Advertisement
Advertisement
Semiconductors
Get more with myNEWS
A personalised news feed of stories that matter to you
Learn more
What many people might not be aware of though is that China was close to the US in the 1960s when it came to early semiconductor technology – so close that it had a fighting chance of leading the industry.

How China is still paying the price for squandering its chance to build a home-grown semiconductor industry

  • This is the first in a series of in-depth articles examining China’s efforts to build a stronger, domestic semiconductor industry amid rising trade tensions with US
  • Here we look at how China came close to the US in the 1960s but lost its way, leaving it scrambling to catch up today

China is self-sufficient in nuclear power generation, it has put a man in space and it is leading in many areas of artificial intelligence.

But when it comes to semiconductor production, it remains woefully behind, spending more on imports of the chips that power the electronic gadgets, PCs and military equipment around us today than it does on oil.

Semiconductor design and production is a notoriously complex business, involving decades of expertise and extreme precision – get it slightly wrong and billions of dollars of investment can go up in smoke.

China has long been aware of the need to develop a strong semiconductor industry of its own but the recent trade war with the US, which threatens to cut off critical access to US components for national tech champions, has added extra urgency.

The trade blacklisting of Huawei Technologies by the US in May on national security grounds and an earlier ban on domestic rival ZTE for breaching an Iran sanctions settlement, has laid bare China’s reliance on the US when it comes to semiconductors, a vulnerability that has not been lost on China’s leaders.

“Technological innovation is the root of life for businesses,” China’s President Xi Jinping said in May on a visit to Jiangxi province, state-run news agency Xinhua reported. “Only if we own our own intellectual property and core technologies, then can we produce products with core competitiveness and [we] won’t be beaten in intensifying competition.”

What many people might not be aware of though is that China was close to the US in the 1960s when it came to early semiconductor technology – so close that it had a fighting chance of leading the industry. However, a combination of political upheaval and misguided industrial strategy has meant that after decades of effort, China still lags.

China must face chip weakness squarely, official says

The story of China’s chip efforts goes back to the beginnings of the integrated circuit (IC).

The IC, a collection of electronic components built onto a single piece of semiconducting material, was invented in 1958 by a Texas Instruments engineer named Jack Kilby. Prior to its invention, engineers had to manually wire electronic components together for each device, with early computing machines taking up entire rooms.

Kilby’s integrated circuits solved this problem by enabling electronics to be miniaturised, reducing costs and allowing complex electronic equipment to be used in areas where weight and space are critical, such as in aircraft or space vehicles.

China was fully aware of the importance of mastering IC technology– in the absence of a consumer electronics industry, chips were seen as vital to national defence.

Over the years China had been misled by the idea that all it takes is equipment to produce a generation of semiconductor fabrication process technology
Jackson Hu, former chief executive at semiconductor foundry UMC

Just eight years after its invention and three years after the first ICs went to market in the US, China managed to create its own IC in 1965, placing the country ahead of would-be competitors like Taiwan and South Korea, neither of which had started developing a semiconductor industry at the time.

However, more than 50 years later, China now trails the US, South Korea and Taiwan in semiconductor technology.

After several decades and billions of dollars of investment, only 16 per cent of semiconductors used in China today are manufactured domestically. Last year, China imported US$312 billion worth of chips to meet domestic demand, exceeding the amount it spends on oil.

As trade war clouds continue to gather, Beijing has stepped up calls for self-sufficiency in core technologies like semiconductors, aiming to produce 40 per cent of the semiconductors it uses by 2020, and 70 per cent by 2025.

But in semiconductors, cutting-edge technology becomes obsolete every two years and analysts estimate that it may take China a decade before it is on par with rivals.

China’s ‘core technology’ push belies its weakness in semiconductors

As such, some have suggested China should focus chip efforts on existing strength in artificial intelligence (AI) to develop AI chips, which are specifically designed to process data for AI applications.

But how did China squander its early efforts to be a big player in an industry now worth almost US$500 billion?

Interviews with industry experts and academics point towards a confluence of factors, including the upheaval of the Cultural Revolution and a consequential lack of human expertise, and initial misguided strategy.

“Over the years China had been misled by the idea that all it takes is equipment to produce a generation of semiconductor fabrication process technology,” said Jackson Hu, who helmed Taiwanese semiconductor foundry UMC as chief executive from 2003 to 2008. “They emphasised the importance of equipment investment, but the truth is that equipment is only one of the necessary conditions.”

In the early years, China simply did not have enough know-how to advance semiconductor manufacturing technologies, said Hu, who believes that China’s lack of talent and experience were the primary reasons for this.

“I started high school in 1967 in Taiwan, which was about the same time that the Cultural Revolution started,” said Hu in an interview. “So for the Chinese youngsters my age, they wasted 10 years of their life, and only when the revolution ended could they return to finish high school and then college.”

According to Hu this resulted in a lost generation of engineering talent for China while countries such as the US and Japan pushed ahead with semiconductor research and development. By 1978, two years after the Cultural Revolution had ended, Hu had already received his doctorate in computer science and started working in Silicon Valley, which by then had over 100 semiconductor companies.

This skills shortage stayed with China, even as the country opened up its economy and doubled down on efforts to boost its semiconductor industry in the 1990s with the 908 Project.

Under the 908 Project, the Chinese government spent 2 billion yuan - a small fortune in today’s money - to build a chip fabrication plant under state-owned Huajing Group, which was appointed as the “national champion” to help advance China’s semiconductor manufacturing.

Lucent Technologies was chosen as the primary foreign partner for the technology transfer project to Huajing, and the now-defunct US firm provided not just process technology but also a design library so that the Chinese firm could design its own chips, according to Paper Tigers, Hidden Dragons, a book on China’s electronic industry by Douglas Fuller, an associate professor at the City University of Hong Kong.

But despite Lucent’s help, the firm’s engineers did not know what products to design and the fab – short for fabrication plant – was empty because there was nothing to produce. The firm basically reverse-engineered designs that were already on the market. As such, the ambitious 908 Project was later regarded as a failure.

“Reverse engineering unfortunately does not lead to technological innovation,” said Monique Chu, a lecturer of Chinese politics at the University of Southampton and the author of the book The East Asian Computer Chip War. “The cultivation of human capital takes time, and developing indigenous technology doesn’t happen overnight.”

Despite an avalanche of money, China still does not have a significant portfolio of intellectual property in integrated circuits.

Back in the early 2000s and at the start of his career, Chinese chip designer Alex Chen joined a semiconductor company called Vimicro. The upstart was founded in 1999 in Beijing’s Zhongguancun by a Chinese scientist who had returned from Silicon Valley and whose venture was backed by the Chinese government.

Local media hailed Vimicro as the firm that would “end China’s chip-less history”. Things looked promising.

But Vimicro’s debut on the Nasdaq in 2005 did not go according to plan – the shares priced at just US$10, a tenth of the price Chen had expected. His dreams of financial independence vanished.

“It was then that I realised that China’s so-called, self-developed semiconductor technology was seen as low-end by the West,” said Chen, who subsequently gave up his stock options, resigned from Vimicro and left for Silicon Valley in search of better opportunities in advanced chip technology firms.

He now works for a leading US semiconductor firm, which was a major supplier to Huawei up until the recent US ban.

There was a bright spot when Shanghai-based Semiconductor Manufacturing International Corporation (SMIC) was founded in 2000.

The foundry was established by Richard Chang – now known as the “father of the Chinese semiconductor” – and it was then that China started to make significant strides in its semiconductor fabrication industry.

But by then, mainland China had already fallen behind South Korea and Taiwan. Taiwan Semiconductor Manufacturing Company (TSMC) – now the largest independent chip fabrication foundry in the world – was established in the late 1980s.

Alibaba unveils its own processor as China seeks self-reliance in chips

Today, TSMC is at the frontier of semiconductor process manufacturing technology. As an independent foundry, TSMC produces chips based on designs provided by its clients. At present, firms like TSMC and South Korea’s Samsung are already manufacturing 7-nanometer (nm, or one thousand-millionth of a metre) chips and planning a shift to 5nm technology.

Semiconductor technology generally follows Moore’s Law, which suggests that the number of transistors per square inch on an IC doubles every two years, meaning that chips become smaller but more powerful and efficient.

Despite being one of the world’s top five semiconductor foundries in the world, SMIC only began to produce 14nm semiconductors this year. That makes it three generations – or at least six years – behind its rivals.

In a renewed push, China in 2014 set up the China National Integrated Circuit Industry Investment Fund (also known as the Big Fund), raising 138.7 billion yuan for investment in China’s semiconductor firms. Investees of the Big Fund so far include SMIC, ZTE Microelectronics and Tsinghua Unigroup.

Professor Fuller remains sceptical of this state-led model though.

“There is a frenzy of building up fabs across China. But we need to ask whether these will all be viable,” said Fuller. “They will likely end up building too many with few qualified people to run them.”

China may be working hard to solve a problem that does not exist, according to Jimmy Goodrich, vice-president of global policy at the Semiconductor Industry Association in the US, a Washington DC-based trade group representing the voice of the US semiconductor industry.

According to Goodrich, China cannot – and should not be trying – to upend the global supply chain for semiconductors.

“Smartphones, laptops, PCs, TVs – a big proportion of these electronics are assembled in China. In reality, what is happening is that more than half of the semiconductors that come to China are re-exported for external trade,” says Goodrich. “So it is not accurate for some policymakers in China to say that there is a huge deficit in chips.”

Taiwan bans China-made semiconductors from cable TV boxes

AI chips could be the light at the end of the tunnel though for China.

Alibaba Group Holding, Baidu and Horizon Robotics are already researching and developing AI chipsets for use in everything from autonomous cars to internet of things devices, which are expected to take off with the advent of next-generation, ultra-fast 5G mobile networks. (Alibaba is the parent company of the South China Morning Post.)

The world’s most-populous nation has access to enormous quantities of data that technology firms use to train algorithms for machine learning, computer vision and natural language processing technologies. That in turn could help tech companies design chips that are optimised to carry out AI-specific tasks.

Moore's Law is less relevant for AI chips as this technology depends more on the efficiency of the AI models that process the data rather than the power of the chip itself.

“Algorithm and standard setting are key factors in designing advanced [AI] chips, China should not be limited by much,” said former UMC executive Hu, pointing out Huawei’s leadership in 5G technologies.

“My understanding is that [Huawei has already built AI and 5G] IC chips faster than some US companies.”

This article appeared in the South China Morning Post print edition as: When the chips are down
Post