The Three-Body Problem of Bitcoin Mining: What is the “Law of the Dark Forest” in the mining industry that has grown wildly for ten years?​

, Author: Leo Zhang, founder of Anicca Research, Compiler: Brother Xiao Mao, published with authorization.

The more transparent things are, the more mysterious they are. The universe itself is transparent. As long as you can see, you can see as far as you want, but the more you look, the more mysterious you are.

Liu Cixin "The Three-Body Problem"

Bitcoin mining presents a situation that combines hardware and software, energy and financial markets. Unseen rules affect every aspect of the mining industry. The effectiveness of mining by each miner depends on a variety of different external factors, most of which are difficult to quantify and are almost impossible to outline and predict.

From a macro perspective, we can identify three main internal factors that drive the development of the entire crypto mining industry: supply mechanisms, climate cycles, and hardware iterations, each of which will affect different parts of the miners' profit calculation formula:

• mining profit

• = mining revenue - mining fee

• = (block reward + handling fee) * Bitcoin price * computing power of miners / computing power of the whole network - (electricity fee + hardware depreciation)

The token supply mechanism affects block rewards (revenue);

Hardware iterations affect miners' computing power, efficiency ratio, and hardware depreciation (capital expenditure).

When Bitcoin was halved for the first time in 2012, most people used home computers and GPUs for mining, and the computing power of the entire network was scattered all over the world. At this time, the dominant force in the market was only driven by the halving of Bitcoin block rewards . As the mining output ratio decreases, the computing power of the entire network plummets, and it temporarily turns to Litecoin, which is more profitable. By the time of the second halving in 2016, commercialized ASIC mining machines and industrialized encryption mining farms have begun to be put into use.

Ahead of Bitcoin's third halving this May, analysts in the crypto market debated the potential outcome. Some analysts speculate that the price of Bitcoin will double due to the impact of supply and demand. Some analysts also speculate that the computing power of the entire network will be reduced by 20-30%. At the same time, due to the coincidence of the timing of the halving, the transition from the climate cycle to the wet season, and the upgrading of mining machine chips from 16 nm to 8 and 7 nm, these three factors will be affected in this halving. Interact and jointly affect mining revenue.

Token Supply Mechanism

Bitcoin is the product of computing power. The foundation of the mining industry lies in the ability to continuously motivate miners to invest in hardware and consume energy to consolidate the ability of the Bitcoin network to guarantee transaction settlement. In the absence of an active market for block space, block rewards are the main source of income for miners. Mining revenues have grown significantly over the years due to rising bitcoin prices, which in turn has shaped mining into a multi-billion dollar industry.

Unlike most tangible commodities, Bitcoin's contract clearly specifies a four-year halving schedule. Every day a varying percentage of the new bitcoin token supply reflows into the rest of the bitcoin economic network. Since miners are the only natural supplier in the Bitcoin ecosystem and also the largest and most stable group of sellers, the profit margin of mining is a key factor in determining the supply side of Bitcoin.

Estimated returns from June to December 2020, based on 6.25 BTC per block and prices from January to May, source: blockchain.info

Obviously, the halving of Bitcoin block rewards will directly reduce the output ratio of miners. Some older generation machines may not boot due to reduced power. During these periods, the buying and selling and consolidation of mining machines will often take place. Miners with cheap power resources will buy some of the older generation of mining machines in bulk at low prices, and the secondary market for mining machines will become more active.

Miners are bivariate call options on the physical form of Bitcoin. The complexity of mining machine pricing and the difficulty of transportation make the second-hand mining machine market very opaque, which is manifested in poor liquidity and high reliance on industry personnel. However, poor price discovery sometimes creates huge arbitrage opportunities for some miners. For example, in late 2018, the price of Bitcoin dropped sharply to the $3,000 level, and a significant percentage of hashrate dropped significantly, sparking numerous headlines discussing a "bitcoin mining death spiral." Some miners took this opportunity to find a large number of Antminer S9s for sale at low prices. In the short four months since then, the price of Bitcoin has entered a frenzied bull run. Not only do these buyers profit handsomely from the bitcoins they mined, but the resale price of these mining machines has also increased by 3 times. This shows that even miner hardware that cannot be turned on still has the nature and value of an option.

image descriptionSource: Jinping GouIn a report released in March

source:, the team expects parts of the network to consist of the following diagrams:

source:

Understand the game process of miners' behavior in Bitcoin halving

image description

*Antminer S9i specifications (13.5TH/s, 1,310W, unit price $21), Antminer S17+ specifications (70TH/s, 2,800W, unit price $1,232); **S9i average age is more than 18 months, S17 + The average age is more than 36 months

Based on the above data, we can calculate the ratio of daily miners selling Bitcoins to the daily output of Bitcoins at various price levels, and the ratio of daily miners selling Bitcoins to the global daily trading volume:

The price range is relatively narrow because the global transaction volume varies greatly with price changes. *6.25 BTC per block. Transaction fees are not included. **Global trading volume of $1.14 billion, source: Bitwise

Note that the data for this analysis are for illustrative purposes only. The above data is based on very rough assumptions: all old mining machines are represented by the Antminer S9, and the new generation of mining machines is represented by the Antminer S17. It goes without saying that miners buy machines from multiple manufacturers, and different manufacturers have different specifications for the same generation of miners. In reality, the composition of the market varies considerably more. At the same time, this analysis is based on a static point in time. The composition of mining machine hardware is very unstable. As the market seeks a new balance after halving, mining machines are actively changing hands. The upcoming wet season is also having a huge impact on electricity costs for most miners. In addition, as more and more miners use financial instruments such as mortgages, futures, and even computing power markets, the selling pressure of miners on the Bitcoin network will be partially relieved or delayed.

By observing the changes in the constituent elements of the mining industry over time, one conclusion that can be drawn is that the generation of computing power is "unequal". Power is unique.

climate cycle

Climate cycles are one of the reasons why the mining industry is geographically concentrated. Over the years, bitcoin mining has inadvertently benefited from massive, if not excessive, investments in hydropower in southwest China. The surplus of cheap electricity, huge power generation capacity, cheap labor costs, and close proximity to mining machine manufacturers make China's southwestern region an ideal place for Bitcoin mining. It is estimated that more than 65% of the global computing power is concentrated in the following provinces.

image description

Bitcoin mining map, source: Emerging Financial Center Cambridge

The booming encryption mining industry has brought benefits to the local power plant business, and many people have built or converted their facilities into computing power centers for mining machine hosting business.

Source: WeChat group

Gradually, the crypto mining industry is restructuring around these climate patterns. Like some ancient ritual, every year before the flood season, the mining industry holds a large-scale crypto mining summit in Chengdu, the capital of Sichuan province. We’ll also see: Some facilities are only open to external customers during high water periods; miner makers plan new product launches before they ship; and miners race to buy the latest and greatest mining rigs.

The growth of Bitcoin computing power during the wet season:

image description

Source: coinmetrics.io

At the beginning of this year, investors and miners returned to a more cautious approach to investing due to macroeconomic uncertainty. Due to the impact of the new crown epidemic on the supply chain, the production of new mining machines has also been suspended. After the Bitcoin halving, more mines shut down, making the supply of mining machine hosting greater than the demand. Many factories in areas such as Sichuan and Yunnan have difficulty finding customers, and this situation continues to lower currency prices. Compared with last year's cost of 0.24-0.26 yuan/kWh for electricity, the estimated average cost at the beginning of the year can be as low as 0.10-0.20 yuan/kWh kWh.

Lower electricity costs will partially offset the impact of Bitcoin’s block reward halving, with many mining farms having to sign contracts with power plants committing to using a minimum amount of electricity. In order to attract business, some companies offer a "joint mining" plan. Miners pay the minimum monthly fee and distribute part of their own mining revenue to hosting service providers, effectively transferring part of the market risk to hosting providers.

But moves like the one above don't mean absolute guarantees of lower operating costs. While hydroelectric facilities are less expensive during high water seasons, these mines are often located in more remote areas where power supplies and internet connections can be unstable and there is sometimes a risk of damage from flash floods.

A few months ago, Sichuan announced the first batch of "hydropower consumption demonstration enterprises", including several bitcoin mining companies. The bad news is that this year has not seen as much rainfall as expected, despite more crypto mining facilities starting operations. At the time of writing, with the scorching heat of summer looming, electricity from the hydroelectric plant is primarily used to keep Sichuan residents fed.

But recently, an area called Muli County in Sichuan issued the "Notice on Reporting the Status of Virtual Currency "Mining" Activities" on May 21. Although many local people do not want to implement the order, it also shows that Fengshui Due to the complexity of the energy situation in the current period, it is not a panacea to expect to reduce miners' electricity bills during the wet season.

Climate cycles are a rather unique phenomenon at this stage of cryptocurrency mining. If mining operations migrate to more diverse locations in the future and are no longer concentrated on a single power supply, the factors of the climate cycle will no longer play such an important role.

secondary title

Iteration of Mining Hardware

In 2017, Shenma mining machines sold 90,000 units, accounting for 7.2% of the total network computing power at the end of the year. In 2018, the computing power of new Shenma mining machines accounted for 9% of the total network computing power at the end of the year. Power accounted for 35% of the total network computing power at the end of the year.

Miner products are a commodity, and while there are specific factors that help miner manufacturers stand out, such as customer service, delivery time, supply chain management, etc., competition forces miner manufacturers to focus relentlessly on two key metrics: price and energy efficiency ratio.

The energy efficiency ratio of the mining machine is measured by the power consumption per unit computing power. The lower the indicator, the less power is consumed for mining. Each generation of mining hardware defines new products by improving the energy efficiency ratio of the mining machine.

image description

Improvements in energy efficiency of mining machines have a significant impact on profitability. Using the current Bitcoin difficulty and the average price of a used Antminer S9i ($21 per machine, EER 97.0J/TH), we can calculate the days to breakeven at various price levels:

The following is the calculation of the latest antminer S19 Pro payback days ($2,407 per machine, energy efficiency ratio of 29.5 J/TH per machine):

Antminer S9 is currently unprofitable in most cases, but miners with extremely cheap power resources can still make a profit. Comparing the two tables above, we can see that below $0.04/KwH, the payback period for miners mining with S9 is faster. Thanks to cheap power supplies, when a new generation of mining machine products is launched on the market, the previous generation of products may not immediately withdraw from the market. The team at Coin Metrics aggregated the estimated usage of Antminer S7 and S9 over a period of time based on the random number distribution. As we can see, Antminer S7 usage is not just going down as the S9 gains popularity:

As mentioned in the "Token Supply Mechanism" section above, miners with cheap power resources will buy some older generation mining machines in bulk at a low price to earn higher profits. This arbitrage opportunity usually arises when the hash rate is relatively high and the price of Bitcoin is relatively low. Although the value of hashrate is largely determined by the price of Bitcoin, there is a discrepancy between hashrate and price due to the delayed hardware market reaction to financial market movements. Manufacturing customized chips will incur one-time engineering costs, and miner manufacturers and their supply chain partners need to amortize this one-time cost according to the number of chips manufactured. Unless the mining machine manufacturer is a long-term cooperative customer with a large number of orders, in most cases, the wafer foundry will require the mining machine manufacturer to pay in full for the reserve wafer order.

Source: BitInfoCharts

Wafer order lead times are typically 12 to 13 weeks, making it more challenging for manufacturers to develop forward-looking business plans. At the end of 2017, many manufacturers did not have enough inventory to sell to the market during the bull market, thus misjudging the duration of the bull market. Manufacturers, including Nvidia, overestimated their order intake in 2018 and had to gradually liquidate inventory at a loss in the second half of the year.

image description

Source: BitInfoCharts

Once the mining market completes the upgrade from 16nm chips to 7nm chip mining machines, the competition for back-end product technology will stop for a long time. The average lifespan of new mining machines will be extended from the previous 2 years to 3 to 4 years, and the failure rate of mining equipment will become a more important factor. In fact, the failure rate, price, and energy efficiency ratio are considered to be key mining machine evaluation indicators.

Frequent mining machine failures will reduce the mining machine's computing power and quickly increase maintenance costs after the warranty period ends. A critical breakdown means lost revenue for maintenance. For example, due to flaws in heat dissipation design, the damage rate of the Antminer S17 series is as high as 20%-30%. The cause of the failure was that the heat sink was loose or displaced, which caused the short-circuit shutdown of the computing power board. The cost of repairing this failure is very expensive-although the Antminer S17 series is based on advanced manufacturing processes, the products are not satisfactory.

More advanced mining hardware means higher upfront capital expenditures, which in turn drives the further industrialization of crypto mining. Soon, more data centers will define their own infrastructure to suit specific mining needs. Surplus natural gas energy mining, immersion liquid cooling and internal monitoring solutions will start to emerge.