Graeme Maxton

Climate Change Economist and Author

China’s struggle for tech self-sufficiency

June 2021

For decades, China’s output of high-end tech has depended heavily on help from US and other foreign firms. The country has invested billions to become more self-sufficient, with some success. Since the US imposed tighter restrictions however, this struggle has become more urgent.

China’s biggest problem is the enormous complexity of much of the cutting-edge technology it must now develop quickly, and the long experience curve needed for success. A second problem seems to be China’s approach, judging by what’s been happening in the auto business, semiconductors and software.

1 – Automotive

A quarter of a century ago, the Chinese government published a detailed plan to become a global leader in the automotive industry. At the time, the sector was characterised by poor quality and lots of counterfeits. The joke was that Chinese firms believed the word ‘copyright’ meant ‘we can copy, right?’. Without understanding the engineering of what they were doing, local firms couldn’t achieve the fine tolerances necessary, and this meant the quality of Chinese cars was far below world standards. Chinese firms were keen to learn however, and there was that plan, so they bought the best manufacturing kit. They were also keen to export to lucrative rich-world markets.

So what is the situation now?

Surprisingly, it’s not much different. There are still many copy-cat cars, and though quality has improved a bit, exports to the rich world are tiny. Many Chinese carmakers still depend heavily on foreign engine technology and electronics, and their vehicles often achieve a low crash-safety score. Reviews still describe them as ‘agricultural’ to drive. In high-end overseas markets, Chinese cars are regarded as not really good enough, illustrating the decades old problem which has plagued many industries in China – Mr Chabuduo – a development mentality which thinks close-enough is good-enough.

Despite lots of state support, China’s carmakers have not had any more success in electric vehicles, which is surprising as these contain many fewer highly engineered parts. Crash test results in China’s EVs can be shockingly poor. Spontaneous combustion and quality are also of concern, with many new cars recalled for repairs.

Other issues include faulty motors, faulty transmissions, faulty odometers and bad odours (a problem to which Chinese consumers are particularly sensitive). Most notably, according to market research firm J.D. Power, problems are far more common in Chinese EVs than in traditional Chinese-made cars.” (Bloomberg).

2 – Semiconductors

Semiconductors lie at the heart of China’s tech ambitions. They’re in the Five Year Plan. The future of China’s telecoms sector, the military, the auto industry, computing and countless other businesses depend on this effort. So hundreds of billions of dollars are being funnelled into the industry, to thwart America’s technology ban, and boost the country’s current 20% self-sufficiency rate.

It’s not going very well.

In late 2020, one of the sector’s champions, Wuhan Hongxin Semiconductor, which received huge amounts of state financing, and was aiming to produce 14nm chips (cutting-edge is 3nm), collapsed after being exposed as a fraud. The founder had been introducing himself to investors as a vice president of TSMC, the leader in the industry. TSMC is a Taiwanese firm.

Another national champion which received oodles of state funding, Tsinghua Unigroup, has defaulted on bonds worth $3.6bn. It is saddled with $31bn in debt, with much of the money spent on property and bitcoin mining. The company has been dumped by its sole high-profile customer, Samsung. Its memory chip factories, key to the country’s drive for 70% self-sufficiency by 2025, have been delayed.

A third contender, regarded as the best remaining hope, is SMIC in Shanghai. It has also run into problems. Having been placed on a US trade blacklist it is limited to making 14nm chips, with its plans for 7nm chips delayed, partly because of the lack of cutting-edge manufacturing equipment available locally. A leadership bust-up and funding issues have added to the company’s woes.

Among the remaining big Chinese chip firms, Tacoma Semiconductor went bankrupt in 2020. Global Foundries Chengdu Water Fab shut the same year, without producing a single chip, and has been sold to a foreign firm. HuaXinTong closed in 2019 with Imaging Device Technologies suspending operations the same year.

Such misfortunes are greatly because making semiconductors is so difficult. It is not a question of throwing money at the problem, which is what is currently happening. It is not about having a good chip design and then building a foundry. There is a black art to the business which takes years to learn.

EUV and all that

China is especially far behind in developing design, manufacturing and testing tools, some of which have only been mastered by a few firms globally. Production yields matter greatly too, and that depends on scale, which local firms don’t yet have. Being able to produce cutting-edge semiconductors below 10nm also depends on having an Extreme Ultraviolet (EUV) lithography machine. There is only one manufacturer of this equipment – ASML in the Netherlands – and it has been blocked from exporting cutting-edge equipment to China.

China makes its own lithography machines though these can only make 90nm chips. It is trying to make a higher spec machine, initially with the help of Japanese technology, which it hopes will be ready in late 2021. To put the difficulty of this task in perspective, an ASML machine costs $120m and contains more than 100,000 components.

As Chinese firms learn how to do this, and with limited experience, they have tried to buy a second-hand lithography machine, and contracted with ASML to buy a previous-generation one (the US is trying to block the sale). This would allow China to be self-sufficient in 20nm chips by 2023.

China has also been stockpiling other chip manufacturing equipment. Without the skills and recipes needed to use them however, this carries risks. Fujian Jinhua Integrated Circuit, a start-up accused of IP theft, spent more than $5bn on such equipment to make DRAM chips. After being blacklisted by the US, and without access to foreign engineers, it was forced to shut shop.

Given its commitment, China will unquestionably make big strides in semiconductor production during the next decade. The country has many notable technology advances – in solar energy, nuclear power, surveillance AI, high speed rail, space exploration and many other areas. It hopes to achieve the same result now. What is happening in semiconductors is only the latest lap of a catchup race which has been going on for more than 40 years. Success could still be hampered by further US restrictions, however.

If so, a different approach may be needed.

3 – Software

When it comes to software, the situation looks better. Huawei, now unable to access Android software for its phones, tablets and other devices, has developed its own operating system in a remarkably short time, called Harmony OS. There are already lots of apps available for download, and they work seamlessly. Huawei now hopes to end the global duopoly in phone systems. (Android and iOS have more than 95% of the market.)

On desktops, there are plans to replace Microsoft Windows on all government computers by the end of 2022 with a locally developed system called Unified Operating System (UOS).

There are other locally developed server and desktop operating systems too, such as Kylin and Deepin, on which UOS is based. Huawei has also released LiteOS, designed for connecting devices to the ‘Internet of Things’, which seems to be a bit like Bluetooth.

With these developments, China hopes its operating systems can take 20-30% of the local market quickly.

This goal looks realistic because Kylin and Deepin are very good systems. I installed Deepin on a laptop and was greatly impressed. It’s stylish, stable and functions well. It comes with the Chrome web browser, an office suite, media software and lots of other useful apps pre-installed. There are thousands of other programs available to download, and all are free. Everything I plugged into the machine, from USB devices, to screens and printers worked perfectly.

There were some quirks in the English version which displayed Chinese characters at times, and there are claims that the system collects data, though I can’t verify this. But then Google, Microsoft, Facebook and countless other systems harvest people’s data too.

For those who want a system that looks like Windows, China’s Kylin is very good too. The 64-bit version even powers China’s supercomputers, the Tianhe-1 and Tianhe-2, which are among the world’s fastest.

There’s a reason

There is, however, a reason why all these new Chinese operating systems work so well.

In the rest of the world, Kylin is called Ubuntu Kylin. The system is a fork of the Linux desktop system, Ubuntu. It has a wrap-around skin, which is easy to install, to make it look like Windows.

The boast that Kylin powers China’s supercomputers is also less impressive when you understand that all the world’s supercomputers run on Linux.

The reason Deepin is good is much the same. It is based on a long-established Linux operating system (Debian-based this time) which is used around the world every day. That’s why there are loads of apps and everything works first time – the system has enjoyed decades of refinement.

Huawei’s Harmony OS, which the firm says is brand new, also appears to be based on an existing system, with a few additions. Despite China’s GlobalTimes calling Harmony a milestone in the country’s bid for technological self-reliance, and the head of Huawei’s Consumer Business Group saying it “is completely different from Android and iOS”, it looks suspiciously like an Android system which has had the word ‘Android’ scored out and the word ‘Harmony’ written in, in crayon.

As one prominent reviewer put it when the system was previewed in January 2021, there is “no discernible difference between Huawei’s ‘all-new’ OS and Android”. Huawei says every line of coding is different on the phone system. In the beta version however, some of the background files still had their Android name, while clicking on ‘system info’ revealed the phone thought it was running Android 10 which is, incidentally, also Linux based.

While HarmonyOS’ official marketing and development paperwork appear to be intentionally written to obfuscate the fact this it is essentially a forked version of Android, it does appear to be based on Google’s mobile OS, so to say that it is replacing Android on the company’s smartphones isn’t exactly correct.” Source

Of course, there’s nothing wrong with using open source systems like this. Amazon’s Fire OS was created the same way. Open source systems are there for anyone to use or develop. They are available in many languages, with lots of apps, and work extremely well.

The problem comes in using these systems while suggesting they are proprietary. It was the same when The Chinese Academy of Sciences announced to great fanfare in 2020 that it had developed its own programming language called Mulan and it turned out to be a copy based on the 30-year old Dutch coding system, Python.

Of course, the Chinese are coming from behind, and lack some of the skills needed. They are also being hampered by US actions.

Now comes the real test

In software, the big test for China’s systems lies ahead – over the next few years.

  • Can Chinese software developers provide ongoing support for the systems they have created?
  • Can they do this outside China?
  • Can they get foreign hardware makers to install their software?
  • Will other Chinese phone-makers adopt Harmony OS too?
  • Will app developers be willing to recompile their software for Huawei’s phones?
  • Will the alternatives to the popular apps which Huawei is banned from using, as well as those that depend on Google Play to work, be accepted by international consumers?
  • Can these firms ensure their systems are secure, and issue updates in multiple languages? (Linux systems are normally very secure.)
  • And, without having built them from the ground up, can they develop and improve these operating systems better than their rivals?

If not, Chinese software firms may need to rethink their approach.

China is struggling to develop its own cutting-edge technology in the auto business, semiconductors and software.