Nobody Knows Exactly How Much Energy Bitcoin Is Using

By the end of this year, Bitcoin may account for a whopping half of a percent of the world’s total energy demand. It doesn’t sound like much, but that is roughly equivalent to the energy needs of Austria, a country of nearly nine million people. This sobering prediction was made by financial economist Alex de Vries and published on Wednesday in Joule, marking the first time that the energy consumption of Bitcoin has been quantified in a peer-reviewed journal.

“Most people are shocked that Bitcoin could use this much energy,” de Vries told me on the phone. “I think that’s the appropriate reaction. Half of a percent of the world’s energy is a huge number, especially for a system that isn’t really used for anything except speculation.”

Over the past few years, de Vries’ blog Digiconomist has garnered a lot of attention for quantifying the amount of energy used in a single Bitcoin transaction (currently the same as an average Dutch household uses in a month). He used this value to quantify the energy consumption of the network as a whole (currently the same as the country of Ireland). Earlier this year, de Vries calculated that the process of securing the Bitcoin blockchain—known as mining—is over three times more energy intensive than actually mining real gold per dollar mined. This hasn’t made him popular amongst the hodlers online, who see the critique as an unfair attack on Bitcoin. In a representative Reddit thread, rebuttals to de Vries’ analysis mostly focus on how plenty of other things waste electricity as well, such as Christmas lights.

“There are some diehards in the Bitcoin community that don’t want to talk about this because they hold Bitcoin and think it’s damaging,” de Vries told me. “But overall I’m getting a lot of really positive reactions from people who are happy that I’m trying to understand this problem which is growing quite rapidly.”

Still, de Vries is the first to acknowledge that his estimates are just that—estimates. It may be impossible to ever know the exact energy consumption of the Bitcoin network, but upper and lower bounds of its energy consumption can be reliably calculated using economic models. Improving the accuracy of these models, however, requires a lot of information that is simply not currently available in the largely unregulated cryptocurrency space, de Vries said.

As detailed in de Vries’ new paper, the Bitcoin network currently consumes at least 2.55 gigawatts of electricity and may consume 7.67 gigawatts by the end of this year.

To calculate the lower bound of this estimate, de Vries took the total mining power of the network and divided it by the energy efficiency of the most efficient Bitcoin miner on the market, the Antminer S9. That means that if everyone on the Bitcoin network was using the most efficient computer chip available, the network would be using a total of 2.55 gigawatts of energy. In reality—where people use old and inefficient hardware—that number is likely higher, although determining just how much higher, or how much it can grow in the future, is really tough.

Besides the difficulty of pinning down which hardware miners are using, its nearly impossible to tell how many individual miners are actually running on the network. There are approximately 10,000 nodes that make up the Bitcoin network, but each node might host dozens or even hundreds of miners. This can lead to large discrepancies in power consumption. As de Vries points out, the same hashrate can be achieved by a single Antminer S9 or about 500,000 Playstation 3s—the Antminer would use 1,372 watts of electricity compared to 40,000,000 watts in the case of the PS3s.

De Vries said he was able to predict the future energy consumption of Bitcoin by using economic models in which Bitcoin can be considered to be a “virtual commodity with a competitive market of producers.” De Vries cites a 2015 paper about Bitcoin economics written by the University of Wisconsin economist Adam Hayes, who predicts that miners will continue to mine “until their marginal costs equal their marginal product.”

This is an economist’s way of saying that more mining power will be added to the network (and hence more energy consumed) until the cost of the energy from adding another unit of mining power is equal to the value of the reward gained by adding that extra mining power, either through the generation of new bitcoins or the collection of transaction fees. At that point, the miner is effectively “breaking even” and it doesn’t make economic sense for them to continue adding more mining power to the network. Once that equilibrium point is reached, the power consumption of the network will effectively level off or decrease as mining will no longer be profitable.

De Vries calculated that about 60 percent of the Bitcoin value produced by a miner on March 16, 2018 went to paying for electricity. Based on that figure, he calculated that the equilibrium point will occur when the Bitcoin network reaches around 7.67 gigawatts of power consumption.

He then looked at reports estimating the number of chips that Bitmain—a Chinese firm that dominates Bitcoin mining and hardware production—is able to produce and found that the company could produce enough chips by the end of this year to bring the Bitcoin network’s total energy draw to 7.67 gigawatts. Since after this point miners would be breaking even, assuming the price of Bitcoin stays the same, it is the upper limit to Bitcoin’s energy use.

The equilibrium limit established by de Vries requires a lot of assumptions. For starters, it’s a static picture of a market that is really dynamic. If the price of Bitcoin changes, this also affects the point of equilibrium—if the price of Bitcoin goes down, so too does the point of energy equilibrium and vice versa. And as everyone knows, the price of Bitcoin can change dramatically day-to-day.

Moreover, de Vries’ estimates assume that miners are rational actors, though he told me it’s possible that some miners will mine at a loss for ideological reasons or to take advantage of stolen computer power. For example, National Science Foundation supercomputers were used to mine between $8,000 and $10,000 worth of Bitcoin in 2014, even though the computing power ended up costing the facility $150,000. A more trivial example of mining at a loss is the rise of cryptojacking malware, which uses a victim’s computer to mine cryptocurrencies. This is only mining at a loss from the perspective of the victim, whereas the hacker is profiting no matter the value of the coin being mined.

In short, there are simply too many unknowns to get anything better than a ballpark estimate of future Bitcoin energy consumption.

“Right now we're just looking at extremely limited data,” de Vries told me. “There is some information available, but its often conflicting. One of the first things we should do is try to get more information about these facilities and then we can better understand them. But for now it's a black box and its really hard to get a good number out of it.”

De Vries believes it may be possible to get better information about Bitcoin’s energy consumption in the future, however. One possible tactic we discussed on the phone would be to require Bitcoin mining operations that are above a certain size to report their energy consumption to a state agency. A similar program is already being trialled in Russia, which will make cryptocurrency miners register with the government starting this year.

The problem with this solution, of course, is that Bitcoin is transnational so this would require cooperation among China, Russia, Canada, and the US to report this information.

As far as mitigating Bitcoin’s effects on the environment, regulators can charge higher electricity rates for large scale mines or cap the amount of electricity a mine can use. In February, a small city in upstate New York became the first in the US to temporarily ban Bitcoin mining to prevent a huge upswing in electricity costs for its residents. Both variable rates and electricity caps are being considered as possible solutions to the town’s Bitcoin mining problem.

For now, however, there are few mechanisms in place to stop Bitcoin from consuming more electricity. Bitcoin evangelists like John McAfee see the price of Bitcoin rising to over $500,000 by 2020, which would significantly raise energy levels consumed by the system as it would be more profitable than ever to mine bitcoins.

However, detractors consider Bitcoin’s economy to be in a bubble phase, and if it pops the price crash would also result in a huge down turn in mining and thus electricity demand. There’s also technical solutions like the Lightning Network on the way, which processes transactions off of the main Bitcoin blockchain and would thus significantly lower the energy cost per Bitcoin transaction. Another alternative is to use wind power and other forms of clean energy for Bitcoin mines.

But before a good solution to Bitcoin’s energy problem can be devised, it will be necessary to understand the full scope of that problem. Given that this will require greater government oversight of a commodity created specifically to avoid government meddling, it’s likely going to be an unpopular proposition.