This Bitcoin mining report recommended by Fidelity gives you a comprehensive understanding of the current status of hash rate and electricity

Data comprehensively explains the current and future of Bitcoin mining hash rate and electricity.

Written by: Sam Doctor, Chief Strategist at BitOoda, a digital asset financial services company Translated by: Perry Wang

Bitcoin mining is fundamental to the Bitcoin network and is integral to Bitcoin as an asset. Despite its importance, mining has been one of the most opaque and least understood parts of the Bitcoin ecosystem. This report from BitOoda aims to increase transparency into the makeup of Bitcoin miners , ultimately helping people understand the current state and health of the system. At the Fidelity Center for Applied Technology (FCAT), we look forward to continuing our research in the field of Bitcoin mining and helping to move the ecosystem forward. We thank the BitOoda team for their work and hope that it will help increase everyone’s understanding of this complex and fascinating part of the Bitcoin ecosystem.

Juri Bulovic, Director of Blockchain Products, Fidelity Center for Applied Technology

Key takeaways

• The total global Bitcoin mining power capacity is 9.6 GW (gigawatts), of which about 50% may be located in China; the United States accounts for about 14%; the hash rate utilization rate is about 67%

• The median electricity cost for BTC mining is 3 cents/kwh (kilowatt-hour), and the mining cost of 1 BTC is about $5,000

• In the next mining machine upgrade cycle, the hash rate could reach 260 EH/s in the next 12 months and 360 EH/s in 24 months, but this would require capital expenditures (Capex) of $6.3 billion, leaving a funding gap of $4.1 billion between the cash flow generated by the industry

• For every 10 EH/s increase in hash rate, the BTC price needs to increase by $1,000 to have a neutral impact on revenue per MWh.

• Cheap electricity, BTC prices, and semiconductor shipments are the three key factors that affect our forecasts.

Bitcoin mining is a secretive industry with little public information . We found that even among the most knowledgeable cryptocurrency investors, there is a gap between their understanding of the mining industry and the potential investment opportunities in this field. Although Coinmetrics, Coinshares and the Cambridge Center for Alternative Finance have published some research reports, many questions remain unanswered. This research report, commissioned by the Fidelity Center for Applied Technology (FCAT), complements existing research, builds on previous work, and attempts to answer some new questions.

Bitcoin Power Capacity Analysis: How Much, Where, and What Price

In this section, we attempt to measure, locate, and evaluate miners’ power capacity and estimate miners’ profitability. We conducted more than 60 conversations with miners, mining machine manufacturers, and distributors, and referenced more than 45 public data sources to try to provide as complete a picture as possible to understand how much Bitcoin mining capacity there is, where it is located, and how much miners pay for electricity.

We then further explored the following questions: How will mining capacity grow in the future as a function of available electricity and mining machine efficiency, and what factors such as BTC price, capital/financing availability, semiconductor technology and performance will limit Bitcoin mining capacity.

Bitcoin mining industry expected to generate at least 9.6 GW of electricity

We arrived at the 9.6GW figure based on the following logic: Bitcoin mining hash rate peaked at 136,098 PH/s on May 10, before the Bitcoin reward halving, and fell to a low of 81,659 PH/s on May 17. We acknowledge that these extremes may be due to luck to some extent - such as lucky time periods or fast block times - which may artificially inflate hash rate estimates, while slow block times may be due to bad luck. Regardless, we exclude luck from the model and make simplifying assumptions to estimate an approximate value for the electricity consumed by the Bitcoin network.

We assume that all hashrate still running at the trough on May 17th came from the more profitable new generation miners - S17-class miners , including Bitmain's Antminer S17, T17, Shenma's M20 , and equipment from Canaan Technology, Innosilicon, Ebang and other manufacturers. We also assume that all equipment shut down between May 10th and May 17th was the old generation of less profitable S9-class miners (such as Antminer S9, Shenma M3).

Please note that we use " S17 class ", "S9 class" and "S19 class" as a general term to include Bitmain 's mining products as well as competitors with similar configurations . The reason we only use Bitmain's models to define the class is that Bitmain has historically dominated the "S9 class" era and has a slight advantage in "S17 class" devices. We also set the Power Usage Effectiveness (PUE) to 1.12 in all relevant calculations, which means that for every 1 MW of electricity used directly for mining, 120kw of electricity is required to run other equipment, including cooling systems, lighting, servers, switches, etc.

Figure: BitOoda divides mining machines into the following levels

Source: BitOoda, Bitmain, Canaan, MicroBT, Halong, GMO, AsicMinerValue.com

The data in the figure below shows that if all the BTC hashrate in operation on May 17 was the newer S17-class miners, then the total electricity consumed by these miners was 3.9GW. Further, if the 54EH/s of hashrate that was shut down between May 10 and May 17 were all older generation S9 miners, then this could explain another 5.7GW of electricity.

These simplified assumptions help us understand the industry more broadly, knowing that the reality is that most of the miners that shut down are S9-class miners , but not all; the small portion of hashrate that remains operational during the trough period may come from S9 miners in certain areas where electricity is very cheap. The reduction in equipment profitability after the Bitcoin halving is the main factor driving the reduction in mining capacity, and another part of the reason is that miners happened to be transferred from northern China to southern China during this period to take advantage of cheap hydropower (see Part 2 for more details on the impact of China's hydropower season). Based on these assumptions, we estimate that the Bitcoin mining industry has access to at least 9.6GW of electricity.

Figure: Bitcoin hash rate and electricity consumption at recent peaks and troughs

Source: BitOoda, Blockchain.com, Kaiko, Coinmetrics

We estimate that the BTC mining industry uses about 67% of the 9.6GW of available electricity, growing at about 10% annually, driving 2.8 million dedicated Bitcoin mining machines. Most of the current mining machines are S17-class, but most of them will likely use the next-generation S19-class machines in the future. Most of the hash rate that has come back online after the May 17 trough is from S9-class machines, which are either operating in areas with extremely low electricity prices or were delayed in being relocated from northern China to Sichuan and Yunnan to take advantage of extremely low electricity prices during the summer flood season .

In addition, despite supply chain delays, the new generation of Antminer S19 and Whatsminer M30 miners have begun limited shipments, and some S17-class miners are also on the way, which have partially contributed to the recovery of the hash rate.

Figure: Bitcoin hash rate, electricity consumption, and number of mining machines installed (data as of 7/1/2020)

Source: BitOoda, Blockchain.com, Kaiko, Coinmetrics

It is estimated that about 50% of the global Bitcoin mining capacity is in China; the United States accounts for about 14%

We used a variety of public sources, as well as confidential interviews with miners, mining machine manufacturers, and distributors to understand the geographic distribution of BTC mining capacity and how much miners spend on electricity. We were able to identify 153 mining farms with a total power of approximately 4.1GW , including 67 mining farms with a power capacity of approximately 3GW, based on anonymous surveys.

Figure: Geographic distribution of mining capacity surveyed vs. estimated total capacity of 9.6GW

Source: BitOoda estimates, Miners, ASIC makers / resellers, public sources

Conversations with miners, mining machine manufacturers, and resellers lead us to believe that we have accounted for the majority of mining capacity in the United States, Canada, and Iceland , but only a small portion of mining capacity in China and the “rest of the world”. In our conversations with miners, we asked them not only about their own capacity, but also how many other mining farms they were aware of in that market, and what total mining capacity they thought was in the region. We know these are approximate, but it helps us find a useful way to estimate the geographic distribution of mining capacity.

Figure: Geographic distribution of mining capacity surveyed vs. estimated total capacity of 9.6GW

Source: BitOoda estimates, Miners, ASIC makers / resellers, public sources

We estimate that 50% of Bitcoin mining capacity pays no more than 3 cents/kWh for electricity, a figure that has been steadily declining over the past few years. Previous evidence suggests that this figure was closer to 6 cents/kWh in 2018. As the network hash rate has increased, revenue per PH/s has fallen, and miners with higher electricity costs have either moved to lower-cost areas or simply shut down their machines.

Figure: Electricity cost curve: Electricity cost vs. share of network capacity

Source: BitOoda estimates, Miners, ASIC makers / resellers, public sources

Our cost curve gives a median cash cost of about $5,000 to mine 1 BTC, with an upper confidence range of about $6,000. This estimate is cash operating expense and does not include depreciation or other expenses for mining hardware.

The curve also shows that a small portion of BTC is mined at a cash cost above the current BTC spot price. We believe this uneconomical mining behavior is driven by multiple factors, such as power purchase commitments , potential incentive payments to shut down capacity during peak electricity demand periods, and the need to acquire Bitcoin in jurisdictions where trading options are limited or expensive.

Figure: Cost to mine 1 BTC based on network capacity at different electricity costs, data as of July 1, 2020

Source: BitOoda estimates, Miners, ASIC makers / resellers, public sources

We note that at the current network hash rate, S9-class miners need electricity costs below 2 cents/kWh to break even, and as hash rates continue to increase, lower electricity prices are needed to maintain their viability. Our cost model assumes that one employee is required to operate 5MW of capacity. Since S9-class miners are less energy efficient than new miners, more equipment is required per PH/s of hash rate, and to achieve the same hash rate, their power consumption, number of employees required, and management costs are all higher than new miners.

An S19-class mining machine only needs 30kW of electricity and 9 devices to get a hash rate of 1 PH/s. If an S9-class mining machine is used, about 70 devices are required, the power consumption exceeds 100kW, and the corresponding maintenance and operation manpower costs, and power management costs are increased to get the same hash rate of 1 PH/s.

Figure: Daily revenue and cash operating costs of different mining machines at various electricity prices at the current hash rate. Note: When estimating the proportion of electricity used for BTC mining, we assumed a PUE of 1.12.

Source: BitOoda estimates, Blockchain.com, Kaiko, Coinmetrics

Labor costs refer to the number of maintenance and operation personnel required to run a large-scale mining facility (>50MW). The data comes from our conversations with miners.

Summary: We estimate that there is about 9.6GW of available Bitcoin mining capacity, with current utilization in the mid-60-70% range. The median unit price of electricity for this total capacity is about 3 cents/kWh, and the median cash cost of mining 1 BTC is $5,000. We estimate that China contributes about 50% of mining capacity ; the United States is second at about 14%. A significant portion of China's mining capacity migrates with the seasons to take advantage of low electricity prices during the flood season, which we will explore in detail in Part 2.

We have some surprising conclusions about the relationship between hash rate growth and China’s flood season

We found that China contributes 50% of Bitcoin mining electricity consumption and network hash rate. Below we will further examine the situation of the Chinese Bitcoin community and the impact of China's flood season on Bitcoin price and network hash rate.

What is the flood season ? Southwestern provinces such as Sichuan and Yunnan receive heavy rainfall from May to October each year . This causes a large amount of rainwater to rush into dams in these provinces, causing hydroelectricity production to surge during this period. Since there is an oversupply of power generation, this electricity is sold cheaply to Bitcoin miners. Overflowing dams need to drain the excess water, so selling electricity at low prices is a win-win for hydropower plants and miners. This low price of electricity attracts miners to migrate from other provinces to take advantage of this. In the dry months, miners pay about 2.5–3 cents/kWh for electricity in northern China, but in the wet season from May to October, the electricity prices paid in Sichuan and Yunnan are less than 1 cent/kWh .

We disagree with the conventional wisdom that low electricity prices during the flood season drive hash rate growth. We believe that the flood season moves the cost curve downward for 6 months of the year, so that fewer BTC need to be sold to support operating expenses, and miners can accumulate capital to invest in mining capacity growth.

As shown in the chart below, there is a significant difference in the increase in average BTC price during the wet and dry seasons, while the growth in hash rate is roughly the same during these two periods. Our chart shows the growth in each period, and finds that the first two periods are likely outliers (further supporting our view), with the average based on a small sample size of the next 11 6-month periods.

Figure: Hash rate and BTC price, segmented by wet and dry seasons, Note: Data since 2014; average excludes the period from November 2013 to October 2014; data as of 7/1/2020

Source: BitOoda, Blockchain.com, Kaiko, Coinmetrics

Capital accumulation is followed by equipment purchases, equipment delivery, and deployment, a dynamic that can be seen in the correlation between rising prices (which support capital accumulation) and growth in hash rate 4-6 months later as the supply chain takes time to deliver the equipment customers have purchased.

China's flood season will bring a downward trend in the electricity cost curve , which will help miners accumulate capital and help promote the growth of hash rate in the future. The increase in capital accumulation will reduce the mining industry's demand for external funds, and the industry itself can also support the future growth of hash rate.

Figure: Correlation between BTC price changes and hash rate changes, Note: Data for the past 12 months, as of 7/1/2020

Source: BitOoda, Blockchain.com, Kaiko, Coinmetrics

We look at the correlation between BTC price changes over the past 15 to 360 days and hash rate changes over the same period last year. We note that hash rate follows price changes with a 4-6 month delay and is highly correlated. This creates a dynamic relationship where capital accumulation is followed by the purchase, delivery, and deployment of new mining machines, and the supply chain takes time to complete the delivery of mining machines.

Available and underutilized power capacity, capital accumulation from within the industry (helped by China’s flood season), external financing, and reduced revenue per PH/s all have an impact on future growth of hash rate. We will explore the future of hash rate in Part III.

Bitcoin hash rate growth forecast: how high, when, why, and what will slow (or speed) it up

We delve deeper into questions such as how much the network hash rate can grow, what factors are supporting this growth, and the capital and financing constraints that could slow this expected growth.

According to our assessment, the Bitcoin network hashrate could exceed 260EH/s in the next 12-14 months as power capacity modestly increases from 9.6GW to 10.6GW , and as mining machines are upgraded, with newer S17 and next-generation S19-class machines replacing older generation S9-class machines. The increase in power capacity takes into account the available power at each mining farm, planned infrastructure spending, and the fact that some higher-cost mining farms may have to shut down operations due to revenue pressures.

Figure: Bitcoin hash rate and power consumption. Note: When estimating the proportion of power actively used for Bitcoin mining, we assumed a PUE of 1.12. Data as of 7/1/2020

Source: BitOoda estimates, Blockchain.com, Kaiko, Coinmetrics

The industry will upgrade to S19-class mining machines by mid-2022. Completion of this cycle could bring the network's hash rate to about 360EH/s . We estimate that the next fundamental equipment upgrade may not occur until the second half of 2022, although mining machine efficiency will continue to gradually improve during this period. We note that if BTC prices remain flat or fall, revenue per PH/s in US dollars will continue to decline to the marginal cost point. The next round of investment and hash rate growth may slow significantly - therefore, our hash rate forecast chart may be delayed, or never realized.

We looked at TSMC 's progress in mining chips and compared it with Samsung and Intel - although Intel does not produce mining ASIC chips . Available data shows that there are huge differences in process technology between different semiconductor suppliers. We note that the next leap in ASIC technology will be the development of 5 nanometer (nm) technology . At this node, TSMC, Bitmain's main supplier, is ahead of Samsung. Although TSMC has received a large number of orders at the 7 nm and 5 nm nodes, its process geometries appear to be similar to Intel's 10nm node.

We believe that Samsung also has a tighter process geometry; therefore, Samsung closely follows TSMC . ASICs are primarily logic chips, so comparisons with Intel also make sense. As the semiconductor industry evolves, we notice increasing differentiation in functional geometry, so that even at the same nominal process node, there are major differences between different chip manufacturers in terms of die density, feature size, and ultimately power and thermal performance.

Figure: Comparison of Intel and TSMC’s process technology

Source : https://www.eetimes.com/intels-10nm-node-past-present-and-future/

Samsung recently announced that the commercial mass production plan of the 3nm process node may be delayed until 2022, and 5nm may become the main production in 2021. We believe that the lack of 3nm production capacity and the possible low output in the early stage will cause the 5nm process to become the main ASIC development and production before 2022. For these reasons, we believe that S19-class miners will account for the majority of miner shipments in the next 24 months , although incremental design improvements can also improve energy efficiency, which will be reflected in new models of machines in this series.

Figure: Once power capacity is utilized and the mining machine upgrade cycle is completed, the growth of hash rate (below) will slow down. Note: When estimating the proportion of power actively used for Bitcoin mining, we assumed a PUE of 1.12. Data as of 7/1/2020

Source: BitOoda estimates, Blockchain.com, Kaiko, Coinmetrics

As shown above, the Bitcoin network can also reach a hash rate of 360EH/s , assuming modest growth in network power capacity and widespread deployment of S19-class mining machines. Improvements in power efficiency (fewer watts per TH/s) could have a positive impact on these forecasts, but a key issue is that the amount of BTC earned per PH/s or per MWh is decreasing - the daily flow of BTC remains roughly constant, fluctuating only with additional blocks and transaction fees. Therefore, if the network hash rate increases, the share of the total hash rate that an individual miner earns will decrease, and their share of the BTC flow will also decrease. If the price of BTC fails to keep up with the growth in hash rate, profitability will decrease, and a new equilibrium may be established at a hash rate that is significantly lower than the value we predict.

The following chart shows the number of BTC obtained per MWh for each level of mining machine: The number of BTC obtained per MWh by an S19 mining machine is roughly equivalent to 3 times that of an S9-level mining machine.

Figure: BTC earned per MWh as a function of network hash rate, data as of 7/4/2020. Power consumption data includes a PUE of 1.12

Source: BitOoda estimates, Blockchain.com, Kaiko, Coinmetrics

The chart below shows how BTC per PH/s has changed (and is expected to change in the future) as a function of the network hash rate over time, factoring in block rewards before and after the halving. Here you can also see that revenue in BTC is declining . Without taking equipment into account and just looking at PH/s, it is clear that BTC price is a key factor in the continued growth of hash rate over time.

Figure: Daily BTC earned per PH/s, over time, and as a function of network hash rate, data as of 7/1/2020, historical data from 1/1/2018

Source: BitOoda estimates, Blockchain.com, Kaiko, Coinmetrics

The dollar value of the BTC mined decreases over time, making it less profitable unless the BTC price increases enough to offset this. As shown in the chart below, the daily revenue earned per PH/s is a function of both the network hash rate and the BTC price . The current Bitcoin network hash rate target is about 124EH/s, and the current BTC price is $9,220, which translates to about $70 per PH/s per day. If the network hash rate increases to 260EH/s, which we expect to reach in the summer of 2021, the BTC price would need to be around $19,500 for the daily revenue per PH/s to remain at the same $70.

If the BTC price is $10,000 by then, the daily revenue per PH/s will be only $36. The middle chart below shows that the energy-efficient S19-class mining machine needs to spend about $37 in cash to achieve 1 PH/s per day at an electricity cost of 4 cents/kWh, but the cost of running an S9-class mining machine at 4 cents/kWh is $133. Even if the BTC price reaches $10,000, the S9-class mining machine still needs to run at an electricity cost of less than 0.5 cents/kWh to break even.

Figure: Daily revenue and cash operating costs for each mining machine tier at different electricity prices under several different scenarios of future hash rate and BTC price. Note: When estimating the proportion of electricity actively used for BTC mining, we assumed a PUE of 1.12.

Source: BitOoda estimates, Blockchain.com, Kaiko, Coinmetrics

The significant capital expenditure required to achieve the potential hash rate will be a limiting factor, especially if the increase in BTC price does not keep pace with the hash rate, and the cash generated within the industry will at least be limited, which can only further increase the dependence on external funds . Moreover, this will also have a negative impact on our hash rate forecast: due to the uncertainty of plans and the decline in investment return expectations, higher cost miners have to exit operations, which will also limit the inflow of external capital.

What if the BTC price remains the same? At what point does the hash rate stop growing? If the electricity price is 1 cent/kWh, the S9 miners can continue to run until the network hash rate reaches 180EH/s. When the electricity price is 3 cents/kWh, the S19-class miners can continue to run until the hash rate reaches 295EH/s. Beyond this point, the S19-class will need higher BTC prices or lower electricity prices to maintain operations. However, these devices will not be able to recover their capital costs at a hash rate of 295EH/s. Obviously, the price increase of BTC is already factored into the capital budget of every miner.

Figure: Daily revenue and cash operating costs per PH/s as a function of network hash rate. Note: When estimating the proportion of electricity actively used for BTC mining, we assumed a PUE of 1.12.

Source: BitOoda estimates, Blockchain.com, Kaiko, Coinmetrics

The capital expenditure (Capex) required to increase the hash rate to 260EH/s in the next 12 months will total $4.5 billion . If the hash rate increases to 360EH/s by mid-2022, it will cost about $2 billion more .

Figure: Bitcoin hash rate and power consumption. Note: When estimating the proportion of electricity actively used for BTC mining, we assumed a PUE of 1.12.

Source: BitOoda estimates, Blockchain.com, Kaiko, Coinmetrics

If the BTC price rises steadily to around $19,000 at a rate of more than 40% per year over two years, then even at an electricity cost of 5 cents/kWh, S19-class mining machines can still operate, but there will still be a funding gap of $4.1 billion between the total capital expenditure of the entire industry and its internally generated cash flow.

Figure: Bitcoin network capital expenditures and internally generated cash flows, data as of 7/1/2020; Y axis is logarithmic scale

Source : BitOoda estimates, Blockchain.com, Kaiko, Coinmetrics

We note concerns that our hash rate growth model requires a large number of new miners to be delivered, and we have received some questions about whether our forecast is feasible. Approximately 60,000 miners need to be shipped per week for the installed base to grow in line with our hash rate forecast. For comparison, Bitmain was able to deliver over 95,000 S9 miners per week in the first half of 2018. Although the number of chips/die size in the S19-class miners is uncertain, we believe that semiconductor/assembly capacity will not be a limiting factor.

In conclusion, we believe that the Bitcoin network hash rate will reach 260EH/s within 12 months and 360EH/s within 24 months. However, this prediction is premised on the fact that according to our model, the BTC price needs to rise, or appreciate by 25–35% annually. We do not model or predict the future price of BTC, but only depict the impact of possible price scenarios on the growth of hash rate, electricity consumption, and capital investment and profitability of the mining industry .

Variables outside this range could delay or accelerate hash rate growth. BTC price and available external funding to bridge the funding gap are two potential constraints that will affect whether the industry can increase Bitcoin mining capacity to 360EH/s, but the production capacity or assembly capacity of semiconductor chips required for mining machines will not be a limiting factor.

Investors need to take these forecasts into account when evaluating mining projects, and keep an eye on the price of Bitcoin. At BitOoda, we are strong proponents of hedging and recommend that investors adopt an active hedging strategy to reduce operational risk - we often say that miners know their expenses for the next 6, 12 and 24 months, but they don't know how much BTC they can mine or what it will be worth. Hedging strategies can help reduce operational risk and stabilize cash flow. Please contact us at [email protected] to view our full report and obtain research methods, sources and more details on the information provided in this article, or to discuss risk management strategies and trading opportunities that we can work with.

BitOoda is a fintech and financial services company focused on digital assets, providing BTC and ETH derivatives, helping companies develop, manage and execute risk management strategies through structured derivatives and proprietary investment products. BitOoda provides financial engineering, research and consulting, intermediary brokerage and risk management services.

Sam Doctor, the lead author of this research report, is the Chief Strategist at BitOoda. Sam Doctor's signature research reports include project evaluations and Bitcoin mining research and analysis focusing on profitability and risk factors. Doctor has 18 years of experience in securities, strategy and cryptocurrency research. He previously worked at JPMorgan Chase's New York and Asia offices and most recently served as the head of data science and quantitative research at crypto research firm Fundstrat Global Advisors. Doctor holds an MBA from the Indian Institute of Management Ahmedabad and previously graduated from the University of Mumbai in Electronics and Semiconductor Engineering. Mr. Doctor holds FINRA Series 7, 63, 86 and 87 business qualifications.