China has been in the center of the news this month, both because of the U.S. semiconductor ban and also the once-every-five-year Communist Party Congress that saw Xi Jinping consolidate power to an even greater extent than expected.
To that end, I am excited to bring you multiple perspectives on all of this news. The first perspective comes from Dan Wang, one of the pre-eminent experts on Chinese technology, and a frequent Stratechery guest. Wang is the technology analyst at Gavekal Dragonomics, and maintains his own site, including his essential annual letters, at DanWang.co. We talk extensively about the Chia chip ban, and touch on the Party Congress; what I think is especially interesting, particularly for those of you who have listened to our previous discussions, is Wang’s perspective on how China, and his view of China, has changed over the years.
As for the next perspective, it’s a bit of a surprise: I’ll email you tomorrow with more details.
To listen to this interview as a podcast, click the link at the top of this email to add Stratechery to your podcast player.
On to the interview:
An Interview with Dan Wang About the China Chip Ban
This interview is lightly edited for clarity.
The Chip Ban Surprise | China’s Current Position | The Trailing Edge and AI | Chip Ban Reactions | Gauging the Actual Impact | Apple and Decoupling | Xi’s Consolidation | Taiwan’s Risk and China’s Future
The Chip Ban Surprise
Dan Wang, it is good to have you back. I think every time you come on I say this is the perfect time to have you here. The last time was in the middle of the Shanghai lockdown, but now with not only the Party Congress but also this recent news around chips. I cannot imagine a better time to have you here. So thank you for being here once again.
DW: Ben, it seems like every time I come onto your show there’s something near catastrophic that happens to China. This is really strange, isn’t it?
Yeah, maybe it’s a bad omen! Let’s start with chip ban. What’s your level of surprise? Let’s just start there. Did you see this coming or are you sort of blindsided by the extent of these new regulations?
DW: One can always tell a story that these chip restrictions were signaled somehow somewhere by someone. For example, the director of the NSC when it comes to China tech issues, Tarun Chhabra, has been writing about this when he was previously at Georgetown and at Brookings, that his deputy also has written about this when he was at Georgetown. But I think the specific timing of when this dropped seemed to have been driven a little bit more by the White House over the past few weeks. And I think it was broadly a bit of a surprise to most of the industry of the scope as well as the specific timing of the week that it actually dropped on.
Was it a surprise to you, not just the industry?
DW: It was a surprise to me, Ben. I didn’t expect the scope and the controls to be so radical, as well as the specific timing of it happening so close to the Party Congress in Beijing.
Yeah, for the record, I agree, especially when you get into the restrictions on persons, there seemed to be no sort of stone left unturned in some respects. And yeah, to do it right before the Party Congress, I mean, it doesn’t exactly speak of a real leniency or attempt for any diplomacy. It seemed set to maximally embarrass Xi. I mean, maybe I’m just too close to China and so I’m seeing it from that perspective. I presume that’s how it would be interpreted in China. Do you think that was the intention?
DW: I think you are a little bit too close to China, Ben, sitting right there over in Taipei, maybe a little bit too close. Maybe we should talk a little bit about what your exit strategy is!
Yeah, we will get to that later.
DW: I think the idea of the optics was probably not the foremost thing on the mind of the administration, that they weren’t really trying to embarrass President Xi Jinping on the eve of his big third coronation, that it wasn’t so much about hitting China when it’s down and has its own quiet period of figuring out internal processes of personnel and the future. I think a lot more of this is driven by White House considerations as well as these broader structural trends: the U.S. government thinks that Beijing has become a much more unfriendly power, even more so than during the Trump years, after President Xi aligned himself so closely with President Putin in February with a no limits partnership.
The view of Washington D.C. in China seems to be that they are going to be very sophisticated technologically and outproduce a whole bunch of stuff and then also possibly threaten the American way of life and everything else, and that these things need to be controlled to a much greater extent. And possibly, I wonder if the midterm elections was something driving here. I tend to think that a lot of these were bureaucratic interagency processes rather than some big well planned move that was really meant to send some sort of message to the Chinese right before the Party Congress.
Yeah, this bit about the bureaucratic machinery moving, I think is an important point. This is something that came up. You were just recently on Odd Lots, but just this week they featured Chris Miller who just came out with the book, Chip War. I thought the most striking part of the conversation was his take that the reason why you see this not only continuity but strengthening in anti-China sentiment relative to the Trump administration is because this is actually driven much more by the bureaucracy than it is necessarily at the political level, I agree that this is maybe the answer here. We’re seeing not just alignment between the Republican party and the Democratic party, but we’re also seeing alignment between the bureaucracy and the politicians. And that’s almost like two parties and when they snap in place then stuff really does happen and it happens pretty quickly.
DW: I do have to give a plug, first of all, to Chris Miller’s book Chip War, which is selling like hotcakes right now. I have to be very friendly towards it because I blurbed it as outstanding, which you can see in the pages. So I’m glad that folks are picking it up and that it is doing as well as it is.
I agree that the permanent staff at various U.S. agencies, especially at the Department of Commerce, is getting a little bit more closely aligned with the higher level political staff, which since the Trump administration have been trying to crack the whip to tell the civil servants to move a little bit faster on implementing a lot of different controls. But I think there is still a little bit of an institutional culture at the Department of Commerce, as well as other agencies that are not too eager to rock the boat. That for the most part, they have successfully resisted a lot of the admonitions from the political staff to create lists of foundational and emerging technologies that designate a much larger swath of Chinese firms onto various U.S. blacklists, because for the most part, the professional staff are still very keenly aware of what is going on in industry. They know that the U.S. Department of Commerce is obliged by law to consider foreign availability.
I also think U.S. industry has done a pretty good job of telling the Department of Commerce that, “Well, look, memory chips aren’t unique to us alone. The South Koreans do quite a lot of that. Lithography isn’t unique to America, that Dutch in fact do it much better.” And then furthermore, the Department of Commerce, specifically the Bureau of Industry and Security, which is the implementer of all of these controls, is very well aware of previous examples where U.S. controls did not create success in broad U.S. initiatives. They remember that the efforts to control space and satellite technologies to China from the U.S. in the mid 2000s, not only failed to prevent China from becoming a major space power in its own right, but could also be challenging U.S. export competition in third markets, because China has managed to figure out a lot of these industries on its own. So I think the institutional bias is against hurting industry too much, and I think that is a delicate dance that Department of Commerce is playing, but I think this stance isn’t quite over it yet.
China’s Current Position
There are a couple things that I want to peel apart about that answer, but let’s start with the Chinese, and their relative strength in the semiconductor stack. So I’ve talked about SMIC’s trajectory before. They’re very strong at 28 nanometer, they’re ramping 14 nanometer. They’ve demonstrated the ability to build seven nanometer, although probably not economically. Is it fair to assume that almost all this to date is being done with now-banned equipment as far as anything 16 nanometer and under? Are all these machines from ASML, from Lam Research, from Tokyo Electron, or how much of what they’ve achieved to date has been done using domestic equipment?
DW: SMIC’s processes are overwhelmingly built on Western equipment, and there are about five companies here that really matter. Three of them are American, KLA-Tencor, Applied Materials, and Lam Research. Another is Dutch ASML, and the final one is Japanese Tokyo Electron. China’s semiconductor production industry is still in its early stages — I would say just out of the starting gate. At best, some of them are quite a few years behind the U.S. as well as Dutch, Japanese competitors. At worst, some of them are plausibly two decades behind the U.S., Dutch, and Japanese competitors. So these do look like pretty substantive controls on SMIC’s ability to continue making progress beyond seven nanometers, but also threatens a lot of its capabilities at 14, 16 nanometers.
SMEE is the Chinese lithography alternative to ASML. Is lithography going to be, in your estimation, the hardest area to catch up on? I know they’ve made deep ultraviolet machines with immersion lithography. Do you have a sense about how far along it is? Are these just sort of direct rip-offs of ASML? How much progress have they made?
DW: Chinese firms are making progress, but at the very best, they are several years behind. And so it’s a little bit difficult to say in which areas Chinese firms are closest, because at the end of the day, you need every part of these equipment to work for you to run a production process. So even if China is closer in some areas and further apart in others, you’re really being held back by your weakest link.
I think it is still worth emphasizing how difficult a lot of these things is. SMIC is not the Chinese company regarded in the very best light of technology progress, that lithography is pretty difficult, that metrology equipment to do a lot of this testing is also immensely difficult. And China is not too far along in EDA tools — that is mostly a triopoly of Cadence, Synopsys, as well as Mentor Graphics — and there are these vast software libraries that the software firms have that prevent China from designing its own EDA.
So functionally, I would think that China is almost just out of the starting gate in pretty much all of these technologies. That even if they can figure out everything to make, let’s say a car work without being able to figure out how the tires work, then they are still really far behind. And so I think the domestic industry is really not yet any sort of an answer to Western equipment.
This is impossible to say, so I will grant that at the beginning. But just sort of off the cuff, what’s your estimation for when China can sort of “catch up”, wherein catch up is build, say, a five nanometer chip, that’s sort of like the level where TSMC needed EUV to do critical layers. Has the timeline of that estimate shifted from where you thought it was like three or four weeks ago, in part because these restrictions are getting so much more into the equipment level and not just the foundry level?
DW: The restrictions from the U.S. now are going to roll back China’s progress to something like 2015 capabilities. So in 2022, China is very plausibly going to be at around 2015 levels of technology. And so when we consider when it can progress from 2015 levels of technology up into something like five nanometers, the first thing to recognize that they’ve been really significantly rolled back. So it’s very easy to gesture towards the future without giving any concrete sense of timelines. But I think that within the next, let’s say 5 to 10 years, China is going to be really, really substantially behind on the domestic manufacturer of advanced chips, in that five nanometers even is going to be quite out of reach for at least the next 5 to 10 years.
I mean, is it possible they just never catch up? You’ve made the point, which I echoed in my article this week that look, it’s a lot easier to walk down a path that you know exists, even if that path is very hairy and there are branches everywhere and there are pitfalls and all these sorts of things, you know it’s possible. That’s a lot different than having to figure out the path in the first place. If you go back through the history of EUV, I mean, everyone forgets about the whole 157-millimeter consortium that was sure that was actually the way to go, and ASML went it alone and figured out the EUV thing. But now you know that EUV is the right path.
Despite that fact, are there so many components, so many things that needed to be done to get this right, and there’s some aspect of you need chips to make chips that we’re actually looking to a decades-long set back or is that unrealistic?
DW: Well, Ben, I read a really good piece by my very favorite technology analyst, and that person is Ben Thompson of Stratechery who wrote a three-step program for why China is plausibly going to pursue a lot of good things.
I just admitted that I borrowed some of it from you, so you don’t need to butter me up!
DW: The first is that the Chinese companies are only walking down an established path, that it is a hard road, but there is a road, and they only have to follow a lot of the roads that American companies have blazed in the past. There has been a lot of vast technology sharing from U.S. companies to Chinese companies that have allowed them to progress to seven nanometers up to this point. And now, China is willing to throw gobs of money as well as infinite motivation to figure out all of these things. And as Ben Thompson laid out, I think these are still pretty good reasons to believe that they are able to figure out quite a lot of these things. Most crucial is that only having to reinvent a lot of existing wheels is an order of magnitude simpler than de novo innovation, that we know that a Micron DRAM chip can exist because one of them does exist. And I still think that the task for Micron or ASML when it comes to EUV, is quite a lot more challenging than what the Chinese have to do.
Now if I wanted to portray the administration’s action in the very best light, I think one can try to make an argument based on the hope or the expectation that China is going to have a really hard time making a lot of technologies, at least in the very near term. Right now in 2022, the mood in China is pretty dreadful, I would say. Between the tech crackdown of last year, the really sluggish economy of this year, between never-ending COVID restrictions, a lot of China’s best and brightest are, I think, quite a lot less motivated to figure out a lot of these technology problems that are pretty strategic for China. If there is just a persistently weak economy, if these lockdowns never really end, one can imagine that quite a lot of China’s best and brightest try to go overseas. We’re seeing some anecdotes of this trend. We’re seeing some of this data that many more Chinese are moving to Singapore in particular.
So it is also possible that if the US is targeting pretty much every segment of the supply chain, everything from the end users, companies like Huawei when it comes to mobile phones as well as 5G stations, if they’re targeting the foundries as well as the memory chip makers between SMIC as well as YMTC, if they are somehow targeting the semiconductor production equipment as well, which is one of the other facts of these rules, then it is possible that a lot of the Chinese efforts are stuck in the mud because every part of the supply chain has broken down. But I would still say that this is mostly a hope that Chinese firms are down and stay down rather than an expectation that they cannot somehow get back on their feet again, because they are now just really, really intently motivated to reinvent a bunch of existing wheels.
Yeah, I think that’s a good articulation of the administration’s case. I think the other thing I would add is, to the extent that AI is a driver, and that we are at an inflection point, there is a potential argument to be made that actually it’s not just that there’s going to be advancements from TSMC in rolling out three nanometer this fall, not just advancements in chips, but also overall systems and systems capabilities that is going to accelerate and make that gap even larger and perhaps produce positive feedback loops. It’s sort of an inflection point if you’re going to move, to move now.
The Trailing Edge and AI
But I would agree with your case, broadly. In the long run, this technology is so obviously critical. China is filled with lots of smart people, and to your point, the path is known. And I think it’s easy to imagine some point in the future, the US tech industry or the silicon industry and the Western silicon industry broadly now, has a new competitor selling probably lower cost just as good or capable equipment and has competition in a way they never would have absent this, because the economics just drove towards ASML having monopoly, drove towards TSMC having a monopoly. And that’s not going to be the case anymore. So hey, maybe the antitrust people, the competitive folks should be cheering this move.
DW: The other thing to note about these moves is that a lot of the thinking about Semiconductors is that if Moore’s Law is not quite dead, then at least it is dragging its feet, possibly on its last legs, and that the international technical consensus roadmap for semiconductors terminates in 2030. So past that point, industry has a little bit of a blind spot on what comes next. Everyone recognizes that a lot of these new fabs from TSMC, the leading edge cost about $20 billions each to build, and that at present most of the use cases are iPhone processors as well as cryptocurrency mining. And that if in the future, progress mostly slows down at the high end, then it is only a matter of time for the laggers to broadly catch up to the technological frontier.
Now of course, I think it’s almost a meaningless statement to say that semiconductors will not advance. Of course they will advance and they will advance in different directions, but if very broadly speaking we don’t need to continue pursuing all of these things, then perhaps it is not the biggest deal in the world that China is stuck out of the highest end.
It’s also stuck at 14 nanometer, which is the most cost efficient node probably in the long run anyway. So there’s some aspect where it’s not the worst place to be stuck if you had to be stuck somewhere, and the US can’t actually grow its lead that large because they will hit a wall.
DW: 14 and 28 are really good levels to occupy: a lot of chips are perfectly good at that level. The US administration hasn’t shown a lot of appetite to tackle China on 28. And so if China is dominant in 14 and 28 nanometers, and all the most advanced chips are being produced in the US, it hardly looks like checkmate that China has a very broad base of chip production, and both the west and China will still be pretty dependent on each other for chips.
Right. That was the point I raised in my article. The US, in part because of the way Intel and AMD developed over time, doesn’t really have any trailing edge capacity at all. And the west has some, but it pales, Taiwan and China absolutely dominate 28 to 45 and sub 45. Given that, to what extent is there actually an increase in risk in this sort of game? Because it turns out, as we saw during the pandemic, those low end chips are pretty important, particularly to a very large consumer economy. If TSMC were taken off the table, the US suddenly has this massive dependency on China. Yes, theoretically the US could make a bunch of low end fabs, but those make absolutely zero economic sense given that number one, TSMC has a bunch, number two, China has a motivation to build more because by building more, they get to learn. And that’s their goal right now. The US has no motivation, no economic driver to build up the trailing edge. Is that a risk for the US?
DW: That certainly seems like where things are going. And so the other question I would have about US strategy is to what extent is this pursuit of the really, really high end all that important? And so here’s where I want to ask you and for your thoughts, Ben, to what extent is this three nanometer process all that important? Right now, there aren’t all that many great use cases, and even though the US administration has taken this view that any GPUs in China is per se, a national security threat, I still am not quite sure to what extent these large language models, all of these AI models are going to herald in this bright wonderful future. A lot of AI presently is really good at generating poems about the beauty of Taipei in Elizabethan verse and also creating wonderful images of the Taipei sunset that are more and more compelling in other ways. But is this really such a big national security threat? Especially since I understand that most military technologies are still much more dependent on the trailing edge as a lot of consumer technologies are. Is this such a big thing that China needs to be deprived from GPUs? What’s your view here?
I think that’s a really good point. Because I think you hear AI, you think Terminator, your mind immediately goes to military applications. But the reality of AI, where AI’s going to make a huge difference, is probably going to be in lots of areas of boring, rote white-collar work where you are just filling stuff out. Images is obviously an easy one. Sometimes you just need to generate an image for an article and you could go get a stock photo or you could get an illustration or you could just generate something, and you could definitely see the language models progressing in that regard from text.
Now obviously a concern in this area, a reason why images are an obvious place to start, is errors, and errors are themselves art. That’s sort of interesting. Errors in paperwork can be overcome, can be checked. Errors in missile systems are a bit of a bigger problem. I think the likelihood that these applications remain fairly deterministic are fairly high. Now I think there are areas around cybersecurity where AI is probably going to have a pretty big impact. But I think it’s fair to question to what extent AI is a real driver, at least in the short term, of military applications.
This gets to your point about GPUs, because that’s the one area where we every additional bit of performance and every additional bit of efficiency does pay off. Because the way to build these language models, the way to leverage them, there is not just a quality and a speed aspect, but a quantity aspect. And the more you can do, the better. And so on one hand, if AI is the motivation, then the GPU focus of course does make sense and that will be a real limitation. But I think you’re fair to question just what difference it’s going to really make.
Chip Ban Reactions
Also, as long as you have put me on the spot, it really bugs me that Secretary Blinken’s has come out talking about these vague increased threats to Taiwan. Number one, has hasn’t presented any evidence, and number two, any of these chip applications and rules are not going to have any impact on the short term. To the extent they have an impact, it’s going to be a few years down the line. So that does make the whole thing sort of smell a little bit, like why are you suddenly talking about this now without evidence, it seems to be tied to justifying this act, but this stated warning, the timeline doesn’t match the actions that you just undertook.
DW: Let’s talk a little bit about the reactions to these regulations because I think the reactions have really caught on in the popular press in the way that previous export control actions, never really have. Now, I think by any measure, these rules are a major escalation. To deny advanced chips to China almost at large, to deny advanced chip making equipment to China, controls on US persons, controls on particular choke points to technologies. I think by any measure, these are very big moves.
But on the other hand, these have been framed from the Department of Commerce as fairly straightforward arms control actions that should not necessarily trigger so many major headlines. Now what I think of is a columnist at the FT, Edward Luce, writing that Joe Biden has fired the first shot of a full blown economic war. I saw headline from the New York Times that Biden has just clobbered China’s chip industry. And there are all sorts of these apocalyptic catastrophist views on how badly China’s chip industry will be affected. And first of all, I tend to think that it is a little bit too soon to tell about how these rules will actually be enforced.
But to get to your point more directly, I think it is a little bit unusual to think that such a major policy change was not accompanied by any real statement from the US president, Joe Biden.
Yeah, that’s a great point.
DW: There were some statements from Jake Sullivan as well as Tarun Chhabra in these think tank meetings that we are now picking up and examining very closely. But for an order of this magnitude to be just simply advanced through the bureaucracy without any statement from the Biden administration, from the president himself, makes me think that, well, perhaps the messaging here does not really correspond with the actual impact of a lot of these rules.
I think that’s exactly right. My read on the lack of messaging is that there seems to be a sense that, “Look, hey, we’re just trying to restrict AI, we want all this stuff to go on.” You can definitely see that in terms of industry, these US equipment manufacturers are getting a third of their revenue from China, and they’re like, “Look, no, we don’t want to risk that, please keep selling. We’re just focused on these sort few things.” That was the thing that struck me right off the bat is that the messaging I’m getting from the administration seems wildly disconnected to the way, number one, how important this seems to me, and number two, to the way I anticipate this is going to be interpreted and received in China. I feel like that comment about a declaration of economic war, is there any reason to think that China would not interpret it this way?
DW: Well, I tend to still think that Beijing has displayed pretty remarkable strategic forbearance over about, let’s say five years of trade warring when the US government has destroyed a national champion, Huawei, and have piled insult upon insult to a lot of Chinese interests. And for the most part, Beijing has been pretty patient about retaliating in kind against US actions. And so here is where I think it is maybe good for us to think a little bit more about what might not be the very worst case scenario from these rules.
Gauging the Actual Impact
To that end, is it possible that we’re over-reacting to these rules that seem so stark, and that the impact might be different than what it seems at first glance?
DW: Well, Ben, I’ve been looking at a lot of these export control actions over the last five years, and I think it is usually a good rule of thumb that there is a little bit less that meets the eye in a lot of regulations. And that regulations around export controls tend to be porous and leaky often, for a couple of different reasons. The first thing to note is that these BIS regulations take the form not as strict prohibitions, but as licensed requirements. And so-
Right. With the presumption of denial, which is sort of new in this case-
DW: There is a presumption of denial. But BIS has always attached presumptions of denial to previous license requirements around Huawei as well as SMIC. And the data that we have from BIS is that even items subject to the presumption of denial have previously received around 90% approval rates in past regulatory actions.
Yes. I was going to ask you this, so you’re on pace. I’m curious if you think that’s going to shift or if that’s going to continue.
DW: So that is a big trend out there, and even for companies that the US government regards as uniquely bad actors, like Huawei and SMIC, even these companies have received through their suppliers, export licenses of around $100 billion, according to the Wall Street Journal.
The critical question then is whether the enforcement posture will shift this time. And the answer is yes, that the Department of Commerce has removed several of these license exceptions, around these presumptions of denial. But I also think that for a rule as complex as this, that it is going to take a little bit of time to see what the actual enforcement posture is, that a lot of these companies will apply for licenses. And if the present trends are at all correspond with a past trends, that there will still be some licenses approved.
I tend to think that companies are very in a rush to over-comply and make these big statements saying, “We are cutting off relationships with Chinese company X and Chinese customer Y.” But often, when they have really understood and digested these rules, when they do figure out that some business cooperation is still possible, they tend not to release big statements, or very easily quotable articles about how they are resuming business.
Third, and most fundamentally, these rules from Bureau of Industry and Security in October, are to a large extent, extra-territorial. There is foreign direct product role application of a lot of these different types of equipment, but we still have the CFO of ASML saying, “Well perhaps the impact on ASML’s DUV sales to China are still going to be limited, because we don’t have a lot of US technology in our tooling.” These rules are going to prompt a lot of companies to consider whether they should design out many more of their US components, such that their products are not subject to jurisdiction when the government limits them from selling to their first, second, or third largest market, which China is for pretty much all firms. And so these rules may not stop companies everywhere from exporting to China, whether these are European companies, or American companies with facilities in Malaysia, Singapore, or Israel. And it is also a possibility that Chinese firms will try to figure out a lot of these technologies over the longer run. And so I think the impact of these regulations do on its face, look really bad, but I think there is still some reason to be cautious that they will not be quite so apocalyptic as the headlines suggest.
Yeah, I agree with that. I think that the point about how these have actually been enforced over time is an important one. I guess the question, though, is moving forward, this does seem to be a clear attempt to shift the bureaucracy. There is, for example, the unknown entity or unknown compliance, I can’t remember what it is…
DW: Unverified user list.
Unverified user list, yes. We don’t know if they’re working with the military and if we don’t know within 60 days, then they’re going on the list, so the defaults are changing in some respects. And do you think that’s going to continue, number one. And then number two, do you see other top level restrictions going forward? I know you’ve talked about the investment side. Is that sort of the other shoe to drop?
DW: It’s possible that there will be some greater controls on AI, and I’m just going off the Bloomberg reporting here. There was a piece last week from Bloomberg that there may be more controls coming on AI technologies in particular. I think the other big shoe to drop is an outbound investment review mechanism, which some folks in DC refer to as a reverse CFIUS bill.
CFIUS screens for inbound investment into the United States with an interagency panel led by the Department of the Treasury to make sure that investments into the United States do not endanger US national security. And there has been an increasing view in the US government that sometimes US investments outbound into particular countries of concern, namely China, may also threaten US national security. That the likes of companies like Apple, or Intel, or financials like BlackRock and Sequoia, really shouldn’t be investing so much, either financially in Chinese startups around semiconductors, as financials may be doing, or building new fabs and building new factory sites in China on the corporate direct investment side. These are the things that could threaten US national security.
My best sense of this is that the Biden administration will release an executive order limiting these types of investments, or at least establishing a transparency mechanism so that the US government is always aware of what investments are taking place in China. And that might happen in the next few months. And it is also possible to me, either before this year, before the National Defense Authorization Act, that the US Congress also manages to legislate something. If not now, then after the midterms when it looks pretty likely that Republicans will capture both houses at this point.
Apple and Decoupling
You mentioned Apple. YMTC, the state-owned Chinese memory maker, built up 128-layer NAND capabilities. This is reported, it was going to be in the iPhone. Originally, I think it was going to be in the iPhone more broadly, then word got out. People were mad and Apple was like, “Oh no, only for the Chinese market.” What I’ve heard — this is just hearsay, not a report — is that Apple did a lot of work with YMTC to develop this capability, which makes a lot of sense because having cheaper memory suppliers is very much in Apple’s interests.
To that end, number one, how much does this ban impact Apple directly? Number two, how at risk is Apple? They’ve always been an obvious candidate for potential retaliation, and the counter has been, “Look, they employ a whole lot of Chinese.” To your point about Chinese forbearance, Apple has been a pretty effective advocate for Chinese interests in the US. But then number three, is their view of China shifting? And again, it’s not just the export controls. There was also the Shanghai lockdown. In fact, they basically couldn’t make a Mac for a month. Where do you see them in all this?
DW: I think Apple could be considered emblematic of a lot of major US multinationals operating in China. If you really had to push me, I would still think that US actions have not triggered a break in Chinese strategic patience. Beijing still has forbearance in part, because it has a lot of need for continuing foreign investment into China this year when the economy is pretty dreadful in 2022, as a result of the property downturn, as well as the COVID zero policies.
So I think Beijing is still pretty intent to pursue its dual circulation strategy, which on the one hand is to make China more self-sufficient from the rest of the world, especially in semiconductors, as well as other strategic technologies. And at the same time, it is to make the rest of the world more dependent on China, which means to continue accessing its market, or procuring Chinese technologies. For example, solar panels, which are now very substantially made in China.
I think that Beijing is not yet really ready for all multinationals to leave, because it still wants their technology, still wants a lot of their employment. And it would much rather that the likes of Apple, or Tesla, or BlackRock, or Sequoia, are advocating in DC for greater economic engagement, rather than thinking that they have absolutely nothing left to lose in China anymore and therefore DC should just do its thing.
And I think it can’t be overstated the degree to which the Chinese economy is just so structured fundamentally different than the US economy, in that exports are still just a massive part of it, and their domestic market is not nearly enough to absorb all of their capacity. And of course the US is the inverse of that. That’s why the relationship is such a mess. But to your point, I think this bit about the China needs this production for jobs and for income, and they need the markets to actually push this stuff into. And so it’s so easy to lock on to the worst possible outcome without considering how deeply, structurally intermingled the west and China still is.
DW: It’s not just a matter of intermingling, it’s also the case that I think Beijing has decided that its retaliation strategy is much more subtle. It is to hug all of these multinationals even closer, instead of to push them into the arms of the hawks. Nonetheless, I detect, as you asked earlier, that many multinationals are deeply souring on the Chinese market. That there is a lot more appetite to diversify their export bound production out of China.
That’s happened a lot this year for I think three broad reasons. The first is that, as you mentioned, the Shanghai lockdown was traumatizing for a lot of people, not just personally, but also for business continuity reasons that a lot of companies around Shanghai, which is one of the major industrial hubs of China, along with Guangdong Province, and the Shenzhen cluster, that Shanghai really couldn’t produce for about at least a month and possibly for two months of severely disrupted operations. And thus, China no longer looks like the most reliable place to manufacture a lot of different goods.
Which is a total sea change from the first year of COVID where it’s like, “Look, this is the one economy that could sort of get their crap together.” And now it’s amazing the degree to which that sentiment is done a complete 180.
DW: And also the second year when Delta slammed alternatives to China, namely Vietnam, as well as India, really, really hard. And there were even companies that moved production to these places that moved them back to China because they thought that, “Well, China’s still the best place to manufacture a lot.” So, they were taken to the cleaners twice.
The other reason that a lot of firms are souring on China is that Russia’s invasion of Ukraine is bringing up these uncomfortable parallels to Taiwan, and Beijing will have an impossible time answering their concerns. So far, I think most of these companies are trying to figure out contingency planning. Very few have actually acted on the Taiwan issue, because there is no real sense of timeline. Could it happen in the next two years? Could it happen in the next 10 years? Could it happen in the next 30 years? Well, so long as there is no certainty on that timeline, then perhaps they don’t need to move too much. They just have to have a plan.
The third big issue is that the Chinese market just doesn’t have that much growth this year. And I think it is a little bit more consensus now that China’s economic prospects over the next decade is not going to improve very considerably, because of a commitment to zero COVID, as well as a lot of discomfort around the reign of Xi Jinping…
And then you have demographics on top of that.
DW: Demographics on top of that, as well as the institutional decay around Chinese governance over the last decade, which makes the place look like it is much more focused on ideology and national security rather than on economic growth. And so the price of participating in the Chinese market keeps going up, because there are just too many unique problems and reputational risks of dealing in this market. At the same time, the future prospects keep going down. And so for this reason, I think the multinationals are going to be slowly and steadily shifting a lot of their export bound production out of China. But I think the key point to stress here is that that process will be a trickle. That process will be fairly slow.
Right. It’ll be a slow down in investment, not necessarily, “We’re just pulling everything out tomorrow.”
DW: It will be a slow down in investment and I think there will be some active divestment, but the active divestment only has to take into account the export bound production. So for a lot of companies, let’s say something like Apple, let’s say around 90% of their supply chain is in China — those are some of the popular estimates out there. And around 20, 30% of its revenue is made in China, well, you really only have to diversify the export bound production around the 60, 70%. It doesn’t really make sense to move all your supply chains to Vietnam and India and then re-import back into China. So, I think there is still some floor on how much companies are willing to move.
I’m glad we spent all the time on the chips and the tech stuff, because obviously given this subject of my publication, it couldn’t be more appropriate. And so thank you for that.
Let’s just do a quick download on the Party Congress, which just concluded. I think the easiest summary is that Xi ran the table, so to speak. I mean, not only does he have another term, but the Standing Committee is all Xi loyalists. And also, it’s not just a change in norms for Xi to have a third term, and a change in the rules, but also some of the norms around some of those appointments has changed as well, so that he could put his loyalist there. I think both of us have always cautioned that it’s wrong to reduce China’s point of view to any one person. It’s a large country with fairly distributed power and political factions and all those sorts of pieces. But is it increasingly correct to just assume that one person is calling all the shots?
DW: I think it would be more correct than before to think that, look, what is really important here is to read the mind of Xi Jinping. And that it is a little bit less valuable to think about what the broader views of the bureaucracy are. He has promoted a lot of his close associates. One might critically call them a lot of his yes men into positions of power. And they didn’t get to where they are by a lot of independent thinking. Now it is possible that they are able to better speak truth to him, but the trend has not been that they got to where they are by giving him raw, unfiltered, independent thinking views. And it is much more of his government than before. I would definitely agree with that.
Is there any reason to be confident in Xi’s competence? I mean, what has he done that actually inspired any sort of faith broadly? I get there are reasons why China’s not opening up for COVID, for example. The medical system’s not great, there’s not much natural immunity. Anti-corruption is, of course, good, but those can also be looked at as ways to question his judgment and actually a play to consolidate power, which seems to be his greatest skill.
I mean, on one hand, you want to assume pragmatism on China, and you put the case forward that China’s been wise to not necessarily retaliate, and to hug multinationals closer. At the same time, to your point, that’s almost been completely undone by mistakes in other areas. I mean this is what makes me the most nervous. At the end of the day you could definitely criticize and talk about issues with the US’s approach, but if we still had Hu Jintao in power and not shuffled off the stage, figuratively and literally, it feels like none of this would be happening. I mean, is that pinning too much on one guy or is that your view as well?
DW: I think that would be pinning a little bit much on one guy. The consensus of the China Watchers is that the authoritarian turn began in the latter third of Hu Jintao’s reign, that the party felt itself getting really dissolute, and that it really needed to crack down on a lot of foreign influences.
And it’s fair to note that the South China Sea stuff started under Hu Jintao as well. This isn’t all Xi Jinping.
DW: It’s not all Xi Jinping. And that after the financial crisis, China decided, rightly or wrongly, I would say wrongly, that it doesn’t have too much to learn from the West anymore, because it gets into a lot of these crises.
I don’t think that there is much way that Xi becomes a radically different person in his third term. I think that would be a fantasy. And I think it is just much more likely that his control is continuous. But I think there is one thing to note that is possible about Xi, which shows that he is capable of strategic retrenchment, and these tactical moves roll back sometimes when it’s necessary. For example he received a lot of criticism around the actual handling of the operation of Belt and Road, he did tamp it down a little bit, and then tried to improve a lot of the processes. When entrepreneurs complained about access to credit in 2019, he convened a lot of them together, and tried, at least, to berate the banks into lending a little bit more.
And I think one thing to note about the Party Congress is not just the personnel, but in his report to the Communist Party delegates, he spoke much more on the development of the economy in this session. There was a very substantial session saying that development is the Communist Party’s central task, and that we must treat it much more seriously. So I don’t think that herald’s a really big strategic shift in the Communist Party’s actions, but I think it does show that there is a recognition that the economy is pretty important for China, and it is pretty important for the Communist Party’s governance. That if they do take this seriously, this soft goal of trying to surpass the US economically, then they at least need to grow quite a lot faster, and at least at a higher rate than the US. So I think I wouldn’t totally write off these concerns that the economy doesn’t matter at all to China anymore. Xi still cares.
Taiwan’s Risk and China’s Future
Two final questions. Number one is what I’ve asked you I think every time, what’s your view of the Taiwan risk today?
DW: Ben, you’re sitting in Taiwan. I think your views are much more important here.
No, I’m inherently biased. That’s the problem. So I need an outsider perspective.
DW: I guess, I agree with the conclusion of your piece, that the risk of this particular action will have gone up slightly, but I don’t think that it is going to be a very big risk estimation upwards. And that it is also plausible to me that Beijing decides to retaliate against the US in a different domain, namely the Russian-Ukrainian battlefield. This is pure speculation, but it could be possible that China decides that its support to Russia, which has been mostly contained to rhetorical and diplomatic means, becomes a little bit more material. That it decides it has a little bit less to lose in this sense.
But I am fairly skeptical of the Taiwan thing happening at the moment because you know as well as me, and much better, Ben, that this would be the greatest amphibious invasion ever attempted. That for any sort of action to begin, there will be a lead up of a year’s worth of military assets, of redeploying military assets onto the coast of Fujian, which faces Taiwan. And these things are all going to be very easily detectable. There will be no secret buildup of the forces. That’s probably not possible.
As an analogy, if COVID Zero ends in China, there will be months and months of public propaganda saying, “Well, maybe this virus isn’t so bad. It’s just the flu everyone,”; perhaps everything that the Republicans were saying about COVID in the early days will end up in the mouths of the communist party. There is a possibility that if China really wanted to open up, there will be a much higher pace of vaccinations as well. You will see it coming.
So I guess on the margin, this might increase the Taiwan risk, but I still don’t think that you need to get your butt out of there anytime soon. But what do you think about this, Ben?
I agree. I mean there is a revealed preference bit here where I’m not moving. On the other hand, I do have a somewhat newly furnished place in the US that I could go to!
I think the Russia-Ukraine point you made is a really, really good one. I wrote earlier this year in an article, Tech and War, that I viewed chips as a bargaining point in gaining Chinese cooperation vis-à-vis Russia. The implication of that is if you take away the chips, then that you’re also sort of freeing up China’s actions in that domain. And I think this point about that being a way to retaliate is a very compelling one.
At the same time, yeah, I think an invasion of Taiwan, in general, it’s going to be so difficult, and the fallout so large. The outcomes will be so catastrophic, not just militarily, but economically, and large for China, that I see it as unlikely. Is there going to be a continued ratcheting up of pressure? Likely, But I still think the relative likelihood is quite low. I’ve always been higher than most Americans, until they discovered Taiwan like two years ago and then they all went to 100. And I’m like, “Now I’m much lower than you even though I’m higher.” So I think we’re probably in the same page in that regard.
DW: That makes sense.
Last question. You’ve been on Stratechery multiple times now, and I’m kind of curious, as all this has unfolded, what, if any, base assumptions have changed for you? Whether that be about China, about the US, about the nature of technology. What has shifted as you’ve observed the way the last few years have unfolded?
DW: First of all, Ben, I’m looking for a medal. I survived Stratechery, what is it, five times now? So I do hope you can give me some sort of ribbon.
Your retaliation was interviewing me and flipping it around. So good job by you.
DW: My views on China have shifted fairly substantially this year after the COVID lockdowns, as well as basically this stasis in China that I observed. I think a lot of the local initiative that the communist party has cultivated over the last decade, a lot of that is gone. And this immensely talented organization, with a lot of bureaucratic resources, is spending quite a lot of its time trying to make sure that people do not go outside and have any productive or social gatherings in any big ways.
The continuous tightening by President Xi on, not just freedom of speech, but also freedom of conscience, is also really affecting a lot of Chinese, who are starting to move more of their physical assets, financial assets, or themselves, abroad. There are a lot more Chinese who are going to Singapore, if they’re among the elites, as well as younger people who are trying to find their way out, even if it is mostly through these pretty difficult and expensive masters programs in various different places because that is their main way of trying to procure a visa.
I think the longer term trajectory of China over the next decade is a lot less bright than before. The focus on ideology rather than growth is really gripping the Chinese state, and that it is now quite possible, and very plausible, that China does not manage to outgrow the US.
Now, the only caveat that I want to put here is that if we think back to the state of the world only 12 months ago, the US and China were in pretty different states. The US looked like it was flailing, Biden’s first year didn’t look very good, especially after the withdrawal from Afghanistan. And over in Beijing, the mood was pretty triumphalist. Beijing felt so confident, that it was willing to smash its most profitable tech companies, and also tackle these systematic risks in property. And these tables have completely shifted this year when the US looks really good after passage of IRA, as well as the CHIPS and Science Act, as well as reopening the economy. And China right now looks like it’s flat on its back with a fairly dreadful economy and never-ending COVID restrictions. So these tables might shift again in 12 to 14 months. I can’t really identify the catalyst. No one ever can. And so I suspect that this story hasn’t really ended here, that things aren’t at an equilibrium now with China being in a rather bad state. These things can shift again and these things may well shift again some time soon.
Dan Wang, it’s been great to have you. I thought this was a great conversation. Thanks for making the time. And I look forward to talking again soon.
DW: Okay, thanks very much Ben. It’s always a pleasure.
This Daily Update Interview is also available as a podcast. To receive it in your podcast player, visit Stratechery.
The Daily Update is intended for a single recipient, but occasional forwarding is totally fine! If you would like to order multiple subscriptions for your team with a group discount (minimum 5), please contact me directly.
Thanks for being a supporter, and have a great day!