stage in the media coverage regarding the popular cryptocurrency. This led to a lively debate, with some frequently returning claims being made. Let’s examine some of the most common ones.
Note: all numbers in this article represent the network’s situation per February 23, 2018.
Update: a new peer-reviewed academic paper titled “Renewable Energy Will Not Solve Bitcoin’s Sustainability Problem” was published in Joule on March 14, 2019.
1. The numbers are wrong
It’s important to realize that there is no way to verify the actual energy consumption of the Bitcoin network. The Bitcoin Energy Consumption Index produces an estimate for this number based on a set of conservative assumptions, but of course there will always be those that argue we should be even more conservative. We’ll probably never know who’s right, but what we do know is that Bitcoin uses an insane amount of energy regardless of who is.
The most conservative method we could possibly use to estimate Bitcoin’s energy consumption is by simply taking the total network computational power, and divide this number by the computational power of the most efficient Bitcoin mining machines. We know how much electricity these machines consume under ideal circumstances, so applying this method we would find an average electricity consumption of around 330 kilowatt-hours (KWh) per unique Bitcoin transaction (or around 22 TWh per year in total). This number doesn’t include any of the additional cooling in a large-scale mining operation, older generations of mining machines, or any of the half a billion people who might be mining Bitcoin without even knowing it (the latter isn’t part of the Bitcoin Energy Consumption Index either), but it provides a lower bound that is still 200,000 times more than an average VISA transaction.
Taking the numbers from the Bitcoin Energy Consumption Index we would end with an average energy consumption per unique Bitcoin transaction that’s a lot closer to 455,000 times the average VISA transaction, but the numbers don’t look good either way.
2. The comparison is flawed
Rather than arguing against the numbers, some will simply state that it’s unfair to compare Bitcoin to VISA and it should be compared to the current financial system. In all honesty, if a Bitcoin transaction replaces a regular VISA transaction, it does seem more than fair to compare the two, but let’s try to go along with the critics for now.
It was recently estimated that the total global banking system may be running on 100 TWh per year, compared to Bitcoin’s 51 TWh per year. That means that Bitcoin only requires half of the amount of electricity required by the global banking system. Case closed? Not exactly.
The first problem with the numbers is that the estimate for the global banking system includes the infrastructure required to support cash (Bitcoin doesn’t replace cash), and other infrastructure costs (light bulbs, cooling, etc.), while the estimate for Bitcoin includes only the cost of mining (there is an infrastructure surrounding Bitcoin as well). But let’s ignore this, as there is a more important matter to focus on.
The number of global non-cash payments exceeds 500 billion per year, or 1.37 billion per day. Bitcoin handles around 200,000 transactions on a good day, meaning that the current global payment system handles 6,800 times more digital transactions per day than Bitcoin does. Even on this flawed comparison, Bitcoin still requires 3,400 times more energy than an average non-cash transaction. This just confirms that there is no way to make Bitcoin look any prettier.
Others try to do so by comparing Bitcoin to gold instead. This comparison is quite far-fetched, as gold doesn’t have an inherent transaction limit, surging fees, or the possibility to duplicate the entire supply at will (eg. Bitcoin Cash or Bitcoin Gold). It also doesn’t make Bitcoin’s sustainability problem go away either, as even gold mining is less energy-intensive than Bitcoin mining.
3. Bitcoin doesn’t use that much energy in total
A popular narrative regarding Bitcoin’s energy consumptions seems to be that even though Bitcoin uses an extreme amount of energy for what it does, in total it’s still only a small part in the grand scheme of things. Being responsible for “only” 0.23% of the electricity consumption of the planet that’s not untrue, as is it well within the margin of error for the electricity consumption estimate for the rest of the world. The obviously flaw in the narrative is that it lacks a proper justification for Bitcoin’s extreme energy-inefficiency, and ignores that Bitcoin is a rapidly growing system. At the current pace, Bitcoin will soon require an increasingly significant part of the world’s electricity.
4. More efficient miners will reduce energy consumption
Based on recent years the trend of increasing energy efficiency has actually led to an increase in energy consumption, so there’s little support for this statement. Increasing efficiency by itself does little to change the total amount of energy consumed. If you can afford to spend $1,000 per day on electricity, that doesn’t suddenly change if you get different machines (although it would allow you to employ more machines).
What did happen was that Bitcoin mining went from being a decentralized activity that anyone could participate in with their CPU/GPUs at home, to something only centralized mining facilities with specialized hardware (mostly in China) could do. This is the natural outcome of economies of scale, and it has led to a drop of the average price paid for a kilowatt-hour of electricity. Residential rates are typically around 12 cents USD for a KWh of electricity, while industrial-sized mining facilities typically pay 4 to 5 cents USD per KWh. The result? The same $1,000 can now pay for more than twice the amount of electricity.
5. Bitcoin will consume as much energy as the whole world by 2020
Bitcoin’s energy consumption is in a bad situation, and certainly not improving, but predictions that the Bitcoin network will consume as much energy as the whole world by 2020 do seem a bit far-fetched. This prediction originally surfaced on grist, and followed from an extrapolation of the recent growth in Bitcoin’s energy consumption as presented by the Bitcoin Energy Consumption Index (the index itself only provides current and historic figures).
The prediction has two requirements to become a reality, the first one being that the Bitcoin mining revenues should be high enough to actually support such a massive electricity bill. At the moment, the revenues are nowhere near enough to actually make that happen, and the price of one Bitcoin should probably increase to well over a million USD to get it close to being enough. At that point, Bitcoin would be worth more than 17 trillion USD in total (the total value of the world’s money is around 90 trillion USD currently).
If the former could somehow be achieved, the actual machines that are going to be running on all that energy still need to be produced. Manufacturers are already struggling to keep up with the current demand for Bitcoin mining machines, so they would need a near-infinite amount of resources and production capacity to actually deliver a massive number of new machines in such a short timeframe.
Altogether, the bottom line is that one should be careful in making extrapolations (especially exponential ones). A more conservative way to make a prediction about the future of Bitcoin mining would be to consider the current profitability of mining. As per today miners are earning around 8 billion USD per year. According to the Bitcoin Energy Consumption Index the costs of mining (mostly electricity) are about 2.5 billion USD per year (thus around 30% of total revenues), making it extremely profitable to add more machines to the network. This means that even though the current estimated energy consumption for the whole network is around 51 TWh per year, there is plenty of room for this number to double (as the current revenues could support it). Such an increase would put Bitcoin in the same range as the Netherlands, Argentina or Sweden in terms of electricity consumption. It will take some time to get to that point, but given current production estimates, this does seem like a plausible possibility within the next year.
6. Bitcoin’s environmental impact is limited because it runs on renewable energy
One interesting line of thought is it wouldn’t matter that Bitcoin is thousands of times less energy-efficient than the current system, if the whole system runs on renewable energy (which is the future we’re heading to). We know this is certainly not the case at the moment (although parts of the network are being powered with hydroelectricity), but could we justify Bitcoin’s energy-inefficiency if it was? It’s actually not that hard to show that the answer to this question is no.
Regardless of the way electricity is produced, the lifecycle greenhouse gas emissions are never equal to zero. This is simply because you have to account for emissions associated with manufacturing, construction, installation and decommissioning as well. Doing so, studies find that even hydropower still has a carbon footprint of 4 grams of CO2eq per KWh. This is significantly better than coal-based electricity which has a carbon footprint of up to 1 kilogram of CO2 per KWh, but it doesn’t take a rocket-scientist to figure out that this improvement (being 250 times less) can never outweigh using several thousands of times more energy per transaction.
For the Bitcoin community the best course of action would now be to stop looking for excuses, and to start addressing the problem.