A data center.Photo: BalticServers/Wikimedia
It takes a lot of electricity to mine bitcoins, and as more miners try to cash in on the crypto craze, the amount of energy required to win new coins increases accordingly. New research suggests the entire bitcoin network could consume as much as 7.7 gigawatts of electricity by the end of this year—enough to power a country the size of Austria. But the new analysis is not without its critics, who say there are many other factors to consider.
By the end of this year, bitcoin could use half a percent of the world’s electric energy, according to research published today in the science journal Joule. What’s more, if bitcoin’s value rises as high as some experts predict, the network could eventually consume upwards of five percent of the world’s electricity.
That’s shocking, to say the least, if not completely absurd. If true, it means bitcoin in its current state is grossly inefficient, completely unsustainable, and an environmental menace. But as critics of the new study are quick to point out, the new paper is predicated on several shaky assumptions, insufficient evidence, and an eye towards the status quo in terms of how new technologies and regulations might alter the situation. The future of bitcoin, and all the alternate currencies vying crypto glory, has not been set in stone.
Blockchain specialist Alex de Vries, the founder of the Digiconomist blog, a senior consultant at the Experience Center of PwC in the Netherlands, and the sole author of the new study, estimates that bitcoin currently requires about 2.55 gigawatts of electricity, and that a single transaction sucks up as much power as an average household during an entire month. By analyzing current trends in bitcoin mining production, and assuming an electricity price of five cents per kWh, de Vries predicts that bitcoin’s energy needs could reach 7.67 gigawatts later this year, placing it in the ballpark of countries such as Ireland (3.1 gigawatts) and Austria (8.2 gigawatts).
“With the Bitcoin network processing just 200,000 transactions per day, this means that the average electricity consumed per transaction equals at least 300 kWh, and could exceed 900 kWh per transaction by the end of 2018,” writes de Vries in his paper, adding, “Bitcoin has a big problem, and it is growing fast.”
The mining problem
Bitcoin requires such a tremendous amount of energy because it needs computers to time-stamp transactions along the blockchain—a publicly validated, network-powered ledger that makes cryptocurrency possible. The blockchain prevents duplicate spending of coins through a process known as “proof-of-work,” but it needs an army of computers to function.
“The main problem is that the energy consumption primarily relates to how agreement on the underlying blockchain is reached,” de Vries told Gizmodo. “Mining makes it a big competitive lottery where the winner—every 10 minutes—gets to create the next block for the blockchain. The built-in reward for this process is fixed, so it motivates participants to constantly add new machines to the network to get a bigger slice of the pie—the more computational power the more you win.”
Currently, the prize for spawning the next block of transactions is 12.5 bitcoins, and 1,800 BTCs are generated each day. At the current rate of $8,190, that amounts to $102,381. That’s a lot of money by any measure, so it makes sense for bitcoin miners, armed with their fleets of computers, to participate in this modern day gold rush. But as more and more bitcoin miners appear on the scene, the demand for electricity increases accordingly.
“Mining power is high and getting higher, thanks to a computational arms race, Roberto Frota Decourt, a cryptocurrency expert at the Unisinos Business School in Brazil who wasn’t involved in the new study, told Gizmodo. “The required number of zeros at the beginning of a hash is tweaked biweekly to adjust the difficulty of creating a block—and more zeros means more difficulty. The bitcoin algorithm adds these zeros in order to keep the rate at which blocks are added constant, at one new block every 10 minutes. The idea is to compensate for the mining hardware becoming more and more powerful.”
By “hash,” Decourt is referring to the output of a hash function—a specific algorithm used to map or simplify data. In the world of cryptocurrency, the hash rate is the speed at which a computer can complete an operation in the bitcoin code, so the higher the hash rate, the better chance a miner has of building the next block and receiving the bitcoin reward. But when the hashing gets harder, as Decourt points out, it takes more computations to create a block and thus more effort to earn new bitcoins, which are then added to circulation.
The bitcoin block mining reward gets cut in half every 210,000 blocks, so the next time this happens the reward will decrease from 12.5 to 6.25 BTC. Decourt likens bitcoin mining to a bingo game.
“In the beginning there were few participants, so players won by completing a row. This was enough to give a block every 10 minutes and each participant won X Bitcoins per day,” he explained. But the number of participants is steadily increasing, so it became necessary to make the game more difficult, he said, and now a player wins by completing two rows. “But now each participant is winning X/2 bitcoins. And you have much more participants consuming energy to produce the same amount of bitcoins. This is why the energy consumption is increasing so much. More miners does not produce more bitcoins, but they consume more energy.”
Usman Chohan, an economist at the University of New South Wales in Sydney, Australia, also not involved in the new study, said developments in bitcoin mining is a lot like the changes seen in conventional mining.
“In the old days, people had to go deep into dark, toxic mines just to pull out a few lumps of coal or gold,” Chohan told Gizmodo. “Today, enormous extractors just cleave the ground, but require much more electricity to do so.:
In the world of bitcoins, Chohan says that even a slight edge in terms of computational power can help miners secure a larger share of the distributed rewards, which is why there’s now a race to build ever more powerful—and thus energy-consuming—mining devices. It’s brutally inefficient, and that’s because the bitcoin network keeps making the calculations harder to solve, which means more and more power is required just to get the same rewards.
As noted, around 200,000 bitcoin transactions are processed each day, which means the ratio of hash calculations to processed transactions, as calculated by de Vries, is 8.7 quintillion to 1 at best.
“This is an indication of the amount of frivolous processing power now being deployed to chase an increasingly prohibitive reward,” Chohan told Gizmodo.
The power of incentive
Bitcoin is an energy hog because it needs a lot of computational power—and troves of electricity as a result—to run the hashing functions. But it’s hard to precisely measure how much electricity is actually being used. The new study is an attempt to provide some estimates around this unanswered, and increasingly important, question.
“In essence I’m taking an economic point of view to figure out where energy consumption is heading. Previous work typically looked at available hardware, and produced results that only said something about the current consumption,” de Vries told Gizmodo. “My findings were based on the current conditions, so bitcoin doesn’t need to increase in value for the conclusion to hold.”
Using economic principles, and the underlying assumption that people don’t like to lose money, de Vries sought to determine the amount of electricity that will be used once it’s no longer profitable to mine for bitcoin. Eventually, an equilibrium point will be reached when hardware and energy costs will equal the value of the bitcoin being mined. To reach his figures, de Vries looked at the hard costs of bitcoin mining, which included production information from Bitmain, a secretive Chinese company and the largest manufacturer of mining machines.
For instance, Bitmain has a mining facility in the Mongolian desert (electricity is extremely cheap in China). Armed with 21,000 bitcoin mining machines, the vast majority of which are Antminer S9 computers, the facility has been rated at between 33.3 to 40 Megawatts. This represents less than one percent of the currently global network hash rate, but it’s from information like this that de Vries was able to estimate how much of a miner’s costs are due to hardware and electricity, and when the equilibrium point might be reached.
“Ultimately this paper tells us that if bitcoin rewards miners with $X million dollars of coins mined per day, then miners are incentivized to spend up to $X million of dollars per day on…electricity, hardware, data center, labor, and so on.” Marc Bevand, a bitcoin investor and entrepreneur, told Gizmodo. “It’s not really a new finding. We knew this already.”
So Many Considerations
Indeed, Bevand doesn’t think the new paper has much to offer, and that de Vries is likely overstating the situation. He says the half-a-percent estimate by the end of 2018 is plausible, but not likely, saying de Vries’ figures are inflated on account of two errors.
“Firstly, his model assumes that manufacturers sell at cost and make zero profits, which allows miners to spend more on electricity. In reality we’ve seen Bitmain, the largest manufacturer, make billions of dollars over the last few years. No one expects them to make zero profits by the end of 2018,” Bevand told Gizmodo. “Secondly, de Vries makes the mistake of assuming that all cryptocurrency mining chips produced by TSMC [Taiwan Semi- conductor Manufacturing Company] on behalf of Bitmain end up in bitcoin mining machines. In reality many chips are fabricated for cryptocurrencies other than bitcoin. This altcoin sector is rapidly growing. It wouldn’t be unreasonable to assume that bitcoin mining machines will represent less than half of Bitmain’s sales by the end of 2018.”
According to Bevand’s own calculations, we should expect bitcoin to use 0.2 to 0.3 percent of the world’s electricity by the end of the 2018, which is about half of what’s predicted by de Vries.
“A major limitation of de Vries’ model is that it depends on guessing bitcoin’s future price as well as the cost of electricity to miners,” adds Bevand. “In the paper he assumes bitcoin maintains its current level at approximately $8,000 and electricity costs $0.05 per kWh. If either bitcoin goes up or electricity costs plummet, the energy consumption should increase, and vice versa.”
There’s also the issue of competing cryptocurrencies to consider, also known as altcoins. Bitcoin utilizes a system known as proof-of-work, but there are at least two other increasingly common ways to build cryptocurrencies: proof-of-research and proof-of-stake. Less energy intensive coins could eventually nudge bitcoin off its lofty pedestal, or undermine its value at the very least.
Another problem with the paper, according to Decourt, is that de Vries’ projections don’t consider the evolution of technologies involved in mining. He points to advances in just two generations of Bitmain’s Antminer machines, showing a 250 percent improvement in equipment efficiency in just two years.
He also pointed to the work of Canadian entrepreneur Bruce Hardy, who found a way to recycle heat generated by mining equipment to produce vegetables and even fish. Perhaps ironically, the energy poured into cryptocurrency could eventually be channeled elsewhere, and not wasted frivolously.
And finally, there’s government intervention to consider.
“The largest unstated assumption about the mining costs mentioned in this paper is regulatory, in that it assumes that bitcoin will continue to remain largely legal,” said Chohan. That said, he believes government interventions are unlikely to be effective in reducing or managing such power use, because the monitoring and oversight costs of such operations are likely to be high, and will likely involve substantial resources of time and personnel.
“It is likelier that, through the ‘market pressures’ of high input costs (i.e. electricity), many of the current mining operations will be out-competed by the largest players,” he said. “In such a situation, detection of such operations will become easier as the players will be more concentrated.”
Moving Crypto Forward
This study claims to be the first rigorously peer-reviewed article quantifying Bitcoin’s energy requirements. But as de Vries himself admits, much of his analysis was based on incomplete information, as miners are hesitant to share information like hardware and electricity usage. If anything, the new study points to the inherent difficulties in quantifying realms like bitcoin, while highlighting the need to study this area more vigorously.
Looking ahead, de Vries hopes that researchers will scrutinize and verify his results, and also collect better data. For that to happen, however, things will have to change.
“A first step would be to require openness from these mining facilities,” said de Vries. “On an international level we should look at whether we can perhaps impose some sort of carbon tax or other measure on these facilities. I don’t think this can be solved locally, since miners will just move elsewhere.”
Despite the limitations of the new study, however, it’s clear that bitcoin consumes a shocking amount of electricity. The number predicted by de Vries may be unrealistic, but even if it’s a quarter or even a tenth this much it’s still a horrendous amount of power—and the world needs to take notice. As our civilization struggles towards sustainability and a cleaner environment, it’s sad to think that a cryptocurrency could somehow get in the way. The estimate provided by de Vries may not be perfect, but it warrants a call to action; bitcoin is a beast that needs to be tamed.