For more than two decades, China has been striving to build up its own semiconductor design capability, hoping to finally lay claim to that essential building block of the modern electronics industrial chain. As the world’s second-largest economy, China has the means to buy just about anything, and its sovereign wealth fund and large state corporations are putting billions into overseas assets. But there are some things its money hasn’t managed to buy, at least not yet, even with the world’s largest foreign exchange reserve. Advanced chip technology appears to fall in that category. The output of the domestic computer chip industry still falls far short of demand, with China importing 80% of the semiconductors used in manufacturing, and almost 100% of high-tech and advanced chips.
China’s status as the world’s biggest electronics manufacturer has made it the biggest user of semiconductors in the world, accounting for 40.5% of global consumption, according to the report, “Continued Growth: China’s Impact on the Semiconductor Industry: 2011 Update,” by PricewaterhouseCoopers (PWC). The semiconductor marketgrew 30% to US$132 billion in 2010, from US$101.2 billion in 2009. But the top 10 chip suppliers in China, accounting for 47% of all chips used, are all foreign makers such as Intel, Samsung and Hynix. None of China’s own semiconductor companies were ranked among the world’s top 20 in terms of revenue in 2011. Despite its huge sway over markets, it may take another 10 to 15 years for China to have a few “fabless” chip design companies – or companies that outsource the actual fabrication of the chip — that could rank among the top 25 global leaders, analysts say. They point to a lack of experienced chip engineers and designers, and the reluctance among foreign partners to share technology due to concerns over intellectual property rights (IPR) and security related issues.
The government is seeking to revitalize its two-decade old plan to develop the industry, having announced plans to invest US$30billion in semiconductor design and manufacturing in 2010-2020, while local governments plan between US$25 billion and US$30 billion in the same period.
The devil will be in the details, says Greg Linden, a research associate at the Institute of Business Innovation at the University of California at Berkley and co-author of the book, Chips and Change: How Crisis Reshaped the Semiconductor Industry. “It is a lot harder to compete in this industry than all other industries where China can compete successfully. This is not making furniture or making textile or even cars,” says Linden. “It is one of the most advanced manufacturing [processes] on the planet.” He says aircraft manufacturing may be an apt comparison, since such work requires that everything function perfectly because lives depend on it. “You cannot screw around with shady engineering, shady manufacturing, shady assembling; it has got to be perfect,” Linden says.
Limited Progress
Currently, about 20% of China’s semiconductor companies are fabless chip design companies that lack manufacturing capacity. Another 30% are wafer fabrication foundries. The rest are simply engaged in packaging and testing.
The limited progress by China’s biggest foundry, Shanghai-based Semiconductor Manufacturing International Corp. (SMIC), suggests the scale of the problem. SMIC, set up in 2000 by Taiwanese chip industry veteran Richard Chang, has reported full-year losses for all years except 2004 and 2010. In the year ended March 31, 2011, its full-year loss was US$245.56 million, compared with a net profit of $14 million the year before. The company, which has seven wafer fabrication plants in China, reported US$1.32 billion in revenues in 2011, down from US$1.55 billion in 2010.
Upheavals in SMIC’s management have taken a toll. “They have changed management several times and have changed the company’s direction,” says Roger Sheng, research director in the Gartner Semiconductor Group, based in Shanghai. He says the company expanded too quickly and thus was unable to establish a high yield rateor keep the factory in good condition. “I’m sure a lot of it comes down to mismanagement, because they seem to change their strategy every year or two,” agrees Linden. “I think it shows that there's a lot more to competing in the foundry business than just throwing money at it.” However, he adds, “I expect they will eventually find their feet.”
The company refused to allow interviews or provide any responses to questions submitted by Knowledge at Wharton. It also would not disclose the amount of its total investments so far. In 2000, SMIC said it planned to invest US$10 billion in its first five years of operation and also reportedly planned US$12 billion in investment in the five years to 2015.
Given its travails so far, SMIC has focused mainly on chips suitable for household appliances, such as air conditioners, rather than advanced chips, and it has not been up to competing with the biggest global players, such as Taiwan’s TMSC (Taiwan Semiconductor Manufacturing Corp.) and UMC (United Microelectronics Co.), says Linden. The same is true for other mainland foundries. Linden believes it is only a matter of time, however, before SMIC will catch up. Korean companies like Samsung, which is now the world’s leading advanced chip maker, took 30 years to get to that level. “China needs patience. Both manufacturing and designing [semiconductors] is a learning process. You cannot really speed up,” Linden says.
Apart from difficulties with technology and management, the extreme concentration of production among a few industry leaders makes it virtually impossible for new comers to break in, says Tomoo Marukawa, a professor and China economy and industry specialist at the Institute of Social Sciences at Tokyo University. TSMC and UMC are both strong both in design (with chip design asset libraries) and in manufacturing (with high production rates and high quality), he explains. “Chinese companies have not been able to create such a business model.”
This problem is not confined to SMIC and other Chinese chip makers, says Raman Chitkara, a partner and global technology leader at PWC. “It is a maturing industry and is dominated by one company — TSMC. Everybody fails before TSMC, whose profit is bigger than the revenues of the other three, UMC, Global Foundries and SMIC,” notes Chitkara, who is based in Silicon Valley. TMSC’s ability to introduce new technology faster than other foundries gives them an edge in quickly commercializing it. “As a result, everyone wants to use them. If everybody uses them, they achieve the scale which gives them sufficient cash flow to rapidly add capacity. The big become bigger and the small become smaller,” says Chitkara, who is author of the in-depth annual report, “China's Impact on the Semiconductor Industry,” which now is in its sixth year.
Sensitive Information
Putting such issues aside, China’s semiconductor designers and manufacturers still face some restrictions on access to technology due to longstanding worries over use of such know-how in military applications. The U.S. government granted SMIC so-called “validated end-user,” or VEU, status in 2007. That designation removed onerous licensing requirements for dual-use equipment and technology by its facilities in China. But in general, mainland Chinese manufacturers are not able to acquire the latest technology. “The biggest difference from the development of the chip industry in Taiwan is the free inflow of semiconductor technology from the U.S. or Japan. Even though the COCOM (Coordinating Committee for Multilateral Export Controls) regulations were abolished long ago, there are still unwritten and tacit regulations for technology transfer to China by the U.S. and Japanese governments,” says foreign semiconductor industry executive who declined to provide his name.
Foreign companies also have strong doubts about IPR violations by Chinese companies, Linden says. “Foundry work requires a lot of trust because your customers are revealing a lot of sensitive information when they give you designs,” he explains. TSMC, for example, helps its customers complete their designs, a process requiring the exchange of much propriety information. SMIC’s loss in a court battle over trade secrets with TSMC further undermined trust over the issue. “That is going to make it hard for them to get the best customers,” Linden says, “Trust is an important piece of the puzzle, and we should not ignore that.”
Acquiring the technology is one thing. Building on that through innovation and basic research is another. Most Chinese chip manufacturers have focused on seeking short-term profits to the detriment of long-term product development, says a former foreign executive long involved in the China semiconductor business. “Once Chinese companies acquire technology from a foreign company, they want to make a quick profit. If the technology is outdated, they’ll seek another technology transfer,” he says. Government officials and industry executives ignored his advice to study and absorb foreign technology transfers and then come up with their own chip technology, he says. Sumio Kuniyoshi, a professor at Nakamura Gakuen University in Fukuoka City, Kyushu, agrees. “You have to watch the market carefully but have a long term perspective when you invest in the semiconductor industry,” says Kuniyoshi, former senior manager of Toshiba Corp.’s China Department.
China originally was seeking to develop basic technology in this area, having made developing semiconductor technology one of the government’s favored 908 and 909 projects (in 1980 and 1995 respectively). But in 2000, Beijing shifted its policy focus from creating Integrated Device Manufacturers like Samsung, which do design, manufacturing and sales, to contract foundries or OEM foundry makers like Taiwan TSMC (Taiwan Semiconductor Manufacturing Co) and UMC (United Microelectronics Corp.). Since then, foundry production has dominated, though the number of chip design companies in China is growing. In “China's Impact on the Semiconductor Industry,” Chitkara of PWC estimated annual growth in the market at 46%, reaching US$5.4 billion in 2010 from US$178 million in 2001. IHS iSuppli research forecasts that annual revenues of China’s fabless semiconductor makers will roughly double from 2010, hitting US$10.7 billion in 2015. “China’s fabless industry in 2010 benefited from booming demand for semiconductors used in cell phones, as shipments last year of mobile handsets in China surged by nearly 60%,” said Vincent Gu, senior analyst for China research at IHS iSuppli in a press release.
Despite that fast growth, most of China’s 500 chip design companies are still small, and experts are not optimistic about their potential to become global leaders. According to PWC, China’s biggest fabless company is HiSilicon Technologies, a subsidiary of Huawei, the world’s biggest telecoms equipment maker, with US$652 billion in revenues in 2010. Spreadtrum Communications Inc., the second biggest in terms of revenue, designs semiconductors for cell phones. It forecasts revenue exceeding US$500 million in 2011, up from US$369 million in 2010. While integrated circuit (IC) design companies are growing quickly, the market lacks a dominant player. “Chinese companies tend to compete on price. So when you tend to compete on price, you really have a financially challenging position. You may or may not have a sufficient growth margin to fund research and development on a broad enough level to broaden your portfolio,” Chitkara says.
Many of China’s 500 fabless companies lack products but are good at winning government subsidies, says Linden. “Most of those 500 companies do not have any kind of sales. Some of them are doing design outsourcing; they may be providing design services. Some of the larger fabless companies listed in the PWC report are providing services but not products. There is a world of difference.”
Scarce Talent
Most chip companies now employ as many embedded software engineers, to provide a full range of services, as they do chip designers. “They have to provide their customers complete packages that can be integrated quickly into end products. This in turn requires a deep understanding of the systems that the market will be demanding at the end of the 18 months it takes to design a chip,” Linden says.
Shortages of engineers and managers with more than just a few years of experience have prevented the industry from achieving a certain level of maturity, he adds. “And even if you could wave a wand and create a plentiful supply of these valuable human resources, you wouldn't have many competition-tested companies that know how to compete in a non-commodity business like semiconductors.”
That may seem strange, given reports that China’s universities turn out a half-million engineering graduates a year. While many very smart and talented students emerge from good engineering programs, many have no intention of pursuing that field as a profession, says Ben Cavender, an associate principal and senior analyst on the electronics sector at the China Market Research Group based in Shanghai. Many opt for an engineering major for the sake of testing into a particular university, but then shift to accounting or consulting once they graduate. Also, engineering graduates often attend two-year colleges rather than four-year universities. “They do not have the knowledge to be able to design a new technology,” Cavender says. Apart from that, Chinese design companies need recruits with at least several years of experience at a well-established IC company. “You have to find people who have working experience at a chip company in the U.S. or somewhere. That is hard to find and they are pretty expensive,” he adds.
Since China lacks a large pool of well qualified chip designers and engineers, its design capability is heavily dependent on recruiting from overseas. Industry sources say SMIC has many Taiwanese, Korean and Japanese engineers. SMIC declined to comment on this issue. “Japanese engineers, who maybe be retired or laid off by restructuring in Japan, have in some ways transferred Japanese technology to Korean companies, helping them improve their technology,” says Hiroshi Koezuka, professor of the college of business administration at Ristumeikan University in Kyoto. “In the same way, many Japanese engineers who are being laid off and looking for jobs may work for Chinese companies, and their technology is likely to improve,” he says.
China should also adopt a more aggressive and integrated strategy for nurturing the chip design and manufacturing industry, says Choong Y. Lee, a professor of marketing and management at Pittsburg State University. He notes that Japan, whose NEC, Hitachi and Toshiba dominated the industry in the 1990s, ceded its one-time lead in this area to Korea’s Samsung by responding too slowly to market trends. Samsung, conversely, made huge investments in the 1990s and 2000s that at the time were viewed by Korean experts and economists as too risky. But by 2005, Samsung had become the world’s second-largest semiconductor maker.In terms of revenue, Toshiba ranked fourth in 2011, according to IHS iSuppli Semiconductor Value Chain Service.NEC and Hitachi merged their other semiconductor businesses with Mitsubishi Electric in 2003, setting up Renesas Electronics Corp., which now ranks 5th but has been in the red for the past seven years and is launching a US$1.25 billion recapitalization and restructuring. The DRAM (dynamic random access memory) businesses of the same trio,NEC, Hitachi and Mitsubishi Electric Corp., were merged in 1999 to form Elpida Memory. It was ranked 15th globally in 2011 before declaring bankruptcy in February.
The Japanese manufacturers made the mistake of disregarding global market trends, industry experts say. “Memory was their mainstay, and when the Korean firms (and Micron in the U.S.) learned to make memory chips that were cheaper but still good enough for the PC market, most Japanese firms failed to adapt. The fabless model hasn't caught on in Japan, so innovation was locked up in large, bureaucratic conglomerates,” Linden says.
Looking ahead, China could still move up the technology ladder given the fast-changing fortunes in this field. Instead of following Japan’s approach in sticking with IDM, China may best follow the example of Korea, which has a couple of big IDMs and a few dozen fabless companies, some of which have had modest success outside of Korea. Much depends on China’s ability to lure top-notch designers from other big industry players in Japan and elsewhere, Koezuka says. “It would be highly surprising to me if there is not at least one Chinese chip company in the top 25 worldwide in sales within 10 years,” Linden says.
