Hass McCook is a chartered engineer and freshly minted Oxford MBA. He has been researching bitcoin over the past several months and recently joined the Lifeboat Foundation's New Money Systems advisory board.
Here, in the first in a series of articles on bitcoin's sustainability, he seeks to analyse the economic and environmental factors involved in the process of bitcoin mining.
There has been a lot of uncertainty surrounding the sustainability of the bitcoin network, with this fascinating nascent technology facing several unsubstantiated claims that it is highly unsustainable from a social, economic and environmental perspective.
I therefore undertook research to disprove or support these claims and provide an order-of-magnitude comparison of the relative sustainability of bitcoin when compared with both the incumbent banking industry, the gold production industry, and the process of printing and minting physical currency.
In fact, my research found that widely-available public information strongly refutes claims that bitcoin is unsustainable, and shows that the social, environmental and economic impacts are a minuscule fraction of the impacts that the legacy wealth and monetary systems have on both our society and environment.
The results of this research have now been turned into a series of articles for CoinDesk (of which this is part 1) which examine the economic and environmental factors of bitcoin mining.
Because bitcoin is resistant to transactional fraud and all transactions can be traced through its public ledger, there are no adverse social externalities or costs arising directly or indirectly from bitcoin mining.
Even though bitcoin addresses are pseudonymous, a good team of detectives will be able to catch a criminal who has not been professionally meticulous in concealing their steps, which is very difficult to do on a public ledger.
The slightest lapse of care will make anyone easily identifiable to authorities, and criminal detection rates will be much higher than the 1% success rate enjoyed by authorities in recovering laundered money (UN Office on Drugs and Crime, 2008).
Economic and environmental impact assumptions
The table below shows some examples of ASIC mining technology that is currently available (Source: Bitcoin Wiki, 2014), as well as energy efficiency per GH of hashing power.
Due to the fast moving and perfectly competitive nature of the bitcoin network, it is assumed that the hashrate and energy usage performance displayed by these units will soon saturate the mining market, and most likely be exceeded to a significant extent every six to 12 months.
** HashCoins have had several production difficulties since the bankruptcy of HashFast, and won't be considered further
*** Minerscube and Extolabs appear to be scam entities, due to their unusually low energy usage and hashrate for the price point, and lack of information about the companies' reputations available on the Internet. Regardless, it is not beyond the realm of imagination to expect to see these hashrates and energy performance figures in 6-12 months' time (the Extolabs performance figures even sooner), which will go further to reduce the environmental impact of the bitcoin network.
The best performing ASIC from a legitimacy, electricity and price point of view is the KnC Neptune ASIC, and the base case will assume that the entire bitcoin network is powered by them to uphold the economic laws of perfect competition and the assumptions listed above.
Environmental costs of bitcoin mining
With a network hashrate of 110 million GH/s, the network needs 0.7333 x 110 million Watts = 80,666 kW. This equates to 80,666 kW x 24 hrs/day x 365.25 days/yr = 707,120,500 kWh/year.
This equates to 2.54 million GJ/year, and 424,725 tonnes of CO2/year.
At $100/MWh, this electricity would cost $70,712,000/year.
The below table shows different energy usage, carbon footprint, and electricity costs assuming a different range of W/GH average network efficiency, but maintaining a network hashrate of 110 million.
Although the bitcoin mining industry should be efficient in theory, and the largest miners would be expected to have the most efficient equipment, this is not easily provable. The difference between the most efficient and least efficient miners is quite clear, and it makes sense that the large professional miners continually reinvest profits in updating their equipment multiple times per year.
For the purposes of this order of magnitude comparative study, I will assume that the industry is not efficient, and average network energy efficiency is at the high end, 1.1 W/GH – 30% less efficient than the KnCMiner Neptune units.
This generous allowance will also cover the impact of producing the ASICs, as several studies show that the gross majority of impact made by electronics happens during their use, and not during production.
Also, 98% of electronic waste is completely recyclable (MRI, 2014).
Economic costs of mining
The mining cycle is difficult to interpret since it depends on the market price of bitcoin.
Similar to large gold miners, when market price of the underlying asset drops, miners tend to hold their assets to restrict supply, causing an eventual increase in price. Miners who can't afford to do this typically shut off their equipment, and exit the mining game.
When market price increases, this draws more miners into the game, increasing network hashrate and difficulty, which requires further capital expenditure from incumbent miners, which also leads to higher operating costs.
So long as market price exceeds mining cost, miners will enter the market, and so long as mining costs exceed the market price, miners will either leave the game, or withhold supply – just as physical commodity miners do.
Difficulty increases have been fairly consistent over the past year, with typical fortnightly hashrate increases of between 10% and 20% (BitcoinWisdom, 2014). Because of this, the useful life of most mining equipment is only about three to six months.
Theoretically, breaking the 110 million GH/s hashrate down into equivalent KnCMiner Neptune ASIC units (3,000 GH each), this results in approximately 36,670 units, at a price of $9,995 each – a total of $366.5m.
Assuming this has to be spent twice a year, capital expenditure (CAPEX) of around $733m needs to be invested in the network. At the current block reward of 25 bitcoins per 10 minutes, 1,314,900 bitcoins are mined per year. This equates to yearly capital expenditure of $733 m/1.3149m bitcoins = $557.45 per bitcoin.
Using the less efficient CoinTerra miner as a benchmark network average, yearly CAPEX would equate to $659.9m per year, or $501.85 per coin.
For the purposes of this report, I will take an average figure of $530 per coin, or $696.9m.
As calculated earlier, yearly electricity operational expenditure (OPEX) for a CoinTerra-average network would be $106.1m, or $80.69 per coin.
Using KnCMiner consumption figures, running costs per coin would be $70.71 million, or $53.77 per coin
For the purposes of this report, I will take an average figure of $67.23 per coin, or $696.9 million.
Adding the CAPEX and OPEX figures results in a cost to mine a bitcoin of $597.23, and a total yearly cost of $785.3m. Interestingly, this is the exact bitcoin price at time of writing.
It should be expected that price of bitcoin should grow proportionally with the cost of network CAPEX and OPEX based on hash rate from this point forward.
This goes a long way to explain the cyclical bubble nature of bitcoin's market price, and gives us insights into local minimum prices after a burst bitcoin cycle bubble.
Now we've looked at the cost of bitcoin mining, it's time we compared it with the cost of generating other stores of value. Stay tuned for the next article in the series, in which Hass McCook examines the sustainability of gold mining.
Environment image via Shutterstock