The Bitcoin Mining Network- Trends, Marginal Creation Cost, Electricity Consumption previously it has been as high 99.6 yet as of 2017 it still remains as high as 91.5 [42]. We therefore assume the price to be the same as our previous estimate.] Bitmain was in fact pricing the S9 according to attempts from other players to enter the market. During this period, where the price is marked heavily marked down, Halong Mining were taking their first round of orders for their Dragonmint Miner “T1” and Ebang Technologies and Canaan mining were taking orders for their latest developments towards the end of the timeframe. The hypothesis would be that they were trying to influence the marketas has been levelled against them multiple timesin an attempt to suppress the competition. The reason for the short time period is the fact that Bitmain only hosts old URL’s for batch prices for a short period of time and beyond that timeframe we would rely on forum literature which is hard to find. Compare [40] with [41] where in the er you can no longer click through Bitmain’s link to a particular batch release and in the latter you are taken to a specified batch. [Update November 2018 Having spoken to a few large miners and researched extensively online we have discovered purchases made in bulk between 250 and 300 dollars. However, Bitmain does make a significant amount of its sales to smaller miners, specifically through its e-shop. The price on the shop is currently 307 and 294 for the respective ‘I’ and ‘J’ models. More recently the price has been hovering around the 500 mark, and from reading Bitmain’s IPO prospectus it is clear they have been making the majority of their sales earlier in the year. Thus, we have taken the price to be about 400 on average for this next leg of the year between May and November 2018. Considering the number of units sold has more 12Batch S9 Price BTC PriceMay 18 US1,222 US9,000Feb 18 US2,725 US9,250Jan 18 US1,995 US13,500Dec 17 US2,320 US16,500Nov 17 US1,265 US8,000Oct 17 US1,500 US5,250Sep 17 US1,750 US4,750Average price US1,825 US9,464Table 11 Overview of Archived Batch Prices of the Bitmain S9 Mining Unit.Innosilicon T2 Turbo US 1,350 – 7/10 This is sourced from the BitcoinTalk overview of currently competitive hardware [38]. The price is pulled from the website of the retailer at the last time available. Therefore, we are quite confident this at least accurately represents the retail price even if it does not capture the second-hand prices. For less popular miners such as this there are not enough second-hand sales to get a good idea of secondary market pricing. MicroBT’s Whatsminer M10 US 1,441 – 7/10 This is sourced from the BitcoinTalk overview of currently competitive hardware [38]. The price is pulled from the website of the retailer at the last time available. Therefore, we are quite confident this at least accurately represents the retail price even if it does not capture the second-hand prices. For less popular miners such as this there are not enough second-hand sales to get a good idea of secondary market pricing. Total Mining Units Bitfily 1,000 – 3/10 This estimate is low because the amount of ination available is equally small. We therefore have little to no ination about sales. Having said that, considering the mediocre specifications of this hardware there is nothing to indicate this unit has sold much more than 1,000 copies. Bitfury Block Box 448 – 4/10 Here we use market estimates of approximately 12 of total hashrate 28 exahash as stated by the CEO of Bitfury to reverse-arrive at 448 by using stated efficiency figures. [Update November 2018 We have carried the assumption over but scaled the assumptions certainty down by a factor of one to reflect the inevitable decrease in certainty as time passes from the last data point.]Avalon 921 US 742 – 7/10 Assuming the officially stated price is accurate. [Update November 2018 We have been unable to find any new price ination as previous ination was taken from Chinese sources. This is significant because the vast majority of Canaan’s customers are domestic; previously it has been as high 99.6 yet as of 2017 it still remains as high as 91.5 [42]. We therefore assume the price to be the same as our previous estimate.] Ebang E10 US 1800 – 7/10 This is sourced from the BitcoinTalk overview of currently competitive hardware [38]. The price is pulled from the website of the retailer at the last time available. Therefore, we are quite confident this at least accurately represents the retail price even if it does not capture the second-hand prices. For less popular miners such as this there are not enough second-hand sales to get a good idea of secondary market pricing. GMO Mining B2/B3 US 1,999 – 9/10 This price is listed on their website [43]. DragonMint T1 US 1,199 – 7/10 This assumption is based on the price of batch 1. We know of no further batches but have seen a lot of people exchanging these for lower prices and so have altered the price down to reflect this as we believe there are not a lot of them around that second-hand market is heavily indicative of the actual price as there’s little evidence of any large-scale miners using these. This is also to match the price on the Bitcointalk.org.13the abundance of wind and solar in the dry season Xinjiang, Northwest [45][46] and the hydropower of Sichuan and Yunnan in the rainy season Southwest [45]; and lastly that they have facilities like it elsewhere in China and the world such as in Anhui and Newfoundland [20]. [Update November 2018 This figure has been brought forward despite Bitmain’s assumed making and selling of units. Their IPO documents state that their private mining operations are limited [4] and we have therefore assumed no additional gear added.] Antminer S15 Private Bitmain Facilities 20,000 – 4/10 Although this miner has only just been announced it is well known that Bitmain does not mind mining on gear before it is released to the public and so we have assumed that it has a significant amount of these mining already. Avalon 841 225,000* – 6/10 This is estimated from extrapolating inferred production runs given the reports of total amount of hardware sold in Canaan’s IPO application [42]. Avalon 921 75,000 – 6/10 This is estimated from extrapolating inferred production runs given the reports of total amount of hardware sold in Canaan’s IPO application [42]. Ebang E10 200,000 – 6/10 This is an estimate back-calculated from the percentage of hardware claimed by Ebang to be produced in 2017 and carried forward. This figure is independently calculated by a company commissioned to do the work for their IPO application [47] and they expect that Ebang accounted for 11 of hardware produced in 2017. This is mentioned on various occasions, first on Private Bitfury Facilities 112 – 6/10 This assumption is reverse calculated from Bitfury investor presentations stating 132 megawatts and subtracting off the known Hut 8 units leaving Bitfury’s own facilities. [Update November 2018 This figure has just been brought forward but knocked down a point as Bitfury have released a new chip and sold some Block Boxes publicly [44] and thus presumably a few privately as well. Bitfury Tardis 1,000 – 5/10 The Bitfury Tardis is a very new miner with their new ‘Clarke’ chips and so it is assumed very few have been sold so far. The only sale we know of is the aforementioned one to Hut 8. Bitfury x Hut 8 85 – 10/10 This ination is available to us by email from Hut 8 and as a publicly listed company we have strong reason to believe this is entirely accurate. Antminer S9 Publicly Available Units 2,150,000* – 7/10 Bitmain’s S9 and other very similar hardware from Bitmain T9’s and all other versions of the S9 are widely assumed by many mining experts and large scale miners to be the vast majority of the network at about 2/3 of all miners in their efficiency class. Antminer S9 Private Bitmain Facilities 230,000 – 6/10 Here we base our assumption on remarks from Bitmain employees and interviews from Quartz articles on Bitmain https// all worth reading and Chinese news sources covering Bitmain. The Chinese sources suggest that the mine in Xinjiang is ‘three times’ the size of the Ordos mine of 25,000 machines; that the Xinjiang mine and the Sichuan and Yunnan mines have a migratory cycle based on 14Hashrates and Power Efficiency per Unit All except GMO Mining – 9/10 This represents a tempered belief in the state of the producers which will have modified only slightly if we believe the real-life specs are different e.g. reading published reviews or forum reviews of trusted members acknowledging there to be a large disparity between the advertised spec and the testing spec. GMO Mining – Hashrates taken from company filings [5]. *Due to the recent drop in hashrate we have assumed 20 of all Bitmain S9s and Avalon 841s to have shut down.page 1. The reason we do not consider it entirely reliable is that as we have seen little evidence of a market share of that magnitude, and that the report supporting the IPO application documents was commissioned by Ebang themselves. GMO Mining B2/B3 16,000– 8/10 As with the previous estimate for GMO we are grateful for their transparency and from multiple public documents their aggregate number of machines deduced from their total hashrate and the efficiency of their hardware. DragonMint T1 25,000 – 3/10 We have low confidence in this figure but we wanted to include an estimate nevertheless. There was a widespread need for a Bitmain competitor and in anticipating this, miners bought up all of the Halong mining products unseen and with a minimum order size of 5 units. At such a small batch size estimate the figure has minimal impact on overall calculations. Innosilicon T2 Turbo 10,000 – 3/10 As with the other smaller companies it is very hard to gather much ination to make a reliable estimate as to the number of miners out there. MicroBT’s Whatsminer M10 25,000 – 3/10 As with the other smaller companies it is very hard to gather much ination to make a reliable estimate as to the number of miners out there. However, the efficiency of this miner is very impressive for its release date and there is significant forum support. Therefore, we have assumed they have put out a maximum output due to the community response. 15[14] Hydrominer, “Hydrominer,“ 2018. [Online]. Available https//www.hydrominer.org/. [15] Coin Factory, “Coin Factory,“ 2018. [Online]. Available https//www.coin-factory.at/. [16] L. Tassev, “B,“ 2018. [Online]. Available https// [17] NR Mine, “NR Mine,“ 2018. [Online]. Available https// [18] HashBrasil, “HashBrasil,“ 2018. [Online]. Available https//.br/. [19] Stratum, “Stratum Blockchain Tech,“ 2018. [Online]. Available https//stratum.hk/. [20] R. Antle, “CBC,“ 27 April 2017. [Online]. Available https//www.cbc.ca/news/canada/newfoundland-labrador/great-north-data-bitmain-technologies-lawsuit-1.4084925. [21] Office of the Chief Economist, “World Bank,“ May 2018. [Online]. Available http//documents.worldbank.org/curated/en/293821525702130886/pdf/125990-WP-7-5-2018-9-39-58-ECAEconomicUpdateCryptocurrenciesandBlockchainMayweb.pdf. [22] Moonlite, “Moonlite,“ 2018. [Online]. Available https//www.moonlite.io/. [23] Genesis Mining, “Genesis Mining,“ 2018. [Online]. Available https//www.genesis- [24] Northern Bitcoin, “Northern Bitcoin,“ 2018. [Online]. Available [5]. [25] N. Marinoff, “Bitcoin Magazine,“ 22 MARCH 2018. [Online]. Available https// [26] B. Akolkar, “B,“ 23 August 2018. [Online]. Available https// [27] JIWO, “JIWO,“ 2018. [Online]. Available https//jiwo.ru/stolica-rossiyskogo-mayninga-irkutsk. [28] Sweans Mining, “Sweans Mining,“ 2018. [Online]. Available https// Works Cited [1] C. Bendiksen and S. Gibbons, “The Bitcoin Mining Network Trends, Marginal Creation Cost, Electricity Consumption and Sources,“ 21 May 2018. [Online]. Available www.coinshares.co.uk. [2] A. Poelstra, 22 March 2015. [Online]. Available https// [3] C. Mora, R. L. Rollins, K. Taladay, M. B. Kantar, M. K. Chock, M. Shimada and E. C. Franklin, “Nature,“ 29 October 2018. [Online]. Available https// [4] BitMain Technologies Holding Company, “HKEX News,“ October 2018. [Online]. Available http//www.hkexnews.hk/APP/SEHK/2018/2018092406/Documents/SEHK201809260017.pdf. [5] GMO Internet Group, “GMO Cryptocurrency Mining,“ 2018. [Online]. Available https//mining.gmo.jp/en/. [6] Hut8 Mining, “Hut8 Mining,“ 2018. [Online]. Available https// [7] Hive Blockchain Technologies, “Hive,“ 2018. [Online]. Available https// [8] Bitfarms, “Bitfarms,“ 2018. [Online]. Available https//www.bitfarms.io/. [9] DMG Blockchain Solutions, “DMG,“ 2018. [Online]. Available https// [10] Coinmint, “Cision,“ 05 June 2018. [Online]. Available https// [11] J. Gogo, “B,“ 2018. [Online]. Available https// [12] S. Haig, “B,“ 2018. [Online]. Available https// [13] S. Das, “CCN,“ 2018. [Online]. Available https// 16[43] GMO Miner, “GMO Miner,“ 2018. [Online]. Available https// [44] Hut8 Mining Corp, “Globe News Wire,“ 12 November 2018. [Online]. Available https// [45] M. Jiansheng, “China Smart Grid,“ 8 March 2017. [Online]. Available http// [46] kidneybeans, “Bitcoin86,“ 12 January 2018. [Online]. Available http// [47] http//www.hkexnews.hk/APP/SEHK/2018/2018062101/Documents/SEHK201806240005.pdf, “HKEX News,“ 2018. [Online]. Available http//www.hkexnews.hk/APP/SEHK/2018/2018062101/Documents/SEHK201806240005.pdf. [29] D. Prindle, “Digital Trends,“ 27 April 2018. [Online]. Available https// [30] Giga Watt, “Giga Watt,“ 2018. [Online]. Available https//giga- [31] Plattsburgh BTC, “Plattsburgh BT