The electricity consumption to mine Bitcoin is currently equivalent to roughly 600,000 households’ usage or US$3.4m spent every day, this will double by end-2018. Current mining costs for Bitcoin imply a marginal cost of production of US$2,250, says James Butterfill, Head of Research & Investment Strategy at ETF Securities. Including rig purchase costs, marginal costs are US$4,300. Current Bitcoin prices appear to be justified only on the basis of mass adoption, which at this stage remains low.
Is Bitcoin useful and how can it be valued?
These are probably the two most important questions when trying to ascertain if Bitcoin is a viable investment. The first question at this point is very difficult to answer, it is similar to a start-up company in that it is potentially a great idea but there is not yet a big enough market to prove its viability. There are early signs of its potential, it has a following and is well known now, retailers are beginning to offer it as a form of payment and some cantons in Switzerland are now accepting it as payment for taxes. We believe cryptocurrencies are an emerging digital asset that has potential given its compelling concept, but is not proven yet, and there remain some very valid concerns over its volatility and its current valuations. Accordingly, like investing in a start-up company, investors should remain cautious.
What type of asset is it?
Bitcoin is a very hard asset to categorise. It has some features of a currency, it is intrinsically a medium of exchange, but it is not currently as stable as the US dollar and has similar volatility to some frontier market currencies. We see it as also having some similar features to a commodity; it is mined using valuable resources to extract (energy and computer hardware), and it is a finite resource. It is debatable whether there is any point to mining it, but a very similar rationale could be applied to gold, given that very little is used for industrial usage (gold does have good industrial properties as highlighted in Disruptive themes behind future commodity demand, but at current prices application in industry are low). Ultimately miners and investors of gold see it as being a store of value and therefore worthwhile, as is emerging for Bitcoin. Bitcoin is essentially mined and minted with megawatts rather than with shovels and minting dies.
How to value Bitcoin
There are varied ways to value Bitcoin, but given it has some similarities to commodities we felt it would be worthwhile calculating the marginal cost of production. While this varies for commodities as supply and demand changes, it is an effective way in understanding the long-term equilibrium price. Bitcoin is exceptional in that the supply is predictable, being determined by the structure of its underlying algorithm. Bitcoin’s algorithm dictates that after a specified number of blocks are mined the reward for mining halves. A linear path for the Bitcoin reward schedule has been established, and this is likely to continue as long as Moore’s law4 for exponential growth in processing power continues. The last coin is likely to be mined by 2130, but 99% will be mined by 2027.
The speed of mining could be accelerated but depends on the success of quantum computing, which could theoretically solve the Bitcoin algorithm far quicker, however this may come at the cost of considerably higher electricity consumption.
The hash rate growth of the Bitcoin network, a measure of the speed at which Bitcoin blocks are mined, coupled with the known power consumption can be used to estimate the electricity consumption costs, the equivalent of the marginal cost of production that is often used to value commodities.
Estimating historical power costs
Bitcoin emerged in late 2009 with enthusiasts mining on their personal computers, which at the time became profitable but was a very inefficient approach. Consequently manufacturers began selling dedicated ASIC5 miners which drastically improved efficiency. These ASIC miners began emerging in 2013, with each new model having a much more powerful hash rate, leading to an explosion in overall network hash rate as more miners joined the network. The power consumption of these commercial Bitcoin miners is well known, as is the overall hash rate. Using historical global power costs we estimate the electricity consumption to currently be 1.5GW/hr, which equates to roughly 600,000 households usage, or US$3.4m spent every day. At the current rate of growth in the Bitcoin network, power consumption costs will be double that of today by end-2018.
Marginal cost of production
To estimate marginal cost of production the total number of coins produced per day is divided by the mining cost: at current consumption and production levels this is approximately US$2,250, well below the current price. However, if the cost of purchasing the Bitcoin mining hardware is factored-in, and assuming a two year replacement cycle, the current marginal cost of production would be US$4,300.
Predicting future marginal cost
As the future network hash rate is likely to follow Moore’s law and the mining difficulty follows a linear path future, electricity costs can be estimated. The scatter chart highlights this close relationship between price and mining difficulty.
Assuming the historical relationship between mining difficulty and costs hold, we believe by the end of 2018 power consumption will be double that of today. We anticipate the marginal cost of Bitcoin will have risen to US$4,230 or US$6,500, including hardware purchase costs by the end of next year. It isn’t until end-2019 that marginal costs would align to today’s price. However, by early 2020 the reward for mining Bitcoins (as dictated by the Bitcoin algorithm) will halve, pushing marginal costs to roughly US$16,000.
Using marginal cost of production is just one approach at valuation, it could be argued that the current high valuations are justified because even if the probability of mass adoption is small, the impact on price would be very large, this is perhaps why we are seeing so much speculation. Another approach would be to use Metcalfe’s law, which states that the value of a network is square the number of users, but the number of users is difficult to determine. At least now, we know what the costs are.
1) The observation that the number of transistors in a dense integrated circuit doubles approximately every two years
2) ASIC – Application-specific integrated circuit