One article to understand the turbulent decade of China's mining chip manufacturers

Abstract: In 850 AD, a delegation led by the Buddhist monk Wuzhen set out from Dunhuang and finally arrived in Chang'an, the capital of the Tang Empire after two years. In the Purple Palace of Daming Palace, Wuzhen brought the good news from thousands of miles away to Tang Xuanzong Li Chen - the recovery of Shazhou. Since the beginning of the An-Shi Rebellion, the Tang court's border defense was empty, and Tubo took the opportunity to occupy the land west of the Yellow River. When Zhang Yichao recaptured Shazhou (now Dunhuang) in 848, the Tang Dynasty had lost this land for nearly a hundred years.

In 850 AD, a delegation led by the Buddhist monk Wuzhen set out from Dunhuang and finally arrived in Chang'an, the capital of the Tang Empire, after two years. In the Purple Palace of the Daming Palace, Wuzhen brought the good news from thousands of miles away to Emperor Xuanzong of Tang, Li Chen - the recovery of Shazhou. Since the beginning of the An-Shi Rebellion, the Tang court's border defense was empty, and Tubo took the opportunity to occupy the land west of the Yellow River. When Zhang Yichao recaptured Shazhou (now Dunhuang) in 848, the Tang Dynasty had lost this land for nearly a hundred years.

The Tang Dynasty, which had fallen from its peak, was no longer able to send troops to support Zhang Yichao's Guiyi Army and could only sit in Chang'an to commend and reward them. However, under the leadership of Zhang Yichao, the Guiyi Army successively recovered more than ten states including Gua, Yi, Liang, etc., gained more than 4,000 miles of land, and a population of one million. They drove all the Tubo forces out of the Hexi Corridor and restored the majesty of the Tang Dynasty. In order to commemorate Zhang Yichao's historical achievements in recovering Hexi, local people excavated a large cave in Mogao Grottoes (now Cave 156) and painted the famous "Zhang Yichao Leading the Army on a Journey".

However, such a hero with great achievements was forgotten in official history because he was far away from the Central Plains. Whether it is the New and Old Book of Tang or the New and Old History of the Five Dynasties, there are only a few records about Zhang Yichao. Today, the detailed historical materials about Zhang Yichao and the Guiyi Army are mostly from the letters and inscriptions in Dunhuang, which makes people sigh.

Back to modern times, against the backdrop of global trade disputes, semiconductors have become the focus of attention of the Chinese people. Whether it is the national large fund with a scale of hundreds of billions or the more than 20,000 new chip start-ups in the past six months, it shows that the chip industry has become a hot spot in people's eyes.

However, just as people only remember the prosperous Tang Dynasty, which was full of resources and abundant natural resources, but forgot the Guiyi Army, which shook the mountains and rivers and recovered thousands of miles of land. In the current hot chip industry in China, there is also a forgotten lone army. This lone army is the mining machine chip manufacturers represented by Bitmain and Canaan Creative. They have created a record of mass-producing the world's first 7nm chip and monopolized nearly 90% of the global mining machine chip market share. As one of TSMC's top ten customers, the order volume once exceeded Huawei HiSilicon; but it is precisely this lone army that has been excluded from the major rankings of the domestic chip industry for a long time, unknown to the public, and without policy support.

Perhaps many people will sneer at the mining chips involved in cryptocurrency, but compared with the semiconductor industry, which has been supported by the country at great expense for nearly three decades but is still weak, domestic mining chip manufacturers have defeated many foreign manufacturers in less than ten years and monopolized the global market share. What basis do we have to mock this lone army?

The success of China's mining chips is by no means the low difficulty of mining chip development as some people say, otherwise why are all foreign mining machine manufacturers eliminated? This is by no means accidental, because in addition to Bitmain and Canaan Creative, there are also many Chinese manufacturers such as Microbit and Wuhan Xindong. In the field of mining chips, we can not only see legendary stories like Bill Gates dropping out of school to start Microsoft, but also see the great era wave that has been stirring for the past decade.

Ideal Times

In November 2008, more than a month after the outbreak of the global financial crisis, Satoshi Nakamoto published a famous paper in the MIT email discussion group - "Bitcoin: A Peer-to-Peer Electronic Cash System"; then in January 2009, the Genesis Block of Bitcoin was mined on a small server in Finland, announcing the official birth of Bitcoin.

In 2011, Zhang Nangeng, who was still studying at Beihang University, first came into contact with Bitcoin. At that time, Bitcoin was still in the GPU mining era. In the Bitcoin world, the higher the computing power of the miners, the more Bitcoins they mined. For this reason, Zhang Nangeng, who had a background in circuit design, designed the world's first FPGA mining machine. Later, Zhang Nangeng promoted his FPGA mining machine in the Bitcoin forum bitcointalk under the ID "ngzhang" and began to become famous in the industry.

However, FPGA mining machines are gradually unable to meet market demand, and people are beginning to turn their attention to the field of application-specific integrated circuits (ASICs). To this end, in June 2012, the Butterfly Labs in the United States fired the first shot in developing Bitcoin ASIC mining machines. If the Butterfly Labs ASIC mining machine is successfully developed, then most of the computing power of the Bitcoin network will be in their hands, and they will have the ability to launch a 51% attack, and Bitcoin will no longer be safe.

In order to break the monopoly of Butterfly Labs, Zhang Nangeng, who claimed that he was "defending the world peace of Bitcoin", also began to develop mining machines. However, Zhang Nangeng was studying for a doctorate at the time, and his heavy academic workload made it impossible for him to focus on the development of ASIC mining machines. For this reason, Zhang Nangeng made a bold decision - to drop out of school and start a business.

At that time, the news that Butterfly Labs was developing ASIC mining machines also reached another talented young man in China, Jiang Xinyu. Jiang Xinyu, known as "Baked Cat", was admitted to the junior class of USTC with the 11th place in the country that year, and then went to Yale University for further studies in 2011. At Yale University, Jiang Xinyu first came into contact with Bitcoin and quickly became interested in it. He then dropped out of Yale and registered his own company Bitfountain in Shenzhen in July 2012, focusing on the research and development of Bitcoin ASIC mining machines.

In the first half of 2013, Jiang Xinyu and Zhang Nangeng successfully developed mining machines and delivered the first batch of mining machines. At this time, Butterfly Labs was still leisurely, and the chip design problem had not been solved, and the delivery date was far away. In September 2013, Avalon mining machine chips were mass-produced and shipped on a large scale, and soon 800,000 chips were put on the market; Jiang Xinyu's Baked Cat mining machine also occupied 20% of the computing power of the entire Bitcoin network at this time, becoming the biggest winner in the field of mining machine chips. At this point, the ASIC mining era officially arrived, and domestic mining machine chip companies took the lead in this field.

Many years later, looking back on this period of time, many people would find it incredible - chips have always been a weak industry in my country. But why were Chinese mining machine chip manufacturers so successful in the early days? In addition to the "human" factors such as Zhang Nangeng's "burning the boats" and Jiang Xinyu's "extraordinary intelligence", there are two more reasons: the particularity of the mining machine chip market and China's engineer population dividend.

(1) The particularity of the mining chip market

In the field of chip design, there has always been a misunderstanding in the market: the bottleneck of domestic chip design lies in backward technology, and taking the Huawei incident as an example, it points out that my country is currently subject to others in terms of EDA software and underlying architecture. However, before 2018, my country's chip design companies were not greatly affected in the use of EDA software and the authorization of underlying architecture, but there were still only a few chip design companies that grew and developed.

In fact, the bottleneck of my country's chip design industry is not only the technical barriers, but more importantly the lack of market support. The embarrassment of most chip design companies in my country is that no customer is willing to use the chips they design, and the problems of the chips can only be discovered and solved by continuous use in practice.

Take Huawei HiSilicon as an example. Huawei phones began to be equipped with HiSilicon-designed chips in 2012. However, the performance of HiSilicon's K3V2 chip was extremely poor at the time, causing Huawei's Ascend D series of mobile phones to be criticized by consumers as soon as they were released: severe heating and slow loading. Even so, Huawei still chose to insist on using HiSilicon chips and continuously optimize them. By 2018, HiSilicon's Kirin 980 chip performance had reached the world's top level. Without Huawei's mobile phones' unconditional support for its own chips, there would be no Huawei HiSilicon today.

Compared with Huawei HiSilicon's accumulation of strength, mining chip manufacturers are more fortunate. This is mainly due to the particularity of the mining chip market: first, most of the mining equipment at that time were GPUs, and the ASIC field was still a blank in the market, which was a new demand and new product in the market; and large companies such as NVIDIA and Qualcomm did not bother to enter this field, which allowed new entrants to avoid competition with industry giants and have the opportunity to grow bigger and stronger.

Secondly, mining chips have strong financial attributes. Ordinary chips are mainly used for various production and life activities, while mining chips can generate cash flow income through mining. In terms of computing power, ASIC is hundreds of times higher than GPU of the same era. Higher computing power means higher mining income: at that time, an Avalon mining machine priced at 8,000 yuan could earn nearly 70,000 yuan a day through mining.

For ordinary chips, users care more about the performance of the chips; but for early mining chips, as long as users can buy a mining machine, it means considerable profits, and the performance of the mining chip itself was not so important at the time. Therefore, once the ASIC mining machine was launched, it was immediately popular in the market, and while the mining machine manufacturers were selling the mining machines, they could continue to optimize the mining machine chips, thus forming a positive cycle.

(2) China’s Engineer Population Dividend

The success of Chinese mining chip manufacturers is largely attributed to China's demographic dividend of engineers. ASIC mining chips are essentially the hardware implementation of the SHA265 algorithm, and the design is not difficult. Especially in the early stage from 0 to 1, it does not require the participation of top talents in the field of integrated circuits. In the past two decades, my country's colleges and universities have cultivated a large number of talents with science and engineering backgrounds, forming a demographic dividend of engineers. According to official statistics, the number of master's graduates in science and engineering in my country has increased from 20,000/year in 2000 to 200,000/year in 2018, with an annual compound growth rate of nearly 25%.

The large group of engineers in the past decade may not be the top in the industry, but they have obvious advantages in labor costs and efficiency. This was fully reflected in the early competition with the American Butterfly Laboratory. When the ASIC mining machine solution was first proposed, the domestic Avalon mining machine used 110 nm process technology, the baked cat mining machine used 130 nm process technology, and the American Butterfly Laboratory used 65nm process technology from the beginning. As the eyewitnesses recalled at the time: compared with the domestic developers who were working day and night to develop mining machines with a high fever; the American developers were all people who loved life and would not focus all their energy on the research and development of mining machines. Butterfly Company opened reservations in June 2012, but did not deliver until a year later, while domestic mining machine manufacturers had already started large-scale shipments.

Industrial Era

The mining machine chip market in 2013 belonged to the "twin stars" era of Zhang Nangeng and Jiang Xinyu. Domestic mining machine chip manufacturers relied on the engineer bonus to get ahead at the starting line, but this advantage was not solid, because the design of mining machine chips in the initial stage was not difficult and the entry threshold was very low. Due to the extremely high profits in the early mining machine market, many new mining machine manufacturers emerged in China, and a number of foreign manufacturers such as Russia, Sweden, and Japan were eyeing them. In the fierce market competition, how did domestic manufacturers win the final victory? This has to talk about a famous company - Bitmain.

Compared with the famous Bitmain, the experience of its founder Wu Jihan is also very legendary. Wu Jihan, who graduated from Peking University with a degree in economics, worked in the financial industry in his early years. He began to pay attention to Bitcoin in 2011 and did a lot of work for it: he became the moderator of the Bitcointalk Chinese community, translated Satoshi Nakamoto's Bitcoin paper into Chinese for the first time, and founded the first Bitcoin information exchange platform in China, Babbitt, with science fiction writer Chang Chi. In April 2013, Wu Jihan resigned from the financial industry and founded Bitmain with Zhan Ketuan, who has a background in integrated circuit design.

Under the leadership of Zhan Ketuan, Bitmain quickly developed a 55 nm mining chip and launched the Antminer S1 in November 2013. With its excellent performance, Bitmain quickly occupied a part of the market share. However, the good times did not last long. In December 2013, the cryptocurrency market began to enter a cold winter, and the price of Bitcoin fell from a peak of $1,100 to a low of $200.

In the long cold winter of 2014, the mining machine market was also greatly affected, and a large number of mining machine manufacturers went bankrupt. The "genius boy" Jiang Xinyu also did not survive this winter. In early 2014, Jiang Xinyu launched the third-generation chip BE300, but it had serious quality problems, resulting in unsalable chips. Jiang Xinyu, who suffered heavy losses, then disappeared from people's sight.

Unexpectedly, Bitmain began to rise in the cold winter: Antminer S2 began to be sold in April 2014; Antminer S3 was launched in July; Antminer S5 went into mass production in December; and Antminer S7 was successfully developed in November of the following year... Under Bitmain's fierce offensive, when the market gradually recovered in the second half of 2015, people were surprised to find that only Bitmain's Antminer was left in the mining machine market. In 2015, Bitmain's annual net profit reached 334 million yuan, making it a well-deserved industry giant.

Why has Bitmain achieved such great success? Many people believe that the design difficulty of mining machine chips is low, which is true at the beginning, but as the mining machine market gradually matures, chip design has also begun to show a trend of specialization and industrialization. The special thing about mining machine chips is that they are mainly used for parallel computing, so computing power is not that important. The key lies in the power consumption of the chip - for two mining machines with the same computing power, the one with lower power consumption will have lower electricity costs and thus have more advantages in the market.

To this end, Bitmain developers have discovered a standard cell suitable for dynamic logic with large area and many tubes after a series of explorations. The special feature of this standard cell is that its trigger is dynamic. Compared with static logic triggers, the advantage of dynamic logic triggers is that their area and power consumption are half of the former.

Similarly, during this period, Yang Zuoxing, a doctoral graduate from Tsinghua University, also began to join Bitmain on a part-time basis and brought with him a full-custom methodology. As we all know, current chip design relies on EDA (electronic design automation) tools, while the full-custom methodology breaks this model. The full-custom methodology uses customized and manual design in the three design links of wafer fab cell library, logic synthesis and layout of RTL design, and strives to carve out every detail design. Although this is inefficient, all aspects of the chip are customized, which greatly improves the power consumption × area optimization ratio of the chip.

Under the dynamic logic solution and full customization methodology, Bitmain completed the research and development of the Antminer S7 series (28 nm) and S9 series (16 nm) in 2015. These two models are far superior to other manufacturers in performance, and their product costs are only half of those of their competitors.

The chip design industry is characterized by its unique market structure. For general industries, low-end products occupy most of the market share, but high-end products have a lower market share. The special thing about the chip market is that there is often a "winner takes all" situation.

With the successful sales of Ant S7 and S9, Bitmain has become a giant in the mining chip market, occupying more than 80% of the market share. Under Bitmain's strong technical advantages, foreign mining chip manufacturers have been eliminated: KNC, bitfury, Spondoolies-Tech, and 21 Inc., which were high-profile in 2014 and 2015, quickly declared bankruptcy or withdrew from the mining chip market. With the turnover of Bitmain's employees, dynamic triggers and fully customized methodologies have gradually spread to other domestic chip manufacturers, but the relevant technologies are still retained in China. In 2017, Japan's GMO company wanted to enter the mining chip market, but its outdated technical design could not compete with domestic manufacturers at all, and soon withdrew from the market due to losses.

Since then, the mining machine chip market has become the world of Chinese manufacturers. Data shows that in 2017, China's three major mining machine manufacturers Bitmain, Canaan Creative and Ebang International monopolized nearly 90% of the global market share, and the rest was also divided up by other Chinese manufacturers such as Wuhan Xindong and BitMicro.

Capital Era

In 2017, Bitcoin ushered in a bull market, with an annual increase of 1,700%. The huge investment returns attracted the attention of the world. For a time, Bitcoin and blockchain became the hottest topics in the market. Many blockchain companies were born like mushrooms after rain. The A-share market saw the rise of "blockchain concept stocks", and the venture capital circle also began to rise to "All in Blockchain". As the most profitable mining machine manufacturer in the industry, it also became a hot company.

In December 2017, when the market was at its craziest, Bitmain’s sales staff were treated like stars in the crowd. Customers gave them gifts and treated them to meals just to get orders for Antminers, and at that time, Bitmain’s orders were already scheduled until the spring of the following year. According to Frost & Sullivan, the market size of ASIC mining machines was only $700 million in 2016, but by 2017, its market size had reached $3 billion; in the mining machine delivery volume that year, Bitmain sold 940,000 mining machines, Canaan Creative sold 290,000, and Ebang International sold 50,000, accounting for 88% of the global market share. Under this circumstance, mining machine companies began to plan to go public.

On September 26, 2018, Bitmain officially submitted its prospectus to the Hong Kong Stock Exchange. In addition, Canaan Creative and Ebang International submitted their listing applications to Hong Kong in early 2018. For a time, mining chip companies dominated the headlines of the capital market, and the outside world finally got a glimpse of the business revenue of these mysterious mining machine manufacturers. Of course, the results were far beyond people's expectations. Their operating performance was actually better than most A-share listed companies in the same industry.

In terms of operating income, Bitmain's revenue reached 16.452 billion yuan in 2017, while the highest operating income of A-share semiconductor listed companies in the same period was Trina Solar with 26.158 billion yuan. Therefore, from the revenue ranking, Bitmain will rank fourth among all A-share semiconductor listed companies, lower than the second-ranked Changdian Technology with 23.85 billion yuan, the third-ranked Nasda with 21.3 billion yuan, but higher than Longi Green Energy Technology with 16.36 billion yuan.

In terms of net profit, Canaan Creative and Bitmain have maintained a net profit growth rate of more than 1 times for three consecutive years. In 2017, Bitmain's net profit was as high as 4.583 billion yuan, Canaan Creative was 361 million yuan, and Ebang International was 385 million yuan. If ranked together with listed companies by net profit, then Bitmain's net profit in 2017 can surpass the entire semiconductor industry - the highest net profit among A-share semiconductor listed companies in 2017 was Longi Green Energy Technology Co., Ltd., but its net profit was only 3.549 billion yuan, nearly 1 billion yuan less than Bitmain's net profit.

However, due to the fact that the blockchain business was too advanced, the Hong Kong Stock Exchange had doubts about the sustainable operating capabilities of the mining machine manufacturers and ultimately rejected the listing applications of the three major mining machine manufacturers. The first attempt of blockchain companies to go public thus ended in failure. Subsequently, the three major mining machine giants turned their attention to the US securities market: On November 21, 2019, Canaan Creative successfully landed on the Nasdaq in the United States, becoming the world's first blockchain listed company; in June 2020, Ebang International successfully listed in the United States; according to relevant reports, Bitmain is also preparing to go public in the United States.

Hangzhou City held a light show to celebrate Canaan Creative's listing as the "world's first blockchain stock"

The ups and downs of the development of mining machine manufacturers have not ended with their listing in the United States. In the near future, they will continue to write their own glory. Perhaps many people will disdain this and think that mining machine chips are worthless except for mining encrypted digital currencies. This is a popular misconception in the market. However, many people do not realize that the value and significance of mining machine chips to the semiconductor industry is no less than the current national integrated circuit fund with a scale of hundreds of billions, which is reflected in at least the following three aspects:

(1) Cultivation of talents in the semiconductor industry

The problem with my country's chip industry is that the industry itself is a talent-intensive industry, but its appeal to talent is far from enough. According to the survey of the "China Integrated Circuit Industry Talent White Paper (2018-2019)", there were 8.2 million college graduates in my country in 2018, and the total number of graduates in integrated circuit-related majors was about 199,000, but only 38,000 graduates in integrated circuit-related majors entered the industry, that is, only about 19% of graduates entered the semiconductor industry.

The main reason is that the salary level in the chip industry is not high. According to statistics, in 2019, the average salary of executives in the financial industry was 22.8 million yuan, the real estate industry was 14.78 million yuan, and the semiconductor industry was only 9.4541 million yuan. Salaries without market competitiveness cannot attract talents, resulting in a large talent gap. According to the White Paper, there will still be a gap of 261,000 people in China by 2021.

In comparison, the average salary in the blockchain industry where mining machine manufacturers are located is very high. According to relevant data from Liepin.com, the average annual salary for technical positions in the blockchain industry in 2019 was 384,000 yuan, higher than the 201,000 yuan in the Internet industry. Before 2018, the salary in the semiconductor industry was much lower than that in the Internet industry, resulting in most graduates of integrated circuit majors flowing into the latter. The rise of the mining machine chip field has attracted a large number of outstanding talents to join the mining machine chip design field, including Yang Zuoxing, the current founder of BitMicro, who we introduced earlier.

There is no doubt that mining machine chip manufacturers, especially in 2018, before the "Everyone Learns Chips" campaign, which strived to localize chips, retained a large number of chip design talents for my country's semiconductor industry with their generous salary packages. What is even more valuable is that compared with university researchers who have never been involved in the "tapeout" link, mining machine chip manufacturers provide an excellent platform for my country's chip practitioners. Whether it is the world's leading 7nm process or the most advanced 5nm process, they can participate in the design and testing at the first time, and accumulate a lot of knowledge and experience in "learning by doing".

(2) Technological accumulation in cutting-edge fields

A big misunderstanding many people have about mining chips is that the technology they have accumulated is useless in other fields. This view may have been correct in the past, but with the rise of artificial intelligence and the increasing failure of Moore's Law, these technologies have shown great value and use.

First, in the field of artificial intelligence, due to the increasing demand for edge computing, AI has begun to move from the "cloud" to the "terminal". Among various platforms, the CPU is good at control, and its design determines that in order to reduce the waste of computing power caused by data transmission during operation, it needs to accumulate a certain amount of data before calculation, which cannot meet the strong real-time requirements of AI; GPU only uses part of the computing resources while bearing high power consumption, and is not an option with a good energy efficiency ratio whether in the cloud or the terminal. Therefore, ASIC chips are becoming more and more popular in high-performance, low-power terminals. In addition to being unable to expand, ASIC chips have advantages in power consumption, reliability, and volume.

Therefore, market analysts generally believe that with the development of edge computing, ASIC will become the core of future AI chips. Ovum's survey report pointed out that in 2016, the cloud (including enterprises, data centers, etc.) was the main revenue area for deep learning chips, accounting for 80%, but by 2025, this ratio will change, the edge will account for 80%, and the cloud will drop to 20%. The market share of ASIC is expected to rise significantly as the demand for edge computing increases. It is expected that by 2025, the market share of ASIC will increase significantly from the current 11% to 48%.

The mainstream ASICs on the market currently include TPU chips, NPU chips, VPU chips, and BPU chips. Due to the long development cycle of ASICs, only large manufacturers have the funds and strength to carry out research and development, and Bitmain is one of the best, having accumulated a lot of technology and experience in the research and development of mining machine chips. In addition, other manufacturers such as Canaan Creative and Ebang International have also made arrangements in the field of AI.

Another application of mining chip accumulation technology is in the field of "Moore's Law". In the 1970s, Gordon Moore, the founder of Intel, proposed that "when the price remains unchanged, the number of transistors that can be accommodated on an integrated circuit will double approximately every 18-24 months." At the same time, computer performance will usually double. However, since 2011, Moore's Law has been difficult to continue to advance. For this reason, many semiconductor companies and professional institutions, including Intel, have placed their hopes on "architectural innovation" to achieve "beyond Moore."

The "full customization methodology" in the field of mining machine chips is a new "architectural innovation" direction: Ant S7, Ant S9, Shenma M3, Shenma M10, designed by Yang Zuoxing using the full customization methodology, each generation of mining machines launched has improved the power consumption × cost optimization ratio by more than 4 times compared with the mining machine chips designed using traditional methods with the same process technology. This is far beyond the improvement brought by relying solely on advanced process technology. Therefore, as Yang Zuoxing frankly said, the full customization methodology has brought innovations in chip design, which is expected to continue to promote Moore's Law for 5-10 years.

(3) Promoting the domestic semiconductor industry

In April 2020, A-share listed company Changdian Technology announced that due to substandard packaging quality, Wuhan Xindong Technology Co., Ltd. filed a lawsuit with the Wuxi Intermediate People's Court, claiming 175 million yuan from Changdian Technology. It is worth noting that according to the financial report, Changdian Technology's net profit for the whole year of 2019 was only 89 million yuan.

Many people were surprised to see this news. The claim amount of this little-known Wuhan Xindong company could eat up the net profit of Changdian Technology for nearly a year. In fact, Wuhan Xindong is also a mining machine chip manufacturer in China, but compared with market giants such as Bitmain and Canaan Creative, it is smaller and more low-key.

Of course, through this incident, many people have begun to realize that the status of mining machine chips in our semiconductor industry chain cannot be underestimated. Downstream wafer manufacturers and testing and packaging factories have played an important role, especially in the field of advanced process technology, where their role is even more important.

Since 2017, Bitmain and Huawei HiSilicon have become TSMC's two largest customers in mainland China. In October 2017, TSMC announced its third-quarter business revenue. Shockingly, the demand for mining chips increased TSMC's third-quarter revenue by $350 million to $400 million. Mining chips even squeezed the production capacity of Huawei HiSilicon's Kirin chips that year. According to Jiwei.com, in December 2017, Bitmain's orders for TSMC's 10nm process exceeded Huawei HiSilicon's Kirin chips.

This phenomenon can also be found in financial data. According to the prospectus and financial reports of the three major mining machine companies, from 2015 to 2019, the sales costs of Bitmain, Canaan Creative and Ebang International reached 8.324 billion yuan, 1.939 billion yuan and 949 million yuan respectively. Among the various costs of mining machines, the cost of wafer manufacturing is about 75%, and the cost of packaging and testing is about 15%. From this, we can calculate the expenses spent by the three major mining machine manufacturers on wafer manufacturing and testing and packaging in each year. As shown in the figure below, in the field of wafer manufacturing, the expenses of the three major mining machine manufacturers increased from 392 million yuan in 2015 to 8.409 billion yuan in 2019; the testing and packaging costs increased from 78.48 million yuan in 2015 to 1.681 billion yuan in 2019, with an annual compound growth rate of more than 1 times.

Take Changdian Technology, which provides testing and sealing services for the three major mining machine manufacturers, as an example. Changdian Technology is the largest testing and sealing manufacturer in China and the third largest in the world. According to the financial report, Changdian Technology's operating income in 2019 was 23.526 billion yuan, and its net profit just turned from loss to profit, reaching 97 million yuan. Even so, most of Changdian Technology's profits actually come from government subsidies. In 2019, Changdian Technology disclosed that it received government subsidies of 103.5877 million yuan. Similarly, from 2016 to 2018, Changdian Technology's net profits were 106 million yuan, 343 million yuan, and -939 million yuan, respectively, but the corresponding government subsidies included in the current profit and loss were as high as 207 million yuan, 351 million yuan, and 155 million yuan. This also means that Changdian Technology, the world's third largest testing and sealing manufacturer, relies heavily on government subsidies to survive.

Obviously, the mining chip market, which has reached billions and is still growing rapidly, is very important for domestic testing and packaging manufacturers.

Similarly, in the field of wafer manufacturing, since mining machine chips require large-scale, high-performance and low-power computing, there is a great demand for advanced manufacturing processes. Since mining machine chips are more concerned about performance and need to seize the market, they are not very concerned about the price of wafers, so wafer foundries such as TSMC are happy to increase orders for mining machine manufacturers and even allow mining machine manufacturers to participate in the testing of the latest 7nm and 5nm process technologies.

Take Canaan Creative as an example. According to the latest news, SMIC, a domestic wafer manufacturing giant, has begun to negotiate with Canaan Creative to cooperate on 14nm mining machines. So how big is Canaan Creative's demand for 14nm process chips? Based on the annual sales records of various types of mining machines disclosed by Canaan Creative and the number of chips inside the mining machines, OKLink Research Institute has calculated its annual demand for advanced process chips. As shown in the table below, in 2019, Canaan Creative's demand for 16 nm chips is approximately 76.09 million pieces, and it is expected that the demand for 14nm chips will remain at the same level.

According to official news from SMIC, as of the end of 2019, SMIC's 14nm production capacity was only 3,000 to 5,000 wafers/month, and it is expected to reach only 15,000 wafers/month by the end of 2020. Considering the wafer yield and the number of wafers on each wafer, we estimate that if Canaan Creative gives all 14nm orders to SMIC, even if the capacity expansion is completed by the end of 2020, SMIC will need 1-2 months of capacity to complete all orders.

Therefore, for domestic wafer foundry giants, orders from mining machine chips are definitely a considerable income, which will also drive wafer manufacturers in mainland China to develop more advanced process technologies.

postscript

In his early years, Bitmain founder Wu Jihan studied at Nankai Middle School in Chongqing. The Nankai series of schools was founded by patriotic educator Zhang Boling in 1904. Nankai Middle School, which has been in existence for nearly a hundred years, has cultivated a large number of well-known scholars and political elites for China. However, in the early years, Nankai Middle School was a private school, and the school funding mainly came from individual donations from the society. However, private donors who were able to donate were mostly politicians such as Cao Rulin and warlord Yang Yide, who donated a lot to Nankai. Students protested, and Zhang Boling said: "Beautiful flowers may be watered with manure."

Today's mining chip manufacturers are in a similar situation to Zhang Boling. Since blockchain technology is still in its early stages, there are many market chaos and people are quite critical of mining chip manufacturers. However, blockchain technology, the underlying technology of encrypted digital currency, may not be the "sewage" that some people say, but the "nectar" of the future. On October 24, 2019, the Political Bureau of the CPC Central Committee conducted its 18th collective study on the current status and trends of blockchain technology development. The national leaders delivered an important speech on this, affirming that the integrated application of blockchain technology plays an important role in new technological innovation and industrial transformation, and taking blockchain technology as an important breakthrough for independent innovation of national core technologies. I believe that with the further development of my country's blockchain industry in the future, people will be more aware of the value of mining chip manufacturers.

Source: OKLink Blockchain