SEMI: At least 38 12-inch wafer fabs will be added globally in the next four years

Today, SEMI released the latest fab outlook. They pointed out that at least 38 new 300mm Fab fabs will be added from 2020 to 2024. This is a conservative forecast and does not consider low-probability or rumored fab projects. At that time, the monthly production capacity of the Fab factory will increase by about 1.8 million wafers, reaching more than 7 million wafers.

As shown in the figure above, Taiwan, China will continue to be the world’s largest contributor to foundry production capacity. Considering TSMC’s influence in the foundry field, we can understand. South Korea is closely followed, because Samsung’s efforts in wafer fabs in recent years, this data is also a matter of course. Mainland China will rise sharply in the next few years.

According to SEMI statistics, during this period, Mainland China will rapidly increase its global share of 300mm production capacity, which will increase from 8% in 2015 to 20% in 2024, with production capacity reaching 1.5 million pieces/wpm. SEMI pointed out that although non-Chinese companies will account for a large portion of this growth, China is accelerating its capacity investment.

“By 2020, these companies will account for about 43% of China’s fab production capacity. It is expected to reach 50% by 2022 and 60% by 2024,” SEMI emphasized.

SEMI said that while China is advancing with leaps and bounds, Japan’s share of 300mm production capacity continues to decline, and the global share will drop from 19% in 2015 to 12% in 2024. The share of the Americas is also declining, from 13% in 2015 to 10% in 2024.

From the perspective of product segmentation, SEMI said that memory will account for most of the growth in expenditures for 300mm Fab fabs. The actual and projected investment amounts show that expenditures in this area will increase steadily in high single digits each year from 2020 to 2023, and will increase by 10% by 2024.

SEMI further emphasized that from 2020 to 2024, the contribution of DRAM and 3D NAND to the expenditure of 300mm Fab plants will be unbalanced. However, from 2021 to 2023, Logic/MPU investment will steadily increase. Power-related equipment will become the leader in the investment of 300mm Fab plants, with an increase of more than 200% in 2021, and double-digit growth in 2022 and 2023.

Extended reading: Misunderstandings on the domestic wafer production line

There is no doubt that due to demand and current development status, a new wafer production line must be launched in China. However, Xinmou Research wrote an article earlier that there are various misunderstandings in the domestic wafer production line, which is worthy of our reference and learning.

The wafer production line has always been the most demanding link in the integrated Circuit industry for talents, technology, capital, etc., and it has naturally become the most vigorous direction for local governments when attracting investment. The driving force of the manufacturing industry is often mentioned, and heavy investment is valued every day, but few governments pay attention to assessing high risks. In the past, Chengdu GF and Nanjing Dekema gave up halfway, and the existing Wuhan Hongxin stopped halfway. Countless cases in the past have told us that blindly launching wafer lines will bring many challenges and hidden dangers to the industry and the government.

　First, the decentralization of resources leads to misjudgments in top-level planning, and even out of focus of the market itself;

The manufacturing industry needs top-level planning and orderly development. Many of the multiple integrated circuit production lines that have been on the news are still in the plan and publicity. However, the resource allocation of the entire industry and the national top-level design will be interfered by the industry’s “prosperity and overcapacity”, which is prone to misjudgment. .

　　Second, the change of local government leadership brings more uncertainty to new projects;

Manufacturing line is an industry that requires long-term investment and continuous support. From planning and construction to profitability, three to five years are common; from preparation to listing, nine years of success are common. Don’t you see, the technological capital strength is as strong as TSMC, SMIC and Hua Hong Group, and it has been a long time to achieve profitability after many years. The ten-year-old operation is not a lie; the pressure of depreciation and amortization is not intimidation. The current production line is mainly funded by the local government, and the government-led projects are extremely risky due to the change of officials and transfers. The first phase is no problem, but what about the second and third phases? After the land is leveled, the piles are built, the workshop is complete, and the equipment is moved, what happens if there is a change in government management, a change in the attitude of the leader, and the follow-up funds are unsustainable? If the successor is unwilling to continue investing, will social funds take over, in what way and at what valuation? How to deal with the window period between the takeovers, daily utility bills, ultra-clean room maintenance costs, and personnel salaries? Please talk about this series of questions before you start, and don’t think about it after you start!

　Third, the challenges brought about by the lack of talent depth have not attracted enough attention;

The manufacturing industry is a highly “people-oriented” industry. The current domestic situation is “sufficient funds and insufficient talents.” Many local governments, especially those in third- and fourth-tier cities, tend to underestimate this problem. The cultivation of talents requires scientific research institutes, university education, environmental atmosphere, and agglomeration effect. The complexity is much higher than the introduction of an enterprise. The project can be signed in one day, 10 billion yuan will be invested in ten days, and one hundred acres of land can be leveled in one hundred days, but the construction of talents often takes ten years. In addition, the talent is not a single soldier, but the entire organization and several teams. Given the weak talent base in many small cities, where does this team building come from?

　　Fourth, the sources of patents and intellectual property rights have become hidden dangers;

The manufacturing industry is a place where intellectual property rights are concentrated. Lawsuits due to intellectual property disputes have begun to increase. This is an overall general trend judgment made by Inimou Research as early as 2016. Many recently announced projects do not make any statement about the source of intellectual property rights, and it is impossible to judge the future technological development route. We are also deeply worried about this. The superstructure also pays attention to this type of risk. Five key points are “whether the project’s talent pool is stable, the marketization of fund raising, whether the project product positioning has market advantages, the risk of technology sources under the US technology blockade, and whether there are core intellectual property rights” and other five key points. It is difficult to answer many of the newly launched wafer lines.

In addition to challenges and pressures, we also found that there are several misunderstandings, misunderstandings and misjudgments in the understanding of domestic government departments regarding the development of the manufacturing industry:

　Misunderstanding 1: Despite the above, I am very successful in other fields, and the successful experience can be copied;

In the past, many local governments did become famous in specific fields, such as panels, lighting, and solar energy. In keeping with the industrial development trend, the government’s large investment and large layout have brought large benefits. It is hoped that it can replicate the successful experience in the past, make large investment, large-scale recruitment, and make another success in the direction of integrated Circuits. But past achievements do not mean that it takes all fields and covers all industries. The integrated circuit is quite different from other industries. It requires careful preparation and unremitting persistence. When we are going fast, we need to think more and ask ourselves: Is the project team reliable? Is the talent pool ready? Has the technology source been determined? Has the humanities environment been built? We call on local governments to screen, think, prepare, and be cautious. Success in other fields may be copied, but it may not be pasted in the direction of integrated circuits! All kinds of dissatisfaction with the monopoly of integrated circuits have been verified by many cross-field experts.

　Misunderstanding 2: Despite the above, more construction is always beneficial to the construction and development of the domestic industrial chain;

The manufacturing industry is indeed the cornerstone of the construction of the entire integrated circuit industry ecosystem, but the rapid prototyping of the industrial ecology requires “quality, order, and degree.” The wafer production lines that have been launched in recent years have been ramping up quickly and homogenization is serious. The integrated circuit industry cycle is an objective law of “everything is going to happen”, and the current development method of “swarming up” will only lead to the collapse of the price system. At that time, no matter whether Party A or Party B, under the cover of the nest, will there be any eggs? In the future, product prices will drop and capacity utilization will be reduced. Naturally, economic benefits will be bad. The “golden nest” that was held in the palm of the hand will become a “heart nest”.

　Misunderstanding 3: Although it won’t work, SMIC and Huahong will come to take over in the future;

There is a misunderstanding in the industry, and even many friends in the front line of the industry: production lines are installed in various places, and they know that it may not be possible, but it does not matter if it fails. Wait a few years for domestic companies to take over. If China’s manufacturing industry is market-oriented or privately dominated, this might well be a possible way out. It is a pity that under China’s special national conditions and the constraints of “state-owned assets cannot be lost”, the expectations are very good, the reality is very skinny, and the reality is difficult to operate.

Let alone whether the built factories and expired equipment will still be useful at that time, just a “state-owned asset cannot be lost” makes it difficult for mergers and acquisitions to occur. Due to the restrictions of state-owned assets management methods, state-owned assets can only be preserved and increased in value, and cannot be sold at a discount. This is a “tightening curse” that has Chinese characteristics and cannot be circumvented in the short term. It is also for this reason that this round of industrial integration has lasted for three or four years, but it was not until 2017 that China Resources Microelectronics acquired the integration case of AVIC Microelectronics, and this case was also a special method of “state-owned assets allocation”. To succeed. Coupled with the merger of Hua Hong and Hongli many years ago, China’s integrated circuit foundries have only two successful integration cases, and they are all internal cycles between state-owned enterprises and state-owned enterprises.

In the past few years, there have been three sales cases in the international foundry industry. One is the Lfoundry 8-inch factory in Italy, with a monthly production capacity of 40,000 pieces. A 70% stake is sold to Zhongke Junxin at a discount of US$113 million, which is equivalent to a 100% stake for US$162 million. The second is Global foundries’ 8-inch factory in Singapore. , With a monthly production capacity of 35,000 pieces, sold to the world’s advanced Vanguard at a price of 236 million U.S. dollars; the third is Globalfoundries’s 12-inch plant in the United States, sold to Onsemi at a price of 430 million U.S. dollars. Some domestic governments have invested tens of billions to make 12-inches. They imagine that if they are not doing well, they hope to sell Fab and look at the price difference above. You can pour a pot of ice water! Not to mention, although the three cases are different, they are all mature and old factories, with technology, customers, and teams. It can be said that there are three; some new domestic factories have no funds, no talents, and no technology. , Can be described as three nos.

In the past, “Haian Lushan, Kunshan Dexin, Dongying Lianxin, Nanchang Jingxin, Zhengzhou Jingcheng, Nanjing Dekema” and other companies, when they encountered problems in their operations, without exception, they thought about solving their difficulties through mergers and acquisitions. There are also people around who want to take over, but “valuation” is always a topic that makes all parties exhausted. In the end, these equipment would rather rust than sell at a discount. In fact, it is difficult to revitalize the equipment even if it is the plant, and no one buys it. Who can integrate it? It can be said that “the halberd has not been sold, and the iron has not been sold.”

　　Misunderstanding 4: Just go on, and then find a relationship operation to become a “national team”;

Another dangerous idea is that integrated circuits are considered to be a race to enclose the land, and the industrial layout is to seize the pit. Choosing a direction, as long as it is necessary for the national economy and the people’s livelihood, put aside the feasibility first. The key is to hurry up, and whoever does it first will drink the soup. As long as I get on, do it, and stand still, no matter how bad I do it, the country will give me the bottom line. In the future, when the trio is drifting away and riding a tiger is difficult, the country will help me get off the tiger.

First, there is no need for central support to bypass supervision, and then the launch of the project will create established facts, and finally “run the ministry to make money” and become a national project. This is a small calculation led by some local governments. Integrated circuits are market-oriented and internationalized, and their opponents, customers, and industrial chains all come from all over the world. Ironing depends on one’s own strength. In the end, whether the local industry can get up or not depends not on how much subsidies the state provides, but on one’s own competitiveness. If the bones are weak, even if the “crying child has milk to eat”, how many subsidies the state has given, will ultimately be unable to support Ah Dou. For this kind of situation, we call on the country to be “willing to do it all.” If you say NO, you will say NO. If you want to eliminate one batch, we will eliminate one batch. Only by getting rid of outdated production capacity can we promote the development of the industry.

Birds take care of each other and then fly, and people think twice before acting. We still call on the government to “analyze” and “prudently” when making decisions. The right way to build production lines is to make decisions and then act. If you make a good decision, making the core becomes a new one, and if you make a bad decision, making the core becomes a bad idea.

Xinmou Research sincerely urges local governments to keep the spirit and clear life, not to be fooled by “experts”, to focus on a ten-year long-term plan, and to make decisions based on impulse, and not to make short-term gains based on immediate benefits. Faithful words do not like to hear, but when everyone is enthusiastic, there must be someone who can sound the alarm. For the sake of benignness, orderliness, and long-term, Ximou Research is willing to shout and shout, and is willing to be a child in front of the emperor’s new outfit, but is afraid that he is afraid of deafening shouts and calling people who can’t wake up pretending to be asleep.

We don’t want the current multi-line construction to end up with “it’s as if the birds are thrown into the forest, and the white land is really clean”. We don’t even want to “bad money drive out good money”, and the real good companies will be disrupted by the industry. The game is forced into the dust, “spattered into mud and crushed into dust”, making “core-making” “worry”. I sincerely wish that Chinese industries and local industries can go more stable, benign, healthier, and longer-term.