CHIP SHOT

On October 4, SMIC (Semiconductor Manufacturing International Corporation), China’s biggest chipmaker, confirmed that its imports of accessories and raw materials from the US had been further restricted, and an export license is needed before they can be sold to SMIC.　

The restrictions in technology imports against Chinese companies comes when the country is still reliant on imported chips. At the World Semiconductor Conference 2020 held in August in Nanjing, Jiangsu Province, Wei Shaojun, vice-chairperson of the China Semiconductor Industry Association, said that China imported US$300 billion worth of chips in 2019 despite its improving capability to manufacture lower-end chips. He predicted that as long as nothing out of the ordinary happens, chip imports would remain at US$300 billion in 2020 for the third straight year. 　

Before the 1980s, China went all-out to develop a semiconductor industry while isolated from crucial technology. Since the 1990s, it has increasingly relied on imports. In the world’s biggest semiconductor market, chips have surpassed oil as its biggest import for several years. This heavy dependence has become somewhat of an Achilles’ heel for China, making it even more urgent to develop a domestic chip industry and become self-sufficient.　

In China, enthusiasm for developing the chip industry, often referred to as the IC (integrated circuit) industry in China, took off in 2014 when the State Council published an outline enhancing IC development and the China Integrated Circuit Industry Investment Fund, which is better known as the Big Fund, to invest in promising companies. In the development spree that followed, semiconductor projects boasting millions of yuan of investment sprouted across the country. 　

However, lack of capital stymied development on several big projects. Other issues arose, such as haphazard investments and redundant development of technologies with low thresholds. This shows a need for State-level deployment and supervision of the industry and a bigger role for government support and guidance in the chip sector, experts said.　

Worse, China lacks equipment and raw materials for chipmaking. Silicon is purified, melted and cooled into an ingot before being sliced into extremely thin “wafers,” the basis for producing a semiconductor. Producing large silicon wafers less than 1-nm thick is a difficult technical process. Wei told NewsChina that most of China’s larger-sized silicon wafers with a purity above 99.9999 percent are imported. 　

Photoresists, a photosensitive material used for lithography necessary to transfer a circuit pattern to semiconductor chips, mainly comes from Japan and the US. China also lags in crucial lithograph machines, which transfer patterns onto a thin film (the photoresist) on the wafer. 　

In addition, the supply of other equipment and materials required to manufacture advanced chips, such as etching and ion implanting machines, are all mostly dominated by companies in Europe, the US, Japan and South Korea.　

Under these circumstances, China relies on imports for core processor and memory chips. A veteran industry insider told NewsChina on condition of anonymity that while domestic companies have made some progress in recent years, they still import over 90 percent of the required manufacturing equipment, raw materials and industrial software.　

China is not totally independent even in design. The market for electronic design automation (EDA), which refers to the tools needed for integrated circuits, is dominated by two US companies and Mentor Graphics, a Siemens business. On September 14, US chip giant Nvidia announced it was buying UK-based Arm Holdings, which designs the architecture for chips used in 90 percent of the world’s smartphones and tablets, from Japan’s SoftBank for US$40 billion. There are worries that this may allow the US government to further restrict the use of Arm architecture in China.　

Driven by market demand and favorable policies, the IC industry expanded rapidly. In 2016, investment in China’s chip designing and manufacturing both surpassed 100 billion yuan (US$14.95b) for the first time. Local governments flocked to establish investment funds to attract projects. As of October, there were more than 50,000 registered chip related companies. In 2020 alone, more than 12,000 chip companies were registered, according to company data platform Qichacha. 　

Behind the rapidly expanding industry are bubbles at risk of popping. In July, Dongxihu District government of Wuhan, Hubei Province, posted on its website that Hongxin Semiconductor Manufacturing Company (HSMC) could halt its project for lack of capital. The company, established in 2017, attracted investment totaling 128 billion yuan (US19.13b). Its project was listed as one of Hubei’s major projects in 2018 and 2019 and had been top of the list of Wuhan’s major projects in 2020. The province has since removed HSMC from its list.　

In recent years, it has become increasingly common to see semiconductor projects fail or teeter on the brink of failure because of broken capital chains. According to a September investigative report by Outlook, a weekly magazine under the Xinhua News Agency, at least six projects across five provinces boasting billions of yuan in investment were halted or declared bankruptcy. 

In May, a wafer fabrication project that started in 2017 with investment from Global Foundries, a leading US wafer fabrication company, and the government of Chengdu, capital of Sichuan Province in Southwest China, suspended business. The planned investment for the project was nearly US$10 billion. This July, Tacoma Nanjing Semiconductor Technology (TNST) went bankrupt after launching a reported US$3 billion government-backed chip project in Nanjing, Jiangsu Province in 2015. 　

What all these failed projects have in common are local governments as their chief financiers while project initiators invest little to none of their own funds. Li Ruiwei, who launched TNST, was involved in three semiconductor projects with local governments in Jiangsu and Zhejiang provinces between 2015 and 2019. Local governments took on all the financial risk. Shen Yinlong, vice director of Nanjing’s economic and technological development zone who helped introduce TNST’s project in Nanjing, said the project, the first one in the zone, was started in response to the country’s call to develop the semiconductor industry. The company sought an investment of 2-3 billion yuan (US$299m-448.5m). The local government declined. Then a similar project involving Li, who has been described in Chinese media as a “speculator,” was launched in Huaiyin District of Huai’an, Jiangsu in 2017. The district government spent almost all of its general public budget that year, more than 2 billion yuan (US$299m), to start the project. 

Gu Wenjun, a semiconductor industry expert, told NewsChina that a current concern is that local governments are both investing and engaging in IC without enough knowledge about the industry. Instead, he called for local governments to set the stage and improve the business environment to attract investors. 　

“Many semiconductor projects are opportunistic. Local governments are so eager to get a project first that some projects are started without acknowledging their situation. Without investment or deep ties to the local area, the projects can retreat whenever they want. Some projects were established simply to cheat money,” noted Shi Qiang, an IC expert from the CCID Research Institute under the Ministry of Industry and Information Technology. 　

According to the interviewed analysts, most places can neither meet the capital, technology and talent-intensive requirements of the semiconductor industry nor invest as much as promised. Only a few projects are sustainable in the long term.　

Also, the industry is struggling with redundant investment in low-level technology. TNST, for example, first promised to build an industrial park for contact image sensors (CIS, chips that can be used in smartphone or security surveillance cameras) in an agreement with Nanjing in 2015. After the project resumed in 2017, both parties realized CIS chips were too challenging and settled for second-best, less-demanding analog chips. Later that year, another of Li’s projects engaged in CIS chip manufacturing launched in Huai’an. 　

Cao Yun, who then worked in STMicroelectronics, a global leading semiconductor company, said TNST’s obsession with CIS chips confused him. “There were already leading Chinese companies developing CIS. Some foundries already owned mature projects and technology. TNST was not competitive enough to enter this arena nor it was necessary in terms of the national development strategy,” Cao said. 　

Investment is also chasing semiconductor power devices used for electricity control and conversion. According to Shi Qiang, many companies just produce, or plan to produce, lower-end power devices, after they fail to produce more advanced models. “There are too many products at the same technological level on the market,” Shi said. 　

There is also hype surrounding projects boasting third-generation semiconductors. Chips made with silicon are considered first-generation. Third-generation semiconductors are made from a compound of carborundum and gallium nitride. The anonymous analyst told NewsChina that third-generation semiconductors are not comparable with other semiconductors and not a replacement. “They have different uses, but their market share is less than 1 percent. Many projects use new concepts [like this] to reel in local governments.”　

Wei pointed out that the planned capacity concentrates on chips of 40-90 nm may cause excess capacity in the domestic semiconductor industry. Combined with a lack of capacity for making advanced chips, redundant development will dilute investment directly. 　

Cao Huanshi, founder of Shanghai-based Qiezi Management Consulting which specializes in IC, observed that few among the 3,000 chip design companies possess their own intellectual property or can produce high-end chips. However, many seek a public listing as soon as they design a new chip. 　

“There is a recklessness among local governments and investors chasing the emerging industry, which causes bubbles,” Shen said. 　

“But we ignored the feasibility of investment,” he said, adding that State- or provincial-level authorities should organize experts to assess the feasibility of future semiconductor projects. 　

Wei Shaojun bluntly called out the mushrooming semiconductor projects in the country as solely being investment-driven. “Few local governments can tell who the target customers of their projects are, which field they specialize in or what their competitive strategy is,” Wei said. He believes there is no top-level plan for semiconductor projects. Guidance from China’s top planning body, the National Development and Reform Commission, which arrives once a project starts, should start earlier, Wei said.　

Meanwhile, experts are calling for more investment in the field, particularly for research and development (R&D). In 2019, China’s Big Fund released over 200 billion yuan (US$29.9b) in a second-phase funding round to fuel the chip industry. At an industrial conference held in Foshan, Guangdong Province in September 2019, Ding Wenwu, president of the Big Fund, said the new funding round would continue to support companies already engaged in fields like etching machines, membrane equipment and testing and cleaning equipment, and would encourage companies to delve into core equipment like lithograph machines. 　

Besides equipment, experts suggest increased investment in raw materials and industrial software. Overall investment is not enough either, according to Xie Zhifeng, a co-founder of SMIC who won the Intel Corporate Achievement Award in 1990. “The first phase of the Big Fund invested US$5 billion a year on average, which is less than the yearly investment of Intel which reaches over US$12 billion,” Xie said. Intel’s annual report shows that since 2015 its R&D spending exceeded US$12 billion. 　

The semiconductor industry is more in need of an increase in sustained investment in R&D, Wei noted. He said that only when R&D investment surpasses 15 percent of a chip company’s sales will its development enter a positive cycle. In the US, the proportion is 17 percent while in the Chinese mainland it is only about 8 percent. What’s more, in China investment plans are usually made every five years. This pace of investment is too slow to allow the efficient upgrading of semiconductor products, resulting in unsustainable investment for R&D, Wei said. He suggested the government increase investment in R&D at the State level and set up a special fund for it. 

China’s IC sector also has a human resources shortage. “We lack talent in general, from leading figures, mid-level and senior managers to R&D technicians and engineers,” Xie said. In July, Huawei CEO Ren Zhengfei visited several prestigious universities in Shanghai and Nanjing, Jiangsu Province, stressing the significance of enhancing cooperation between companies, colleges and research institutes in scientific innovation and talent cultivation. 　

Under these circumstances, mobilizing the country’s strength to develop core technologies, as proposed in the new policy to enhance the IC and software industries released by the State Council in August, is regarded as feasible by experts. The policy sets out preferential tax and duty rules for the semiconductor industry. In the 1970s, Japan put forward a plan to develop its IC industry to narrow the gap with the US. Its government mobilized five tech giants, including Toshiba and Fujitsu, to make technology breakthroughs. Between 1980 and 1986, Japanese companies’ share of the IC market increased from 26 to 45 percent. 　

But in mobilizing different forces for this purpose, the government should ensure those charged with overseeing technological breakthroughs are qualified, Wei said, adding it should coordinate the entire process instead of just taking notice of the results.　

Wei said that in the first decade of the 21st century, China’s IC development was mainly market-driven, while the government’s role was absent. “In the past 10 years, though, the government started playing a role and emphasized original innovation. If future policy is well-crafted and sufficient investment is pumped in, China’s IC industry might realize self-sufficiency in most aspects by 2030. By 2040, it may lead the race in some fields,” Wei predicted.