The US clampdown on AI chip sales to China has reignited the debate as to whether China can become independent on semiconductors, but I think that all the evidence points to China making very little headway towards this goal.
This is not a new issue – it was 2015 when China put together its Made in China 2025 strategy, a part of which was to ensure that 70% of China’s semiconductor consumption would be met by local production by 2025.
But progress to date has been woeful – we are already more than halfway through the ten-year period, and only 9% of domestic consumption is being met with domestic production.
Not for lack of trying
This has not been caused by a lack of effort, as the Chinese state is offering more money in subsidies than any other country in the world to companies who are willing to build a fab there. Geopolitics aside, this would mean that a leading edge fab built in China could become more cost-competitive than anywhere else.
However, so far, this has not had many takers, and those that did have locked up their facilities like a fortress to prevent intellectual property leakage.
The result has been that most of the money has been poured into domestic companies such as Tsinghua Unigroup and SMIC, but this has produced very little in the way of results. Instead, Tsinghua Unigroup defaulted on its loans in 2020, and its CEO has been detained as part of an investigation into corruption and waste.
This is a sure sign that the Chinese state is extremely unhappy with progress to date and is looking for someone to blame for the billions of yuan that have gone down the drain.
China started too late in chip game
The recent move by the US to limit access to cutting-edge AI chips comes on top of bans on shipping cutting-edge manufacturing equipment and EDA tools to Chinese companies, and will almost certainly galvanise China to try even harder.
This is where the debate begins – some observers think that developing the semiconductor independence it needs is just a matter of time and money. If semiconductor technology was static, then this might be a possibility. But semiconductor manufacturing is continually evolving and I think that the Chinese are too far behind to have a realistic chance of catching up.
I think that this is not through any fault or because China is not very good at developing technology, but simply because it started too late.
Semiconductor manufacturing started in the 1970s and the giants of the industry have been building on and evolving their technology ever since. China did not really begin to look at semiconductors before 2000 and so it was 30 years late to the game.
Furthermore, Taiwan, the US, Japan and The Netherlands have taken steps to minimise the transfer of intellectual property to China, meaning that its progress has been even further slowed down.
China’s tech strengths lie elsewhere
The practical upshot of this is that China will never catch up, and for as long as electronic circuits are made from silicon, it will be at a large disadvantage.
However, when it comes to other technologies like AI, autonomous driving, robotics and quantum computing, it is a completely different story, as China began development roughly around the same time as everyone else. There is good evidence that China has some of the best AI algorithms available, highlighting just what it can do when given a level playing field.
I suspect that this is why the US is continuing to increase the pressure, as limiting China’s access to silicon chips will slow its development of the technologies that China is really good at. The net result will be to further damage China’s aspirations to become a global technological power, meaning that it slips even further behind in the race. China has a big hill to climb.