China's Semiconductor Struggle
China's Semiconductor Struggle
Chips being the new oil, they became a critical war resource
Recently, the USA has gone all-in in trying to stop China from turning into a world superpower. A keystone of this strategy has been denying China access to semiconductors.
If you are utterly unfamiliar with the semiconductor sector, you might be interested to read the industry primer I wrote a few months ago. It is not required to understand this article, but it can help.
Escalating sanctions
It started with simply banning China from accessing the most advanced chip manufacturing machines, EUV (Extreme Ultra-Violet) lithography.
This should have gotten China stuck to, at best, the 10-12 nm (nanometer The p a millionth of a meter) chips. Enough for data centers and cars, but not enough for advanced AI.
Problem is that China’s SMIC, the response to Taiwan’s chip juggernaut TSMC, has managed last summer to make 7nm chips even without EUV.
This was followed by a much more aggressive set of sanctions:
First, in October, Washington forced any American working in CHina to choose between keeping his passport or his job. The move is said to have essentially decapitated most Chinese chip firms, which had paid a lot to attract American engineers and managers.
Second, in November, Chinese chip companies are essentially now banned from using any US tech or patents. It bans the export to China of any semiconductor-related software, chip-making machines, and components using even remotely American technology or patent (in practice ALL of them).
Something that's really weird about semiconductors is that in a lot of ways, the way that they have been made is almost identical to when the industry was birthed 50 years ago. It's almost all the same steps but very miniaturized.
That's one of the reasons why America has so much long-term IP: they invented it and it's almost the same, just a billion times smaller.
The strategy shifted from “let’s slow down China's progress in semiconductors” to “let’s kill China's ability to make any on its own”.
Semiconductors, especially advanced ones are vital for running a modern developed economy (factory robots, car computers, data centers, power grid, high-speed trains, planes, etc…).
Chips are also essential for missiles, planes, tanks, and warships.
If the sanctions succeed, China will never be able to challenge the US.
Dependence on foreign chips is a blade to China’s throat the US always had, but, until now, it was not so blatantly threatened with.
China’s chip production
As I have explained in the “blockading China” report, this is just what sea powers like the US does. They leverage their control of trade, key resources and advanced technologies to strangle any wannabe competitor.
And if a war ensues, they use the same weapons of embargo and blockade to win said war.
I explained how China is intensely preparing on the basic supplies front: food, transportation, energy, and metal.
Chips are something they seem to have been aware of as a weakness, but have failed to solve so far. SMIC is able to make 7nm chips, but yield, quality, and production volume are sub-optimal.
The centralized, state-sponsored project “National Integrated Circuit Industry Investment Fund” has been a failure, and is generally considered as corrupt. The fund consumed $45B for little results.
In total, around $100B has been spent for little results in total production capacity or reducing dependence.
In August, some of its top managers came under investigation for corruption, something that could result in a death sentence in China. Obviously, having let Xi be vulnerable to November sanctions did not score them many points.
Overall, China has “okay” chip manufacturing capacities (“foundry”), but is performing poorly in advanced chips. It is also utterly reliant on foreign technology for the machinery that goes into these foundries. Companies like the Dutch ASML or American Applied Materials.
(as a side note, if you are interested, I covered here the investment case for Applied Materials for my client FinMasters).
Re-shoring production?
I must admit I am not the most expert person when it comes to chip manufacturing. So I prefer to rely on actual experts to judge how likely is it for the US to move TSMC foundries.
After enough research, I come to the same conclusion as Bill Bishop, a reliable expert on China who himself relies on Stratechery/Ben Thompson, a top publication regarding tech.
The conclusion seems to be that the US will keep TSMC foundries in the US. And it will not actually provide any extra safety, as keeping them running will depend on supply and expertise straight from… Taiwan.
It seems China is stuck struggling to build an independent semiconductor supply chain. And the US is stuck staying dependent on Taiwan.
No recovery?
I think the recent sanctions will be initially devastating to China’s chip industry. Especially the ban on “components”. This means not only no new manufacturing machines for Huawei or SMIC foundries. But also no spare parts.
These are extremely advanced, precise, and fragile machines. They require regular maintenance and repairs.
In addition, this is a VERY small market. Chip-making machinery is an oligopoly of just a few suppliers. In the short term, the rest of China’s industry might be able to smuggle foreign-made chips from third countries.
But SMIC will not be able to do that for its foundries’ equipment.
Defecting allies
At a first glance, we could think that this is pretty much over. The US has waved its ban warhammer on China's semiconductor capacity, and it will take it years if not decades to recover.
This is a distinct possibility. But this is far from certain. The problem is that yes, China is highly dependent on foreign supply. But said suppliers are also highly dependent on China. After all, “China is the world’s largest semiconductor market”, a 34% of the total market.
Suppliers have no interest to comply with Washington and plenty of reasons to keep selling to China.
Chip-making equipment
The first crack is coming from the most advanced supplier, ASML. It had agreed to keep away from China its most advanced technology , EUV .
But it refuses to give up the sales of its less advanced DUV technology, used for 16nm chips and larger. It is also the kind of equipment SMIC modified to produce 7nm chips. ASML is supported in this by the Dutch government.
Same for the Japanese Tokyo Electron, Canon, and Nikon, who are still selling equipment to China.
This should give SMIC plenty of time to pile up equipment and spare parts to keep running in any conditions.
These companies are not worried about China copying these machineries. At best it will take years. And by then, they will be legacy technology not worth defending anyway. Better to make money from them now.
Advanced chips
Meanwhile, South Korean manufacturers are still selling their most advanced chips to China. Korea extracted from the US a full one-year exemption from the sanctions.
Samsung is apparently planning to sell as much as possible in that time interval, even promising large supplies of the latest 3nm AI chips to Baidu (China’s Google).
Chinese manufacturers and tech companies are likely to build up inventory ahead of time.
The way forward
It is hard to determine the future scenario on these topics, with so many moving parts.
Diplomacy
On the diplomatic front, China benefits from the fact that almost all non-US, non-Taiwan suppliers are refusing to comply. This way, the USA has to be wary of antagonizing - all at once - the Netherlands, South Korea, Japan. This surprising revolt of US allies gives some breathing room for China.
The reason pushed forward by these manufacturers is that without the Chinese market, they would lose a competitive advantage and the money needed for R&D.
There is a lot of truth to that. If they lose the profitable Chinese market, their innovation would likely slow dramatically, which in itself would help China catch up.
If asked, the rest of Asia would prefer peaceful cohabitation with China. And they are getting increasingly reluctant to obey the US orders to start a trade war with the giant of the region, which has by far the largest population and economy.
Nevertheless, considering the diplomatic influence of the USA, it is possible these countries will be brought in line anyway by 2024.
Domestic production
On the industrial front, China has really struggled to replicate the Taiwanese, Korean and Japanese tech. There is some reason to believe this will persist. Too high centralization and government interference are likely to blame, and this is not going away.
I expect some local production to be established, enough for sure to support the need of the military and critical industries like utilities. But it seems China will stay vulnerable to foreign chip supply for a lot longer than it would like.
Geopolitics: Enter Taiwan
As I said before, the impulse to move Taiwanese chip production to the US put China on a deadline. Currently, a war in the region would be catastrophic for China due to insufficient channels to import food, metals, and energy.
But if China fails to grow its semiconductor autonomy quickly, the US will get its own domestic production running, making it less vulnerable to the loss of Taiwan supply, BEFORE China can have its own.
Overall, I see a very dangerous zone in 2-4 years in the future. This is when a multiple of bad decisions might be taken:
China might prefer to attack Taiwan before TSMC transfers chip production to the US, putting everybody on an equal war footing: no chip supply for anyone. Especially if by then, it will have secured supplies of food and energy.
The US might feel they have a strategic advantage over China ONLY until China grows its domestic chip production. So they would have to “act” fast. This would be the “right” time to push for full Taiwan independence and force China to attack in a weak position.
Many possible miscalculations of the other party's intents:
China could see hardening sanctions as a prelude to war and do a panicked preemptive attack, Pearl Harbor style and not dissimilar to the Russian attack on Ukraine.
The US might believe they are still too vulnerable to Taiwan chip supply being destroyed, so they might decide to militarize the island heavily, causing an attack in reaction before it can be too much reinforced.
Any confrontation in the South China Sea or other areas going out of control due to increasingly reckless actions taken by both sides’ navy and air forces.
Tensions somewhere else (Europe, Middle East, India) trigger accelerating mutual sanctions and slipping toward a blockade of China, itself an automatic cause for going to war for China.
Investing takeaway
Chip manufacturing is for me a too risky business at the moment. Yes, it fell from its COVID highs, but the current price does not reflect the real risk.
In addition, exposure to economic slowdowns could hurt the share price of chip makers as well.
At most, I could see a case for Applied Materials (which I did in the aforementioned report for Finmasters) as they supply foundries and make their machines in the US. They are likely to benefit from the effort to build new foundries and would benefit even more from a Taiwan attack, making it an asymmetric bet.
There is also a likely window of opportunity for Chinese consumer stock. Between post-Covid reopening and before a new flaring up on international tensions. Especially with “Ukraine 3.0” holding the attention of the world.
So I expect the most battered stock to rebound, as long as we are not too close to the Blockade of China.
For full disclosure, I bought accordingly some Alibaba stock myself last week.