Mining serves an essential function in the Bitcoin protocol by securing the distributed network consensus through proof- of- work. The immutability of the Bitcoin blockchain is a direct result of the cost of mining as any attacker attempting to rewrite or append fraudulent transactions to the blockchain would need to acquire and operate enough hash power to outpace the entire honest network. The combined capital and operational expenditures of such an endeavour, combined with its dubious benefit for the attacker, makes such attacks prohibitively expensive to undertake in practice.
Using provable work as a mechanism for establishing distributed consensus is still a novel and uncommon approach to systems requiring reliable synchrony between participants, such as monetary applications. Even so, over the last five years alone the Bitcoin mining industry has grown from a sector dominated by hobbyists to a multi-billion-dollar industry with individual participants whose profits match those of multinational industrial conglomerates
Running a relevant Bitcoin mine is now an undertaking on the order of operating a large- scale data centre. Thousands of individual mining units often featuring multiple circuit boards containing many dozens of chips are needed to even make a dent in the Bitcoin hashrate. Mines must secure industrial-sized power supplies to run not only the mining hardware itself, but also their substantial cooling requirements. Modern large-scale mining operations often require power supplies ranging into hundreds of megawatts (MW) and the total mining network is estimated to draw multiple gigawatts (GW).
While much criticism has been levied at the energy expenditure of proof-of-work systems it is in fact this energy expenditure that keeps the system secure. There have been multiple previous attempts at quantifying Bitcoin’s energy use and while some have been well-founded (2), other frequently cited attempts have been less accurate (3). In this paper we examine the current and projected size, composition and energy expenditure of the combined Bitcoin network as well as its associated costs. Using these figures, we arrive at a range of estimated marginal costs of Bitcoin creation, given a range of assumptions, before finally taking a closer look at the network’s energy sources and rough geographical distribution.
Due to the limited nature of publicly available data relating to Bitcoin mining, in the making of this paper we have been forced to adopt a range of assumptions. We consider this paper our first iteration of several where we will continue to improve on both models and assumptions. Within the limits of our knowledge we have set these assumptions as close as possible to what we believe to be the actual figures, but caution readers that no matter how well-founded these assumptions are, they are still assumptions. Where deemed valuable to the reader, we have performed sensitivity analyses to show how our calculated results are affected by varying the assumptions. The remainder of this section will shed some light on our rationales for making these various assumptions whereas full documentation and deeper explanations can be accessed in the Appendix.
First, we begin with our sampling range. We have chosen to sample all publicly known Bitcoin mining hardware with shipping dates after 1 January 2014. The year of 2014 widely considered the beginning of the industrial era of Bitcoin mining as signified by the advent of large-scale deployment of mining hardware featuring Application Specific Integrated Circuits (ASICs), designed purely for SHA-256 hashing. While we acknowledge that some Bitcoin ASICs were released before this date, widespread industrial- scale mining operations were uncommon, and even the largest mines rarely exceeded single digit megawatts (4). In line with our hardware sampling range, when extrapolating the future hashrate, hardware efficiency and hardware costs, we have calculated our regressions from data in the same time range.
Second, we have been forced to make assumptions with regards to BITCOIN HASHING INV. that is not related to the pure electrical demands of running the hardware. Such BITCOIN HASHING INV. non-exhaustively includes rent, cooling cost, maintenance and administration. Due to the largely private nature of most large-scale Bitcoin miners, such figures are – for obvious reasons – not publicised. In this instance we have chosen to rely on figures from comparable non- mining data centres and the educated guesses of individuals involved in the mining industry. Rather than attempting to know the unknowable, we instead perform a sensitivity analysis with a considerable input range to showcase the effect of a large assumption variance on overall marginal costs of creation.
Third, there exists no reliable source of the total amount of deployed mining hardware. We have therefore made assumptions based on a combination of various publicly available information and industry estimates, and overall worked within reason to estimate figures that correspond with the pseudo-measurable hashrate.
Finally, and relevant to all previous assumptions, we are often forced to assume that people are telling the truth. We recognise that the Bitcoin mining industry is full of unknowable information for participants residing outside of industrial entities, poorly researched opinions, and outright misinformation, sometimes even on the part of manufacturers advertising performance and total market share. In this setting it is important to keep in mind that none of the relevant ASIC manufacturers are publicly traded entities (although the main chip foundry is), and listed miners with strict disclosure requirements are only just getting their feet wet.
Thus, we have chosen the approach – to the maximum extent possible – of don’t trust, verify. When that approach inevitably fails, we examine the information available, judge the sources on their merits and only include data that falls within a reasonable level of rationality. Again, when assumptions make significant impacts on our calculations, we perform sensitivity analyses to illuminate their effects. Since they cannot be wholly avoided we prefer instead to be perfectly clear with regards to their overall influence on results.
Based on our best estimates for market- wide electricity BITCOIN HASHING INV. we have chosen $0.05/kWh as our mid- range value. On top of that we estimate an average of 30% of electricity BITCOIN HASHING INV. as cooling and other (C&O) BITCOIN HASHING INV. to cover all other costs including, but not limited to, rent, cooling, staffing and mining pool fees. To that we should add that we believe, and have anecdotal evidence to suggest, that 30% is at the higher end of the cost spectrum, making it a highly conservative number. Furthermore, we estimate that the average mining gear is depreciated (CAPEX horizon) over 18 months. A further discussion of the evidence to support these assumptions can be found in the Appendix.
Official Website : https://bitcoin-hashing.com/
Facebook : https://www.facebook.com/Bitcoin-Hashing-618720298778267/
Telegram : https://t.me/BitcoinHashing
Instagram : https://www.instagram.com/bitcoin_hashing/