New changes are brewing amid ups and downs in the global semiconductor industry

rain

New changes are brewing amid ups and downs in the global semiconductor industry [Copy link]

Compared with the same period in 2005, the beginning of 2006 seems to be different: semiconductor inventory is at a suitable level, the traditional post-holiday depression phenomenon has not reappeared, and semiconductor fixed asset investment has remained at a normal level of 20% of semiconductor annual sales. It is believed that there will be no overcapacity. The global semiconductor industry is expected to have a good start in 2006. Here we review some major events in the global semiconductor industry in 2005.

　　2005 was volatile and 2006 is still uncertain .

　　After another peak in the semiconductor industry in 2000, after three years of energy accumulation, it finally grew by 28% in 2004. However, the industry is surprised that by the fourth quarter of 2004, the growth had only lasted for four quarters. It is said that this is due to the semiconductor inventory. The industry generally believes that the industry has entered a new round of decline cycle again. Heavyweights including Morris Chang also predicted that the global semiconductor industry will show a negative growth of 2% in 2005.

　　Has it really entered a decline cycle? By the first quarter of 2005, at least the inventory problem has returned to normal levels. At this time, the industry expected that this cycle would be particularly mild and short, and was sure that it would recover again in the second half of 2005, so many market analysis agencies raised their forecasts

. When the second half of 2005 really began, the semiconductor industry placed its hopes on seasonal consumption growth and structural adjustments at the end of the year. However, soaring oil prices and typhoon attacks once again dampened the confidence of American consumers, casting a shadow over the global semiconductor industry. Market analysis companies such as iSuppli once again revised the growth rate for 2005. This repeated cycle caused the situation of the global semiconductor industry in 2005 to surprise market analysis companies.
　　For example, iSuppli has been constantly revising its forecast for the global semiconductor industry in 2005. In December 2004, they predicted that the global semiconductor industry would grow by 4.7% in 2005, but adjusted it to 6.1% in March 2005, and adjusted it to 5.9% on September 8, but adjusted it again to 2.4% on October 12, and finally revised it to 4.4% on December 1.

　　People always set goals and confidence in anticipation. The situation of the global semiconductor industry in 2006 has once again become the focus. iSuppli predicts that the global semiconductor industry will have less fluctuations in the next few years. They reiterated that the growth in 2005 was 4.4%, while the growth in 2006 was only 4.3%. The reason for the slight increase in 2005 and 2006 is that the sales peak brought by the previous PC replacement has become history in 2004. It should be noted that the current mild growth is the result of a long-term effect. In the past 27 years, the annual average growth rate of the semiconductor industry as high as 17% has become history. WSTS predicts that the growth of the global semiconductor industry will be 8% in 2006 and 10.6% in 2007.

　　To objectively analyze the trend of the global semiconductor industry in 2006, due to the lack of killer products, most people believe that the current growth of the semiconductor industry is driven by multiple factors, not just by the demand for PCs. The industry generally believes that consumer electronics, wireless products and LCD TVs will be the three major categories of terminal products that will drive the development of the semiconductor industry in 2006, but the problem is the lack of "superstars" with outstanding performance.

　　In summary, the performance of the global semiconductor industry in 2006 is expected to maintain a slight growth trend again. However, the current semiconductor industry has become very fragile and cannot withstand the impact of major global storms. It is expected that there will still be variables.

Pure foundry industry encounters IDM challenge.

　　It has been less than 20 years since Taiwan first established the foundry model in the world. However, due to the strong vitality of the foundry industry, its role and status in the development of the semiconductor industry have become obvious.

　　Although the global foundry industry has also experienced different stages of starting, growing and expanding. However, the situation of TSMC and UMC dominating the market has continued to this day, accounting for about 70% of the market share.

　　In recent years, the process technology progress of the global foundry industry has slowed down. The main reason is that when the process technology transitioned from 0.18μm to 130nm technology, it encountered technical difficulties due to the use of copper interconnection and low-k dielectric materials. In addition, the semiconductor industry needs time to digest the expansion of production capacity after the bubble in 2000. In general, the global foundry industry will face new changes. According to market research companies, the average annual growth rate of the global foundry industry should be twice that of the semiconductor industry, close to 18%-20%. However, the results in 2005 were really disappointing. According to statistics from the foundry industry in Taiwan, the global foundry industry may only grow by 1%-2% in 2005.

　　At present, the competition in the global foundry industry can be basically divided into two parts: the competition in high-end foundry, the key is where are the customers of 90nm and below technology? When high-end customers are looking for foundries, the key is not low prices, but the extension capabilities of technology in the development of next-generation products. In fact, they are looking for a partner for co-existence.

　　The other part of the foundry industry competition is mainly about price. It is important to note that the profits in the global foundry industry are becoming increasingly thinner, and this price decline trend will intensify, so the next step of industry mergers is inevitable.

　　Another sign in the global foundry industry is that Samsung, Toshiba, Infineon, IBM, and other Japanese IDMs have recently begun to join the ranks of high-end foundries. Although this change is only a capacity adjustment from the perspective of the IDM itself, it is estimated that it will not pose a threat to TSMC and UMC in the near future, but the global foundry industry is indeed facing further reform pressure, and some even doubt whether the fabless-plus foundry model can compete with IDM.

300mm becomes mainstream, and the introduction of 450mm is delayed

　　. Driven by Moore's Law, the trend of increasing the diameter of silicon wafers has never stopped. In order to achieve the transition from 8-inch to 12-inch silicon wafers, the world has invested a total of US$20 billion in equipment and materials, and some people in the industry predict that it will take 30 years to get a return on investment.

　　The global 12-inch silicon wafer production has once again sounded the horn of advancement since 2004. By the end of 2005, there were 46 12-inch silicon wafer production lines in the world, including 13 in the United States, 6 in Japan, 11 in Taiwan, China, and 4 in South Korea.

　　The progress of the global 12-inch silicon wafer has slowed down compared with expectations. According to statistics from the International Semiconductor Capacity Organization (SICAS), by the fourth quarter of 2004, the global weekly output of 12-inch silicon wafers was 53,400 pieces. If converted to 8 inches, it is 120,000 pieces per week, accounting for 10% of the corresponding global total silicon wafer output (converted to 8 inches) of 1.26 million pieces per week. This shows that even if it is predicted by the end of 2005, the global output of 12-inch silicon wafers will only account for 15% of the global total silicon wafer output.

　　The current status of the global 12-inch silicon wafer production has the following characteristics: All newly built chip factories have been 12 inches and below 65 nanometers since 2006, and it is no longer economical to build new 8-inch factories. Huang Chongren, chairman of Taiwan Powerchip Semiconductor, said in January 2005 that from the perspective of the memory industry, 2006 will be the year of crossover to 12-inch factories, and by then 8-inch factories will no longer be competitive in mass-producing standard DRAM.

　　According to a report by the Information Network Company of the United States on December 15, 2005, in 2000, only 20% of the world's total investment in 300mm semiconductor equipment was spent, but by 2004 and 2005, this figure had risen to 60% and 70% respectively. The life cycle of each size of silicon wafer can add up to about 20 years. Therefore, although 300mm chip equipment is becoming more and more mature, it is believed that 200mm equipment still has strong vitality and will not withdraw from the stage of history immediately.

　　According to ITRS's original forecast, the introduction period of 450mm silicon wafers should be between 2010 and 2014. VLSI reported on August 19, 2005 that the various R&D funds required to develop 450mm silicon wafers were about 102 billion US dollars. If a 450mm silicon wafer production line has a monthly production capacity of 150,000-200,000 pieces (equivalent to 8 inches), it would require an investment of 4 billion to 5 billion US dollars.

　According to historical experience, the effective life cycle of each silicon wafer size is about 20 years. For example, the 8-inch silicon wafer, which was prepared in 1986, did not enter the profit period until 1997. It can be predicted that the real profit cycle of 8-inch silicon wafers will be replaced by 12-inch silicon wafers in 6-7 years.

　　Therefore, the semiconductor industry currently believes that the mainstream of 450mm silicon wafers will be postponed to 2020-2025. Stave Newberry, president of Lam Research, called on everyone not to discuss this topic in the next three years.

　　Under the new situation, the global semiconductor equipment industry is facing a new period of mergers and repositioning. The increasing complexity of technology, the gradual reduction of total market demand, and the pressure of equipment price reduction are the dominant factors. However, since the global 300mm chip production line has just entered the mature stage, it can match 200mm in terms of cost performance. Therefore, it can be expected that the semiconductor equipment industry will still have a certain vitality in the next few years.

　　Whether the production of 450mm silicon wafers can continue is a serious topic that cannot be answered for the time being. Intel is the only active promoter in the world. Even if it is determined in the end, the possible development model is that chip manufacturers will develop their own equipment or chip manufacturers and equipment manufacturers will work together again to jointly promote the development of the semiconductor industry.

Process technology crosses the 45nm barrier

　　According to the 2004 ITRS report, the global semiconductor industry should enter the 90nm node and the 65nm barrier in 2006. At present, from the overall technology trend, it has not dragged down the industry, that is, immersion lithography, copper interconnection and low-k dielectric materials, although there are also a lot of challenges, but basically can keep pace with the industry.

　　The industry has been convinced that 193nm immersion lithography has achieved comprehensive success in 65nm and 45nm technologies.

　　According to various reports, immersion lithography has successfully passed the 45nm process technology, and for the next 32nm technology, it is still difficult to make a conclusion whether to continue to develop large numerical aperture immersion lithography technology or adopt EUV technology.

　　At the seminar held by SemiJapan and Cymer in December 2005, there were different views on how to achieve 32nm technology. At present, many basic researches on EUV technology are still lagging behind. For example, when evaluating EUV technology in 2004, there were still three major difficulties, namely mask defects, photoresist sensitivity and light source power. After efforts in 2005, the resolution of photoresist and the roughness of lines have become the most important contradictions. The industry predicts that EUV technology will not be introduced to the market until 2009. If immersion ArF lithography is used, secondary exposure must be used, which will increase costs and prolong the cycle. Therefore, from the perspective of accelerating the development of new products, the call for EUV technology is gradually increasing.

　　According to the latest news, Germany's Xtreme Technology Company claims that it has increased the output power of EUV light source from the initial 120W to the current 800W level, and it is expected to reach 1000W by 2010. Carl Zeiss of Germany has provided ASML with the first EUV optical system, so the EUV prototype (Alpha) that ASML is developing has been provided to IMEC in Belgium and Albany University in New York for trial use in the second quarter of 2005.

　　As for the high-k dielectric material problem that solves the leakage current of transistors, it is still difficult from the process point of view. It is estimated that it will not be adopted at 45nm (including 45nm). According to industry predictions, it may be adopted at 32nm.

　　The industry is always exploring what size will not be able to cross, 22nm or 16nm, that is, when will be the end point of silicon materials? At present, there is a saying in the industry that it is likely to be 16nm.

The transfer of the semiconductor industry chain is intensifying

　　. The semiconductor industry has only been born for half a century, but it is already facing the proposition of discussing the final end of silicon materials, that is, how far Moore's Law can go. Any industry has a life cycle, and the semiconductor industry is no exception.

　　Looking at the quiet changes in the semiconductor industry in recent years, it is found that the transfer of the global semiconductor industry chain is accelerating. Successful semiconductor factories always develop industries with higher added value, and transfer eliminated industries to many emerging countries. The president of market research company VLSI believes that "the chip manufacturing industry is quietly withdrawing from the United States." Motorola is a successful example of fading from the chip manufacturing industry. After its structural reorganization in 2004, the new company Freescale was quickly listed within a year and turned losses into profits.

　　Another famous American company, LSI Logic, has sold its 8-inch chip manufacturing plant in Gresham and turned into a fabless company. The purpose is to be responsible to shareholders, reduce chip manufacturing costs and reduce the risk of transitioning to 90nm and 65nm process technologies.

　　Recently, there is also a very famous semiconductor company Avago, which is a company spun off from Agilent. It mainly produces chips for various products such as mobile phones, network equipment and printers. The current annual sales revenue is 1.8 billion US dollars.

　　Another feature of the global semiconductor industry chain transfer is the IP trade to the high end of the industry chain. Many powerful and innovative companies have now taken steps forward, with the fundamental purpose of obtaining greater profits.

Japan rekindles hope for reorganization

　　According to the latest estimates by market research company IC Insights, there are about 7 Japanese companies in the ranking of the top 20 semiconductor manufacturers in the world in 2005. Together with Elpida, Hitachi, Mitsubishi Electric and other large companies that did not enter the top 20, there are nearly 10 Japanese semiconductor manufacturers in total.

　　Although the overall strength is still second only to the United States, ranking second in the global semiconductor industry, compared with the rapid development of other regions, it seems to lack competitiveness and is showing signs of decline.

　　In 2002, due to the crisis of the Japanese semiconductor industry, it was reorganized once, and Renasas and Elpida were created. The purpose was to try to restore Japan's glory in the 1980s, but the practice of 3 years has not been effective, and it is likely to lead to a new round of mergers and acquisitions.

　　Sato Fumiaki, who has been voted as the top stock analyst by Nikkei Financial News for 5 consecutive years, has a very insightful and innovative view on the urgent need for mergers and reorganizations in the Japanese semiconductor industry.

　　Sato called on Japanese semiconductor giants to merge and reorganize, and merge into new companies with larger economies of scale in the fields of memory, system chips, and wafer foundry, so as to revitalize Japan's semiconductor industry. Sato believes that the most ideal merger combination plan for Japan's semiconductor industry is to form one memory factory and two IDM system chip factories. The specific possible combination plans are: Toshiba, NEC Electronics and Sony will be combined into one; and Renesas Technology, Fujitsu, and Panasonic will be combined into another, and jointly establish a joint venture to produce IDM system chips. In addition, Elpida will cooperate with Toshiba and Spansion to produce memory.

　　Sato believes that if the Japanese semiconductor industry is unwilling to do so, then another option for Japanese semiconductor factories is to completely abandon wafer manufacturing and become fabless companies, otherwise the prospects of Japan's semiconductor industry are unpredictable.