China's chip industry is roiled by a series of defaults and management squabbles. Douglas Fuller, professor at the City University of Hong Kong, says what's behind the turmoil, how the incoming Biden Administration will affect the trade wars, and why political risk in the global semiconductor sector remains high.
At SMIC, the People's Republic's largest semiconductor manufacturer, there has been a mysterious departure from the top management this week. The news is all the more surprising since China has set the development of its own chip sector as a top priority and has promoted its ambitious plans with massive investments in recent years.
What is behind these dramatic events? «In China, you can never predict when the state is going to encroach», says Douglas Fuller. Few Western observers know the Chinese technology sector, and the semiconductor industry in particular, as well as the economics professor at City University of Hong Kong. His 2016 book, «Paper Tigers, Hidden Dragons: Firms and the Political Economy of China's Technological Development» has established itself as a standard reference.
In this in-depth conversation with The Market/NZZ, which has been edited and condensed for clarity, Mr. Fuller discusses the current turmoil in China's semiconductor industry and says what's next for the telecom equipment behemoth Huawei as well as the broader Sino-American trade conflict under President-elect Biden. He also explains why the growing tensions around Taiwan pose a significant risk to the global chip production and thus to the world economy.
Professor Fuller, a series of defaults and chaos at the top of China's biggest chip companies are shaking the country's semiconductor industry to its core. How do you interpret the developments?
I’ve always thought that if China had one big advantage - and it’s an advantage other countries have used in the past to advance in the semiconductor industry - it’s about willingness to spend money. But I also always thought that eventually, with all the problems with very high leverage in China, they are going to have to deleverage. So the question has always been what it means for China’s industrial policies. I think it means that those industries, which have been heavily reliant on sheer volume of investment, are going to be in trouble.
So what are the consequences for the chip sector?
I don’t have any pretense to be able to see into the inner workings of Zhongnanhai, the central headquarters for the Chinese Communist Party. But there seems to be at least a willingness to flirt with the idea of not only letting Tsinghua Unigroup get into trouble with its domestic bonds, but also with international lending. So this could be the start of a serious recalibration in how much money China is going to spend in the chip industry. Or, it may be a dramatic effort to signal to everyone that they have to scale back a bit. It may just be one of these «kill the chicken to scare the monkey» kind of things, as the Chinese say.
Just two years ago, with US-China tech tensions heating up, Tsinghua Unigroup’s management vowed that the company would spend $100 billion over the next 10 years to build a world-class domestic semiconductor industry. How serious is the current crisis surrounding the colossus?
Not too long ago, Tsinghua Unigroup wanted to buy Micron. But where were they going to get the money for that? From the Chinese financial system. So the fact that they are unable to get the money to pay the principal on various bonds signals that, for whatever reason, they’re being cut off. There are one or two possible explanations for this dramatic, salient failure: One, the Chinese government doesn’t really want to spend less on the semiconductor industry, they are just not going to spend it on Tsinghua Unigroup. Two, they want to spend less money in general and this is just the biggest firm out there.
How does this affect the race for global chip dominance?
Tsinghua Unigroup was probably the number one vehicle through which China was pushing investment in the IC industry. From what I understand, Tsinghua Unigroup does not have a majority of the shares of Yangtze Memory Technologies. So you can have Tsinghua Unigroup get into trouble and still foster YMTC’s NAND chip production. It’s similar in the case of ChangXin Memory Technologies, China's first DRAM chipmaker, pushing ahead with its projects in Hefei. These firms are very dependent on access to copious amounts of patient capital to ride out the vicious swings in the silicon cycle in that particular segment. This is something that Korea did very well in its move to take over the DRAM market from the Japanese and maintain their lead. The good thing is that these firms seem to be professionally run by people who are private entrepreneurs. But in China, you can never predict when the state is going to encroach. Just look at the blocked IPO for Ant Financial.
What does this say about government subsidies in the global semiconductor industry in general?
Look at the early days of TSMC in Taiwan, the world’s leading semiconductor foundry. It was basically owned by the Taiwanese state, but they gave TSMC founder Morris Chang all the leeway he needed as a successful entrepreneur. One thing that helped him is that the Taiwanese state didn’t want to spend a lot of money. He had to try to figure out how to make do without much money, get partners. Essentially, the Taiwanese government knew enough to know that they didn’t know anything about the semiconductor industry. In the end, TSMC became spectacularly successful. Maybe for ChangXin or YMTC that's what can happen. But there’s always the danger in China that there will be just too much state interference. President Xi Jinping’s recent rhetoric is all about national security. He never seemed to understand anything about the economy. That’s not necessarily a great environment for China’s chip industry going forward.
China is years behind when it comes to leading-edge semiconductor production. Are Chinese manufacturers ever going to be able to compete with established firms like TSMC, Samsung, or Intel?
A lot of people in the West like to talk about how the Chinese government is very farsighted. I think that’s at best an exaggeration, and in some aspects may be totally wrong. There has been a lot of waste going on in China’s chip industry, particularly on the fabrication side where every local government wanted to encourage to have a fab in their area. For a while, people were very worried that in the memory market you suddenly see something akin to solar panels where China just wiped out the competition. But the problem is that the technical barriers are much higher. In solar panels, the technical barriers aren’t high at all which is why you had every other Chinese city funding solar panel makers. I’m not saying that YMTC and ChangXin won’t succeed. But in the short-term, they’re not going to be huge players in this field.
This week, an abrupt departure in top management has caused uncertainty at SMIC. What is going on at China's largest chip foundry?
Here’s the key question: Is it really in the interest of SMIC as a commercial enterprise to try to get to the cutting-edge of technology? Arguably not. They’re much better being what they were before: A fast follower, a couple or more generations behind, still making money. It’s very hard to be a TSMC. Basically, only Samsung is able to do what TSMC does and make money off of it. SMIC has never been as operationally efficient as TSMC. Also, it’s a smaller firm with less money to spend on R&D, and it will be much harder now because I don’t think they are going to get advanced EUV lithography equipment from ASML in the Netherlands. So it’s quite a tall order to insist that they go for broke and try to catch the technology frontier. And this seems to be very much a policy from on high.
President Xi has declared that China aims to cover 70% of its semiconductor demand from local production by 2025. How realistic is this goal?
I don’t think it’s a very realistic goal, unless semiconductor consumption totally collapses in China. There are still a lot of multinational firms running fabs in China. Intel left because they sold their fab. But SK Hynix, TSMC, UMC, Texas Instruments and a bunch of smaller firms all have fabs in China. Going forward, because of political risk, some of these multi-nationals will think twice about production in China. Both Korea and Taiwan are investing a lot of money in their own chip industry. Behind the scenes, they are probably going to try to put pressure on their firms to not have new investments in China. And, if you’re an American company you have to ask yourself: Do you really want to be putting a fab in China right now, given how bad Sino-American relations are?
Over the course of Donald Trump's presidency, the trade war has increasingly become a conflict over technology. What can we expect from Joe Biden in this regard?
Biden’s approach will perhaps be a little bit looser in terms of the export controls. It’s also firm dependent. SMIC was caught red-handed doing things they weren't supposed to do in terms of working with researchers of the People's Liberation Army. We can argue whether these things really matter in the long-term when it comes to transferring technology to the Chinese military. But that’s an area where I don’t see that China has any allies in Washington, even if semiconductor capital equipment firms want to sell to Chinese firms.
However, particularly the share price of equipment manufacturers such as Applied Materials, ASML, Lam Research and KLA rose sharply after Biden's election victory.
It’s maybe hopes that the entity controls on Huawei will be relaxed. Before the US election, the stock price of these firms was already pricing in some of the worries if the Trump Administration expands these controls to other firms in China. Now, in the view of investors the thinking is: All right, we’re going to have a much more settled market with less political risk than we would have had under a second term of the Trump Administration.
Huawei is at the center of the tech war. What will happen to the leading Chinese telecom supplier under the Biden Administration?
First of all, the situation has already evolved. What the US military was really worried about was Huawei telecommunication equipment in the telecommunications infrastructure of first world allies. Now, a lot of those countries have rejected Huawei. Even Germany is turning away. So it looks better and better in the perspective of the US military. Also, I guess the US military never worried about the developing world because that’s more of a ZTE play, China’s other large telecommunications gear vendor. ZTE is going to sell 5G equipment all over the developing world, and the US apparently doesn't care. That’s why I have to assume that they primarily cared about Japan, Korea and European allies.
Does this mean the US death sentence against Huawei is basically overturned?
The Huawei controls were badly designed and not multilateral, so they were actually undermining the future of US-based semiconductor industry activities, which is a bad public policy outcome. Overall, Trump was probably the least persuasive president to any of America’s allies. Because of their own calculations and the actions of the Chinese government, they've turned against Huawei. With that context: Does it really make sense to try to kill off Huawei’s telecommunications infrastructure business, if Huawei anyway can’t sell to the countries the US government doesn't want them to sell to? That would argue that the US will loosen up these entity lists controls a lot, and there will be a huge American business constituency for that.
Taiwan is at least as important for the semiconductor industry today as the Middle East was for the oil industry for a long time. U.S. industry leaders such as Nvidia, AMD, and Qualcomm have their chips manufactured by TSMC; nearly 70% of all semiconductors worldwide are manufactured on the island or pass through it during the production process. Doesn't that pose a high risk from a geopolitical perspective?
That’s right, but Biden did not run on being tougher on China than Trump. The one exception would be the humanitarian issue around Xinjiang which is a separate issue. It’s not all surprising that some Chinese firms ended up on an entity list. The bad odor around everything going on in Xinjiang creates a very powerful alliance from left to right, from human rights activists to people who just hate China because it’s China. This has nothing to do with commercial considerations. On top of that, most of these firms are pretty small. So some loosening up of these controls and no dramatic expansion is what I would guess the US approach is going to look like. That’s as long as China doesn’t react in some dramatic and unforeseen way, for example starting to fire missiles at Taiwan.
After breaking the resistance in Hong Kong, China is now putting Taiwan's army under increasing pressure with irregular military tactics. How big is the danger of open conflict?
China does have people who obviously watch Taiwan very closely. They interpret it through a very CCP kind of lens, but a lot of them have a pretty decent understanding of Taiwanese politics. They may not know how they can try to change things, but they know the trends: It’s just obvious that the vast majority of the Taiwanese are completely unenthusiastic about any sort of re-unification with the PRC as it’s currently constituted. The Taiwanese median voter basically wants to keep the status quo which is de facto independence.
What does this mean for the future of Taiwan?
The Chinese realize that claims of common bonds, or at least eventual unification are weak in Taiwan. So they’re increasingly worried. What’s more, they look at mismanagement in the West, whether it was the global financial crisis of 2008-09 or the fact that nobody in the West can control Covid. So they’re increasingly confident about their system, and there is a lot more investment in the military. For all these reasons, China is becoming more aggressive about Taiwan. Problem is, the Taiwanese haven’t done a great job of having a very robust defense. They’re facing this very dangerous international situation, but they’re not addressing it the way let’s say Israel does in the Middle East. The Taiwanese government is pretty frugal. It’s kind of a center right country. They don’t like to pay high taxes, and that sort of dictates that they’re not going to spend a lot of money on defense. But that’s the one thing that should be a big priority.
How long can this go on? Will China at some time decide to risk a «hostile takeover» as it did in the case of Hong Kong?
One of the reasons China is willing to be so tough against Hong Kong is probably they knew they had no chance of winning the hearts and minds of Taiwan. Taiwan was not interested in the One Country Two Systems «thing» anyway, so getting rid of this idea wasn’t going to affect public opinion in Taiwan. China will continue to ratchet up pressure and try to steal Taiwan’s remaining allies. Then again, the ultimate move of an actual invasion of Taiwan has such high costs.
In Hong Kong, we've seen arrests of pro-democracy protesters and opinion leaders. How worrisome is this for you personally?
I haven’t been personally affected. But you don’t know what will happen, so I can’t say there are zero worries. I guess I’m unclear how worried I should be.
Let’s get back to the semiconductor sector. You've been following the industry for quite some time. What advice would you give when it comes to investments in semiconductor stocks?
I’m not a stock picker, so I can’t tell whether a company is worth its stock price. But I can tell what a good underlying company is. For instance, I still think TSMC is a great company, but they have this political risk issue. ASML is a great company as well. That’s probably already reflected in the stock price, but it looks like ASML is going to have a really nice, long run. Even if they don’t sell anything to China, other manufacturers will buy their EUV equipment which is essential for cutting-edge chip production. So they are very well positioned to do well over the next five to ten years.
What other companies are attractively positioned?
It’s complicated because there are these political risk factors. If they weren’t around, I would say KLA, the American equipment vendor, looks very solid in terms of having a strong hold over some niche areas that are likely to grow. They are very strong when it comes to a lot of the instrumentation at the end of the fabrication process, particularly for high-end process nodes. What’s more, any firm producing automotive chips will have a market that just keeps growing. In the medium-term, that’s a very promising area.
In contrast, the market leader Intel is at risk of losing the technological race against Samsung and TSMC. Can Intel catch up after its latest production blunder?
I’m sort of sympathetic to the idea that by moving away from being run by engineers to being run by bean counters Intel kind of lost its way. Here, the US government could play a role in helping them. If the US is serious about having its own player in the EUV based chip production, I think the government should step in. Because even if you’re a very large firm like Intel, as a pure commercial consideration in the short-term, given the operational risks involved, you have to ask yourself how worthwhile such a strategy is.
And why should the U.S. government intervene?
It’s a legitimate argument that the US shouldn’t rely so much on fabrication being made in Taiwan since China doesn't recognize it as an independent state. They have always said they want a peaceful solution, but when push comes to shove, they’ll go to war over Taiwan. And, it looks like it will be harder and harder for the US to help defend Taiwan. So maybe the US should be building fabrication in America.
TSMC has announced that it will build a plant in Arizona.
I’m not too enthusiastic about spending a lot of money on a TSMC fab in the US. I have great respect for TSMC, but they’re a Taiwan based firm that’s very Taiwan oriented. The Taiwanese government still has one of its seats on TSMC’s board. By design, the Arizona fab they’ve announced to build in Arizona is going to be behind the times. It’s not going to be cutting-edge. So the US is spending a fair amount of money for not that much return. TSMC has a track record of setting up fabs in the US and letting them get further and further behind.
Does this mean that only Intel would be able to defend a leading position for the US in semiconductor manufacturing?
I would say either Intel and perhaps GlobalFoundries which has that weird governance structure. I guess it’s officially owned by the UAE, but it has significant assets and a management in the US. Frankly, Samsung was first doing its cutting-edge processors for Apple in its Austin, Texas, fab. So they have a better track record than TSMC in the US. But in this case, it’s probably better for the US to use an American or quasi-American firm if our interest is to build up capacity in these cutting-edge nodes. This seems to be the tenor of these two bills that are in Congress, but in my opinion, they aren't large enough. Semiconductors is one of those industries where you are either going to put enough money in or not. I think the Europeans are doing it the right way. Their plan to invest 35 billion € in the sector seems like a more serious effort to me.