Are Chinese chip manufacturers being strangled by Japanese suppliers for commercial or political purposes?

Recently, Japanese wafer giant Sumco decided to cut wafer orders from Yangtze Memory Technologies (formerly Wuhan Xinxin), a Chinese mainland memory chip manufacturer, and give priority to supplying to TSMC, Intel, Micron and other large manufacturers. With Japanese Sumco giving priority to supplying manufacturers in the United States, Japan, and Taiwan, Yangtze Memory Technologies may face a shortage of wafers. This is not good news for Tsinghua Unigroup, which is trying to achieve domestic substitution in memory chips.

Japanese Crown Prince Naruhito visits a semiconductor company

Mainland China's 12-inch wafers are heavily dependent on imports

In recent years, 12-inch wafer factories have sprung up everywhere in mainland China, and domestic and foreign companies have invested huge amounts of money in building factories in mainland China. After receiving investment from the Big Fund, SMIC expanded its factory and built three new 12-inch production lines in Beijing and Shanghai. Huali Microelectronics followed suit and started building a 12-inch production line, and plans to master the 14nm manufacturing process around 2020. Tsinghua Unigroup has invested more than 100 billion yuan in building storage factories in Nanjing and Wuhan respectively. Hefei has also built a storage chip factory in cooperation with Yukio Sakamoto, the former president of Elpida in Japan.

Among overseas companies, TSMC invested 19.5 billion yuan to build a 12-inch wafer factory in Nanjing, and plans to start mass production of 16nm process in the second half of 2018. Samsung, Intel and SK Hynix have built 12-inch wafer production lines in Xi'an, Dalian and Wuxi respectively, mainly for the production of memory products including 3D NAND and DRAM. GlobalFoundries established a joint venture in Chengdu to build a factory. UMC established Unichip Integrated Circuit Manufacturing (Xiamen) Co., Ltd. in Xiamen and plans to introduce 28nm manufacturing process.

In 2016, China's demand for 12-inch wafers was more than 400,000 pieces per month. In 2017, the demand for 12-inch wafers will exceed 600,000 pieces per month. As the above-mentioned wafer factories are put into production one after another, the demand for wafers by Chinese companies will surge, and it is expected to exceed 1 million pieces per month by 2020. However, the global wafer supply is basically monopolized by foreign companies. Even as the demand for wafers by Chinese companies increases year by year, foreign companies have strangled the wafer supply chain. Let me explain here that manufacturers such as TSMC and Intel do not produce wafers themselves, but only process wafers into various chips. Wafers as raw materials are provided by manufacturers such as Shin-Etsu and Sumco.

At present, the world's top few silicon wafer suppliers are Japan's Shin-Etsu, Japan's Sumco, Germany's Siltronic, the United States' SunEdison, and South Korea's LG Siltron, with market shares of 27%, 26%, 14%, 11%, and 10%, respectively. The combined market share of the top five giants is as high as 88%.

Mainland China manufacturers mainly produce 6-inch wafers, and the self-sufficiency rate of 8-inch wafers is less than 10%, and the self-sufficiency rate of 12-inch wafers is even lower. Therefore, Sumco's cancellation of wafer orders from Yangtze Memory Technologies, a mainland China storage chip manufacturer, is the result of being controlled by others in the supply chain.

The main reason for canceling Yangtze Memory's orders is commercial

Currently, Yangtze Memory Technologies Co., Ltd., a subsidiary of Tsinghua Unigroup, has developed a domestically produced 32-layer stacked 3D NAND Flash, which is expected to be put into mass production after 2018. In contrast, foreign manufacturers such as Samsung have already achieved 64-layer stacked 3D NAND Flash. Moreover, the market share of Samsung and SK Hynix's memory chips is very large, and they can spread the cost by relying on their market share. As a latecomer, Yangtze Memory Technologies Co., Ltd. not only has to catch up with the technological gap, but also has to face issues in terms of yield and cost.

Therefore, for now, Yangtze Memory does not pose a threat to foreign manufacturers such as Samsung, SK Hynix, Toshiba, and Micron. The reason why Yangtze Memory was cut off by Sumco is mainly due to commercial reasons.

In the past few years, the global wafer supply has generally exceeded demand. For example, in 2015, global wafer suppliers produced a total of 76 million 12-inch silicon wafers, but the market consumed only 57 million. With fierce market competition, silicon wafer suppliers have become the current top five manufacturers after a series of mergers. In addition, the recent severe global silicon wafer shortage has caused Intel, Toshiba, TSMC, Micron and other major manufacturers to increase prices to buy wafers (there are rumors that TSMC has increased prices by 10-15% to grab orders). In contrast, Yangtze Memory in mainland China is too small and has a small order volume. For suppliers, it is natural to give priority to serving high-quality customers with larger orders. In this context, Yangtze Memory can only stand aside.

In addition, the sharp increase in memory chip prices since October last year is also one of the reasons why Japanese suppliers give priority to supplying manufacturers in the United States, Japan, and Taiwan. After the Samsung Note7 spontaneously combusted, the price of memory chips skyrocketed, and the output value of a single 3D NAND chip reached 5,000-6,000 US dollars. Samsung and SK Hynix benefited greatly from this. In the fourth quarter of 2016, Samsung Electronics' operating profit was 7.8 billion US dollars, a year-on-year increase of 50%, despite the continuous spontaneous combustion of Samsung Note7 and huge losses. SK Hynix's revenue in the first quarter of 2017 was 38.4 billion yuan, a year-on-year increase of 72%, and its net profit reached 11.6 billion yuan, a year-on-year increase of 324%. With the skyrocketing price of memory chips and several major manufacturers around the world scrambling for orders, the weak Yangtze Memory is even more unable to compete with international manufacturers for production capacity.

Achieving wafer self-sufficiency takes time

There are two conditions for achieving domestic substitution of silicon wafers. On the one hand, it requires massive amounts of capital, and on the other hand, it requires breaking through patent barriers.

In terms of funding, it takes about $400 million to build a 12-inch polished silicon wafer production line with a monthly output of 200,000 wafers, while the total investment for an 8-inch silicon wafer production line with a monthly output of 200,000 wafers is about $200 million. Therefore, unless the state invests, private capital has little willingness to enter this industry.

In terms of technology, foreign wafer suppliers have applied for a large number of patents to protect related process technologies, and as latecomers, it is difficult for Chinese companies to bypass these patent barriers. Due to economic interests and political factors, the possibility of Chinese companies obtaining technology licenses from foreign wafer suppliers is very small, which increases the difficulty for China to catch up.

In addition, how emerging wafer suppliers in mainland China can survive in the fierce commercial competition is also a problem. After decades of commercial competition, a monopoly pattern has been formed by several major companies. Since the five major wafer suppliers occupy a dominant position, it is difficult for mainland Chinese companies to commercialize even if they achieve technological breakthroughs. In fact, some companies or scientific research institutions have developed 12-inch wafers, but due to the low yield and the inability to spread the cost through production capacity, they are at a disadvantage in commercial competition. Shanghai Xinsheng Semiconductor, founded by Zhang Rujing, the former founder of SMIC, is still far behind Japan's Shin-Etsu and Japan's Sumco.

The existence of a gap does not mean that we will always be controlled by others. Compared with the manufacturing of some core semiconductor equipment, the technical difficulty of manufacturing raw materials such as wafers is relatively low. More problems lie in circumventing patent barriers, achieving commercialization, and surviving in the fierce market competition.

In the past, TSMC, Intel, GF, UMC and other manufacturers were the major customers of wafer suppliers, and the market demand for wafers from SMIC, Huali Microelectronics, Wuhan Xinxin and other manufacturers in mainland China was relatively small, which led to foreign wafer suppliers being able to get the moon first because of their proximity to the water. However, due to the small market demand in the mainland, few people in the mainland would do this. However, with the construction of a large number of factories in mainland China, the market demand for wafers has risen rapidly, and wafer suppliers in mainland China will have a good opportunity for development.

There is still a long way to go to achieve a counterattack in memory chips

Not long ago, Huawei P10 caused a flash memory scandal. This situation was caused by Huawei itself, but also by the fact that Huawei was controlled by others in terms of memory chips. After all, the NAND Flash market is dominated by the ToggleDDR camp led by Samsung and Toshiba and the ONFI camp led by Intel and Micron. Foreign giants such as Samsung, Toshiba, SanDisk, Micron, and SK Hynix occupy more than 80% of the market share. Without domestic suppliers to choose from or use as bargaining chips with international manufacturers, it is difficult for Huawei to fully control the memory chip supply chain.

Tsinghua Unigroup's two 100 billion yuan investments in Nanjing and Wuhan have given Chinese people hope of breaking the monopoly of international giants in memory chips. According to public reports, Gao Qiquan once said that Yangtze Memory will start mass production of 64-layer stacked 3D NAND flash memory in 2019, striving to shorten the gap with Samsung and other large manufacturers within 2 years.

（3D NAND）

However, even if Yangtze Memory can really start mass production of 64-layer stacked 3D NAND flash memory in 2019, it still has to face the possibility of shortage of foreign wafer suppliers. The industrial chain of the integrated circuit industry is very long, involving raw materials, equipment, design, foundry, packaging, testing and other aspects. If China wants to no longer be controlled by others in memory chips, it needs to face not Samsung, Toshiba, Micron, SK Hynix and a few companies, but to defeat the entire semiconductor industry division of labor system dominated by the United States. This task is very arduous and there is a long way to go.

As a winner of Toutiao's Qingyun Plan and Baijiahao's Bai+ Plan, the 2019 Baidu Digital Author of the Year, the Baijiahao's Most Popular Author in the Technology Field, the 2019 Sogou Technology and Culture Author, and the 2021 Baijiahao Quarterly Influential Creator, he has won many awards, including the 2013 Sohu Best Industry Media Person, the 2015 China New Media Entrepreneurship Competition Beijing Third Place, the 2015 Guangmang Experience Award, the 2015 China New Media Entrepreneurship Competition Finals Third Place, and the 2018 Baidu Dynamic Annual Powerful Celebrity.