Elsevier

Economics Letters

Volume 184, November 2019, 108492
Economics Letters

The Bitcoin mining breakdown: Is mining still profitable?

https://doi.org/10.1016/j.econlet.2019.05.044Get rights and content

Highlights

  • Our analysis shows that after June 2018, mining is only profitable for professional miners located in those countries where electricity costs less than 0.14$/kWh.

  • We estimate that the marginal cost of bitcoin stands somewhere around 1,952 US dollars for facilities located in the countries with the cheapest electricity costs, that also employ the most efficient producing (mining) technology.

  • According to our calculations, June 23rd 2018 was the date at which bitcoin mining was no longer worth continuing for amateur miners outside China, while November 25th 2018 was the cutoff date for professional miners using the most efficient existing devices. We predict increased centralization of professional mining facilities in China.

Abstract

We provide an updated estimation of the energy consumption of the Bitcoin network, and a calculation of the evolution of the production cost of Bitcoin over time. Using these data, we conclude that since June 2018 Bitcoin mining is no longer profitable for commodity miners without access to electricity prices below 0.14 $/kWh. This phenomenon explains why many Western miners have dropped out of the circuit, further increasing the centralization of mining activity in China. In addition, we estimate that the marginal cost of the production of bitcoin is around 1,952 US dollars. Below this price the cost of mining would not be profitable, even with the most efficient equipment and the lowest possible price for the energy required. This could lead to a massive exit of the biggest mining players, with unpredictable consequences for the future of this cryptocurrency.

Introduction

Bitcoin has been the first successful attempt in history at issuing a decentralized electronic means of payment (Nakamoto, 2009). Since its introduction in 2009, it has achieved tremendous popularity: in November 2018 there were 17 million bitcoins in circulation, with an exchange value of 90 billion US dollars.

Cryptocurrencies enjoy several desirable properties. Firstly, unlike the traditional payment methods, transactions linked to cryptocurrencies do not require the involvement of any trusted intermediary or third party. Secondly, it is not necessary to reveal the identify of the users in a transaction. The Bitcoin network solves the problem of digital identity by identifying users pseudonymously, in such a way that there is no connection between the operation carried out and the user’s public identity. Privacy is preserved. Thirdly, there is no possibility of a double expense, i.e. the same bill being used twice by the same agent (except in the case of counterfeiting or fraud). At this respect, the Bitcoin network makes public and encourages the revision of flows within the network through the mining mechanism. The mining mechanism is based on a cryptographic proof denominated proof-of-work (Preneel and Juels, 1999).

Essentially, a proof-of-work (PoW) is an easy-to-check proof of computational effort, and was originally intended for discouraging spam (Dwork and Naor, 1992). In Bitcoin, PoW is used to avoid the double-spending problem, and to build a consensus mechanism to approve valid transactions and reward miners. This construction is designed to be a computationally expensive task, because an approximation better than brute force, i.e., trial and error, is not known. When a miner eventually finds a valid solution, the transactions in a given block are permanently recorded in the blockchain, and the miner is rewarded with a fixed amount of new bitcoins (12.5 at the time of writing).

But all solutions come at a cost. In this case, the strength of Bitcoin is also one of its weaknesses. By design, the proof-of-work mechanism is computationally and energetically expensive. This is one of the greatest criticisms that Bitcoin has received in recent years (see Truby (2018)).

In this paper we address the following question: given any market price for bitcoin, what is the maximum acceptable price of electricity that would just keep the miners in the market?

In order to calculate the overall energy cost of the Bitcoin network, it is necessary to characterize the hardware equipment used for the mining process. At its inception, and until approximately mid-2011, Bitcoin could be mined using general purpose hardware such CPU, first, and latterly GPU cards, which were especially well-adapted to intensive hash calculation (Bedfor Taylor, 2017, Taylor, 2013, Malone and O’Dwyer, 2014).

From this year onwards, due to the growing popularity of Bitcoin, it was necessary to specifically design hardware, in order to provide very high hash rates with a low energy requirement. Table 1 summarizes each hardware generation, with the characteristics of the average and most efficient device in each time period (Miche Zadé, 2018). However, in any case, in this work we will only focus on the ASIC generation, which is the latest generation, the one used nowadays.

The two key factors in the characterization of the Bitcoin mining equipment are its hash rate and the power usage. The first one, denoted as Hrate, makes reference to how fast the equipment is and is usually measured in terahashes per second (TH/s). On the other hand, the power usage, P, determines the operation cost of the hardware and is closely related to its efficiency, ε, which is usually measured in Mhash/J, i.e., the number of megahashes calculated with one Joule of energy.

For example, the most efficient mining equipment currently available is Ebit Miner E11++ (ASIC Miner Value, 2019), with a power usage of 1980 W and a hash rate of 44 TH/s. Taking into account that 1J=1Ws, the efficiency of this device is ε=44106Mhash∕s1980W=22222Mhash∕J.

This study uses the historical data from the creation of Bitcoin, on 4 January 2009, until 31 December 2018, which has been extracted from QUANDL, a popular site with freely available financial datasets (Quandl, 2018).

Based on the values for price, hash rate, difficulty and reward per block, our work derives other quantities such as the cost of production of Bitcoin for different electricity prices. All these data are available as a supplementary dataset to this paper, together with the source code of the programs implemented.

Section snippets

Bitcoin variable cost of production

The first estimated quantity is the variable cost of production of a bitcoin, CBTC. To do this, we start with the cost of calculating just one hash operation with a mining equipment of efficiency ε (measured in Mhash/J), and an electricity cost of k (measured in $/kWh): Chash=εk

Taking into account that 1 J = 1 W s Chash=1εWsMhash$kWh=kε3.61012$hashtaking into account that 1 J = 1 W s the other hand, the expected number of hashes to find a valid block is D232, where D is a parameter

Maximum electricity price for system sustainability

Let us place ourselves in the shoes of a bitcoin miner. We take the fixed cost of production (buying the hardware) as sunk costs, since this is a cost that was assumed in the past and the miner cannot go back in time. The miner has to make the following decision: Either continue mining bitcoins or cease activity. This decision will entirely depend on the variable cost of production, i.e. the electricity cost. A natural question then arises: what is the maximum electricity cost for system

Conclusions

This paper has studied the evolution of the production cost of Bitcoin across time. The sharp drop in Bitcoin prices, after years of a mounting race of arms in hash power, constitutes a challenge for miners and the Bitcoin community.

We have defined a marginal cost for bitcoin linked to electricity prices and the hash calculation technology available. According to our estimations, June 2018 was a major tipping point. Since then, bitcoin mining is no longer profitable for miners whose electricity

Acknowledgments

Research supported by UAM -Grant Thornton chair on Blockchain. The authors gratefully acknowledge helpful comments and suggestions made by Simon F. Olsen form Lykke Corporation. The usual disclaimer applies.

References (12)

There are more references available in the full text version of this article.

Cited by (0)

View full text