We should also pay attention to the development of the global semiconductor industry

fighting

We should also pay attention to the development of the global semiconductor industry [Copy link]

Analysis on the Development of Global Semiconductor Industry

2005-01-13 Source: China Electronics News

Market conditions: excitement and confusion coexist 　　The most fundamental reason for the cyclical changes in the semiconductor industry is the growth in demand for global terminal products. The average annual growth rate of the global semiconductor industry is about 10%. But 2004 is different from the past. Because in 2000, the global semiconductor industry reached its sixth peak and created a historical high. In the following 2001 and 2002, the semiconductor industry fell rapidly and the global semiconductor industry was reshuffled. According to past analysis, 2003 should rebound, but due to SARS and the US-Iraq war at the beginning of the year, the global semiconductor industry seems to have entered a period of adjustment again. So 2004 is the long-awaited year of growth. At the beginning of 2004, various consulting companies predicted that 2004 would be a good year, and even before July, there was still cheering. However, the so-called cyclical changes in the semiconductor industry are not entirely as expected. After less than three quarters of happiness, Lehman Brothers, a well-known American investment bank, lowered its future expectations for the semiconductor industry on August 13, and lowered the investment ratings of some large semiconductor manufacturers such as Intel, Xilinx, ON Semiconductor, and Fairchild Semiconductor, citing the unstable market environment at the time. Lehman Brothers pointed out that the sales revenue growth rate of the semiconductor industry in April and June exceeded 40% for the first time in four years, and the shipment growth rate exceeded 20% in five of the last six months. Although the inventory level of OEMs does not seem to be too high compared with historical conditions, Lehman Brothers believes that the high inventory level in the distribution channel and the small room for inventory reduction indicate that the growth rate of semiconductor unit shipments may decline. At the same time, considering that the growth of equipment utilization may peak in the second or third quarter of 2004, the lack of component price increases, and the lack of "killer applications" to increase demand, the semiconductor industry will face a decline. So far, the industry has two different views on the entire semiconductor industry. In summary, one side is basically represented by investment banks and consulting companies, who believe that the inventory is too large. For example, Morgan Stanley believes that under normal circumstances, the average inventory days of semiconductor manufacturers are about 60 to 70 days, but in the second quarter it has reached 74 days, which is higher than the normal level. The rise of the semiconductor industry will show a slowing trend. The other side is represented by IDM manufacturers such as Intel, TSMC (later changed its view) and Applied Materials, who believe that the industry is still in the rising cycle of the semiconductor cycle. However, the mainstream view in the industry can still be described as uncertain, especially for the prospects of the semiconductor industry in 2005. By the end of 2004, the former view seemed to have a slight upper hand. But the vocabulary used is very scientific, not recession, but correction. Looking at the global semiconductor industry in 2004, it is very appropriate to describe it as "high aspirations at the beginning of the year, confusion at the end of the year." It's just that the rising cycle of this cycle seems too short, only about five quarters. 　　 The positioning of semiconductor equipment manufacturers is unclear 　　From the analysis of the semiconductor industry chain, because the equipment industry has a 6-month lead time, it is sensitive to the ups and downs of the industry. There is no doubt that the global semiconductor equipment industry has achieved another bumper harvest in 2004. According to the latest report from SEMI, the sales of the semiconductor equipment industry reached 35.31 billion US dollars, an increase of 59.13% compared with 2003. Why is it said that "the positioning of semiconductor equipment manufacturers is unclear"? The possible reasons are as follows: 1. The development trend of the global semiconductor equipment industry has slowed down significantly. In 2004, the semiconductor equipment industry achieved a rare brilliant achievement, with an increase of 59.13%. According to historical experience, if the growth rate of the equipment industry is greater than 50% in the previous year, it will definitely decline in the next year. Therefore, the forecast for 2005 and 2006 is that after another 2 years of adjustment, it will not be able to recover to the level of 2004. 2. The investment recovery period of 300mm wafer fab is as long as 30 years. Iddo Hadar, the managing director of corporate strategy business of Applied Materials, said in November that so far, the industry has accumulated 20 billion US dollars in expenses for deploying 300mm silicon wafer technology and equipment. Hadar said that the payback period for deploying 300mm silicon wafer manufacturing technology may be as long as 30 years. Therefore, Hadar said in a joking tone: "Most companies that invest in this manufacturing process may not survive for long." Applied Materials also predicts that semiconductor manufacturers are expected to adopt 450mm silicon wafer manufacturing process between 2011 and 2015. When asked whether IC manufacturers can get a return on investment in 450mm manufacturing process, Hadar said: "It is possible to get a return, but the payback period may be longer than the life of the company." 3. Only 10% of the equipment manufacturers in the world can meet 80% of the market demand. Dean Freeman, an analyst from Gartner/DataQuest, believes that only 10% of the equipment suppliers are needed to meet the needs of the IC industry in the next 10 years. The analyst believes that the field of chip manufacturing equipment and materials is becoming more and more concentrated, and only three manufacturers will have a decisive monopoly in a specific field. This trend and the expected downturn in the IC industry in the next few years will undoubtedly accelerate the restructuring and mergers and acquisitions in the semiconductor industry in the future. Gartner also believes that today's semiconductor equipment industry is overcrowded, and for this reason, 40% of semiconductor equipment manufacturers may disappear in the next 10 years. 4. Doubts about following Moore's Law The world's top semiconductor equipment manufacturers have been working hard under the drive of Moore's Law. At this point, we should ask a question: where is the return on investment? Semi's semiconductor process roadmap (ITRs) is like an invisible pusher, such as achieving 90-nanometer mass production in 2004 and 65-nanometer mass production in 2007. When the industry entered 90-nanometer mass production, the equipment industry must be at the forefront, so the equipment manufacturers have already entered 65-nanometer and below, and have invested a lot of manpower and financial resources for this, otherwise it would be impossible to occupy the next market. In a market economy, it is impossible not to mention returns, otherwise how can the industry continue to progress. However, today's global semiconductor equipment industry is facing a decisive period on how to track and develop. As it approaches the limit of Moore's Law, the investment required is very huge and the risk is also high. The key is that there may be only about 10 manufacturers in the world who can afford 65-nanometer equipment. Moreover, in the initial stage, it was only in the research and development stage, and the order quantity was small. As a result, the price of semiconductor equipment remained high, and even Intel said that the price of the equipment was too high. This contradiction will become more and more intense. In the end, only the first and second manufacturers in the world will continue to invest and follow Moore's Law, or like Intel, it is very likely to manufacture equipment by themselves. 130 nanometers and below process technology is blocked. Driven by Moore's Law, the semiconductor industry continues to reduce the size to meet the market demand for cost reduction. However, in the process of size reduction, the most commonly used method is to use a photolithography machine with a shorter exposure wavelength. From 0.5 microns, 0.35 microns, 0.25 microns, and even 0.18 microns and 0.15 microns, there were no major obstacles along the way. However, when entering 0.13 microns and 0.11 microns, the semiconductor industry encountered a "hurdle", that is, the process technology was not easy to achieve, and the outstanding performance was the decline in the yield of the device. The main reason for the analysis is that the replacement of the stepper lithography machine from 248 nanometers to 193 nanometers and the introduction of corresponding new materials and new structures have caused the inadaptability of process technology. For example, Bai Peilin, a spokesman for Nanya Technology in Taiwan, said that due to some problems with the 0.11-micron process technology, Nanya Technology's output increase in the fourth quarter of 2004 dropped from the originally expected 50% to only about 30%, and such problems occurred again on November 16, 2004. Although Nanya Technology's 0.14-micron and 0.11-micron process technologies have reached the mass production stage, Nanya Technology's production cycle has increased by 2 to 3 weeks on the basis of the original 60 days due to the use of 193-nanometer lithography equipment and new materials. Huang Chongren, chairman of Taiwan Powerchip, also said that it can be seen that the biggest difference between using an 8-inch plant to produce 0.11-micron technology and using a 12-inch plant to produce 0.11-micron technology is that there will be a considerable difference in the output qualification rate. Therefore, by 2005, even if the DRAM price falls, the manufacturing cost reduction of DRAM factories using 8-inch factories to produce 0.11-micron technology will be relatively limited, and they may even end up in a loss-making state. In this way, there will be no funds to continue to produce 12-inch factories that help stabilize DRAM prices. Micron Technology reported on November 16 that its 8-inch wafer factory used 0.11-micron technology, causing about 5,000 8-inch wafers to be scrapped. In addition, Intel also exposed similar problems. For example, its latest 3.6GHz Pentium4560 chip product was originally highly anticipated, but the chip brought unexpected troubles. As a result, 3.6GHz chips were once out of stock in the market. For this reason, Dell's Dimension 8400 was forced to cancel its plan to equip the model with a 3.6GHz chip. Then, on July 30, Intel officially announced that the 4 GHz chip originally planned to be launched in the fourth quarter of 2004 would be postponed to early 2005. According to Kevin Krewell, editor-in-chief of Microprocessor Report, and Nathan Brookwood, an industry observer, 5 GHz microprocessors may not appear until 2007 or later. IBM also encountered trouble in producing the new G5 microprocessor, forcing Apple to postpone the release date of its new iMac product from July 2004 to September. The reason was that IBM's contracted products could not be delivered as scheduled due to low yield. In summary, it reflects that in 2004, when semiconductor process technology advanced to 130 nanometers and below, it encountered problems such as declining yields, which slowed down the speed of industrial development to a certain extent.　　DRAM regroups 　　Intel's predicament seems to represent the predicament of the PC industry. Like processors, DRAM products also faced embarrassment in 2004. DRAM, once a barometer of the semiconductor industry, attracted much attention from the industry in 2004 as before. However, this time it seems not to be the ever-changing DRAM prices that guide the market, but a lawsuit. Since 2002, there have been constant rumors about the conspiracy to fix prices involving almost all major DRAM manufacturers such as Hynix, Infineon, Micron and Samsung, and finally there was some progress in 2004. Infineon Technologies AG has agreed to pay a huge fine of $160 million, while four lower-level employees face fines and jail time. A sales manager from Micron Technology may also face fines and jail time, and a sales manager from Samsung Electronics is also involved. Regardless of the final outcome of this case, the decline of the DRAM industry is an indisputable fact: Samsung's latest 70-nanometer process is used for flash memory, not DRAM; Infineon has increased its investment in flash memory and automotive electronics, and outsourced more DRAM production to wafer foundries such as SMIC; Micron has focused more on flash memory and mobile phone CMOS sensor chip products. ISuppli predicts that the global DRAM market will grow by 56.2% in 2004, reaching US$26.7 billion, while it will only grow by 1.3% in 2005. Perhaps 2004 is the last glory of the DRAM industry. After the passion, DRAM manufacturers will become more pragmatic. They will no longer only think about increasing storage density and market share, but will pay more attention to a rich product portfolio and flexible production conversion, as well as paying more attention to the needs of market segments and product profits, hoping to help stabilize global DRAM prices. 　　SMIC enters the IC high-end club　　In the global semiconductor industry, China's semiconductor industry performed exceptionally well in 2004, with sales increasing by 58% from 35.1 billion yuan in 2003 to 55.7 billion yuan (forecast value), equivalent to only 6.75 billion US dollars, accounting for only 3% of the global semiconductor industry's sales of 218 billion US dollars in 2004. China's semiconductor industry is a veritable "junior", and it is not enough to attract global attention for the time being. But SMIC is different. Some people abroad call SMIC a dark horse in the global foundry industry, with sales revenue of nearly 1 billion US dollars in 2004, and it will soon become one of the three major drivers of the global foundry industry. In 2004, SMIC took big actions in all aspects, which some people called "unconventional thinking". Although SMIC is still a "newcomer", it also wants to squeeze into the "high-end club" of global semiconductors. For a semiconductor foundry based in China, such achievements have been achieved just four years after its establishment, which is enough to attract strong attention from the global industry. The following events of SMIC in 2004 shocked the global industry. 1. It was listed on the New York Stock Exchange and Hong Kong Stock Exchange simultaneously, raising $1.8 billion. 2. China's first 300mm chip production line built in Beijing started trial production in September. 3. SMIC's 90nm chip with independent intellectual property rights will be put into trial production in the first quarter of 2005. 4. SMIC's production capacity reached 120,000 8-inch silicon wafers per month at the end of 2004. SMIC is a new chip factory based in China. It is blocked by high technology such as the United States, but SMIC is still determined to participate in the competition of the global integrated circuit "high-end club". Intel 　　is in trouble and Samsung is trying hard to catch up　　. 2004 was a year of trouble for Intel. In January, Intel ambitiously announced at CES that it would enter the LCD market through LCOS chips, but sadly announced in August that the plan would be shelved. Due to problems such as heat dissipation and power consumption, Intel announced in May to cancel the next-generation desktop PC and server chips Tejas and Jayhawk. In June, Intel was forced to recall 10% to 15% of its previously released Grantsdale and Alderwood chipsets because a production defect was found that could cause the system to freeze or fail to start. In 2004, Intel had planned to launch the 4GHz Pentium 4 processor, but at the end of July, Intel admitted that it could not deliver the processor in the fourth quarter of 2004 as originally planned. By October, Intel quietly sent a letter to customers informing them that the 4GHz Pentium 4 plan would be postponed to the first quarter of 2005. It is undeniable that Intel is still one of the most profitable companies in the world. The latest ranking forecast of the top 10 global chip manufacturers in 2004 released by market research firm IC Insights recently showed that Intel continued to maintain its first place, with a market share that was more than double that of Samsung, which ranked second. But this cannot conceal Intel's current predicament. In fact, Intel's announcement to replace the "Intel Inside" strategy with the "Intel Anywhere" strategy means that Intel has realized that it is entering a "strategic turning point" and has begun to actively adjust in the face of the new industrial environment. Whether Intel can successfully transform and maintain its dominant position in the information industry after the integration of 3C is still full of uncertainty. Compared with Intel, which has suffered many disasters, 2004 is a bumper year for Samsung. The huge demand for flash memory in mobile phones and consumer electronics products, as well as the stability of DRAM market prices, have enabled Samsung, the dual dominant player in the flash memory and DRAM markets, to achieve a growth of nearly 50%, with revenue reaching US$15.1 billion. Although Intel's revenue of US$30.5 billion in 2004 is still twice that of Samsung, it is worth noting that Intel's revenue in 2003 (US$27 billion) was almost three times that of Samsung (US$9.6 billion). 　　IBM is innovative　　"Blue Giant" IBM always represents typical American cultural concepts with innovative thinking. Since the beginning of this year, IBM first sold its hard disk business to Hitachi, and then sold its PC business to Lenovo in China. One of the possible explanations is that these are non-core businesses, that is, they have fallen from high-tech industries to traditional industries. IBM's logic is very clear, abandoning traditional industries and moving towards the higher end of the industrial chain, such as services and intellectual property trade. No wonder IBM once developed a new IC variety, but waited for the right time to sell the IP intellectual property instead of rushing to mass-produce it. The American culture is to use knowledge and services to gain greater profits, so IBM's global image is always innovative. Recently, it was reported that IBM will sell its chip manufacturing part in the near future and become a real fabless company. 　　Semiconductor materials have gone through a difficult period. 　　The question worth pondering is that in 2004, the global semiconductor industry followed the peak of the cycle and achieved once again the best performance in history. However, the corresponding semiconductor material industry, including masks, silicon wafers, chemical reagents, photoresists, CMP abrasives and special gases, are not optimistic. In the words of the industry, they are all "large investment and low profit" businesses. In 2004, the total growth of the semiconductor material and chemical reagent market reached 22%, and sales increased from US$10.6 billion in 2003 to US$13 billion in 2004. However, according to the forecast of Information Network, it will face a brake in 2005, with a growth rate of 7.7%. In recent years, it is very strange that even in the rising cycle of IC, the two major categories of mask and silicon wafer companies have almost no profits. Therefore, when the industry enters a downward cycle, the companies are very difficult, so that when the industry jumps again, the companies are unable to rebound. In October 2004, a major event occurred in the mask industry when Japan's Toppan Printing merged with the US Dupont for US$650 million. As a result, Toppan-Dupont became the world's largest mask manufacturer. Prior to this, the global mask supply was mainly divided by four manufacturers: Dainippon Printing (DNP), Toppan, Dupont and Photronics. As for the silicon wafer industry, the industry unanimously believes that the traditional silicon wafer development model has become a thing of the past and a new model must be established. Lehman Brothers Consulting Company in New York said that the only way is to accelerate the reorganization and merger of enterprises. In a relatively stable market, there are too many suppliers, resulting in fierce market competition. For example, at the beginning of 2004, the world's third largest silicon wafer manufacturer, Germany's Siltronic, planned to go public in March to raise US$750 million, but was forced to give up the listing due to various difficulties. According to market research forecasts, the global silicon wafer market will increase from 6.313 billion square inches in 2004 to 6.596 billion square inches in 2005, 6.784 billion square inches in 2006, and 7.201 billion square inches in 2007.