Chip War: The Fight for the World's Most Critical Technology

foreign good evening Welcome to our International speaker series my name is Angela Zhang I'm the director of the Philip cage one Center for Chinese law today I'm going to invite the author Chris Miller to give a book talk on his super exciting new book the chip wall the fight for the world's most critical technology and you probably have read in the news this book has been a sensation since it came out um and and it's because it's on an extremely timely topic that is of great interest to our audience our our talk has attracted an unprecedented number of of um of audience today so before I give the floor to Chris let me give a very brief introduction of our speakers so we we have offered Chris Miller with us he is an associate professor of international history at the Fletcher School of Law and diplomacy and top at Tufts University um he is also um Jean uh Kirk Patrick visiting fellow at the American Enterprise Institute and the Eurasia director at the foreign policy Research Institute um his research focused on technology geopolitics economics International Affairs and Russia and I'm also super excited to introduce the discussion for our book talks tonight it's a particular he is a senior vice president for China and Technology policy Elite at ASG Paul is a leading expert in this area and I am so glad he can join us tonight he advised clients in technology financial services and other sectors as they navigate complex political and Regulatory matters in China and around the world although Paul looks very young he has a lot of experience um he was most recently founder practice hat and managing director of the geotechnology practice and Eurasia group and previously he spent more than 25 years in senior positions in the U.S government analyzing China's rise as a technology what power and Advising senior policy makers on the broad range attack related issues so I really cannot think of a better person to comment on the book today so without father a dude let me give the floor to Chris thank you very much Angela for the invitation and the introduction and thanks also uh to Paul for agreeing to join this conversation um share some thoughts with you on uh on my book chipboard but perhaps before I begin that a bit of background as to how it is that I came to write this book as Angela mentioned my background is in Russian studies and my previous three books have been on the history of Russian economics and politics which most of you might not think would be a natural background for writing a book on semiconductors in the role in international Affairs about five years ago when I started this project I was initially planning to write a book about missile technology uh during the Cold War uh and the the US and the Soviet missile race during the 1950s 60s and 70s and the more I learned about missile technology the more I came to realize that the interesting parts of developing missiles was not the rocket engines nor the Metallurgy involved or there's some complicated metallurgies you get more sophisticated missile types but the guidance computers inside of missiles were the most fascinating part in the development of advanced guidance computers was one of the greatest challenges that not only of the Cold War or the arms race but of all of Science and Engineering over the past Century or so and the guidance computers that were involved in are necessary for missile development in the 1950s and 60s not only were important for missiles themselves but it gave birth to the computing power that we take for granted today the first major order for semiconductors from the early 1960s was for the guidance computer and the Apollo spacecraft which guided the first mission to the moon the second major order for semiconductors after they were invented was for the guidance computer on the Minuteman 2 intercontinental ballistic missile which was designed to carry nuclear warheads in the United States through Targets in the Soviet Union so there's been a direct and deep interconnection between military demands for miniaturized computing power and the technology that's inside of all of our smartphones and PCs and although today we think about chips primarily in the context of consumer devices the reality is that there's been a deep interconnection between the production of military trade power the demands of militaries around the world and advances in Computing and what I'd like to suggest to you and this is what I realized I began researching this book chipwar is that in fact military concerns defense concerns security issues have been Central to the chip industry over the past 60 years obviously not the only Factor business models economic issues consumer demands have been hugely important as well but it struck me when researching this book that there was a lot more discussion of the economic imperatives of the consumer demands than there was of the interrelationship between uh military concerns and the electronics Industry and the Computing industry uh as well and so the main argument of Chip War is in fact that there are multiple Wars or battles going out of the industry battles between companies not battles uh between uh countries but also battles between militaries for the ability to secure access to the most Advanced Technologies what I'd like to suggest to start is that although today we associate tips with with things like smartphones or data centers and we rely on ships for almost everything in our daily lives from Cars to dishwashers to microwave ovens in fact as I mentioned the origins of chips is in missile guidance systems uh and you can't understand how the chip industry has developed without taking into account it's deep into relationship with the military especially the US military which has funded many of the key advances in chip technology we wouldn't have had a shrinking of computing power and a reduction of the demands for energy usage in computing power unless there had been such a need for smaller more power efficient computers inside of missiles and if you think back to what a computer looked like in the late 1940s a computer was the size of a room uh required vacuum tubes light bulb-like devices that turned on and off uh to function uh and and consumed an enormous amount of power uh and so the most advanced computer in 1945 for example the NEX computer at the University of Pennsylvania I had only a tiny fraction of the computing power in your smartphone and it would have been completely incapable of um of sending a missile anywhere because it was simply too big to put on a missile and so the key challenge over the past several decades has been miniaturizing computing power so that now we have Computing inside of our iPhones for example that is many times more powerful than computer that guided for example the Apollo spacecraft to the moon and that is largely driven by defense demands and also defense r d dollars which have shaped the direction of the industry today just like in the early Decades of the Cold War military systems are fundamentally reliant on computing power even relatively old defense systems like the javelin anti-tank missile uh pictured here which has been played a big role in the Russia Ukraine war over the past several months are fundamentally reliant on semiconductors uh each Javelin missile for example is over 200 semiconductors inside and this is an old technology first designed with several decades ago and newer Technologies are even more dependent on Advanced Computing if you look at how defense planners in the US or other countries Envision the future of warfare uh they Envision uh battlefields that are involving more Computing uh distributed across different systems more secure Communications more need for intelligence and surveillance more need for processing power distributed across the battlefield and all of that that means computing power all of that requires semiconductors whether it's semiconductors that are high power processors semiconductors that are sophisticated memory chips or complicated digital signals processing semiconductors the wars of diffuser will be even more reliant on semiconductors than today and this is something that certainly in the US defense planners are thinking about but also in other countries in China for example there tremendous amounts of discussion about the application of artificial intelligence to military systems or autonomy to military systems and you can't have an autonomous military system unless you've got a ton of processing power very large amounts of memory and a lot of signals processing in that system and all of that requires Advanced semiconductors to work so wherever you look around the world you'll find that militaries are planning and in many cases designing in some cases even rolling out systems that are far more dependent on semiconductors than ever before because the advanced Computing on which they will rely is only possible thanks to semiconductors so as a result of this the uh the the struggle uh for um the the most prominent position in the semiconductor industry is not uh only driven by firms uh it's not only driven by products it's also increasingly driven by uh security and Military demands and this is a a shift compared to the past couple of decades although in some ways it's a return uh to the origins of the industry since the 1990s or so the chip industry has been largely able to ignore security and geopolitical questions uh in large part because most of the industry was focused either in the United States or in a small number of countries that the US uh felt were close Partners or Allies South Korea for example uh Taiwan Japan and the Netherlands um but as tensions between the US and China have grown uh both uh Washington and Beijing have placed a lot more emphasis on trying to increase their influence in semi-factor Supply chains and domesticate where possible some of the technology and the production capacity that they believe they need so in the United States there's been a couple of major moves in the past several months first uh Congress has passed the chips and science act which will devote several tens of billions of dollars to incentives for manufacturing uh Advanced chips primarily in the United States as well as a lot of new money towards r d in the chip sector in addition DARPA the defense Department's Advanced uh projects research agency several years ago launched an Electronics Resurgence initiative design again just like in the early days of the chip industry to uh put more money into advanced chip making and ship design Technologies specifically to enable military systems and then finally as I think I'll probably discuss uh later on the Commerce department has rolled out a series of pretty uh dramatic controls on the export of ship making equipment and certain types of chips uh to China so there's a a lot of movement in the US when it comes to the chip industry largely driven by security and defense concerns that's the big shift that's happened over the past couple of years and similarly in Beijing there's more concern than ever before about semiconductor supply chain so China spends as much money importing semiconductors each year than it does importing oil uh and policy makers in Beijing realize this is a a vulnerability and I've been trying for some time to address it since around 2014 the Chinese government has flagged this as a major policy priority and it's launched a series of different industrial policy initiatives designed to put substantial sums of money into the chip industry and this takes place both at the national level uh where there's the the big fund for integrated circuit development but also at the provincial and local level where many provinces and local governments also have various initiatives to support uh the semicular industry and so estimates of the size of the amount of money flowing from government and government backed sources uh into the semiconductor industry in China uh are are dramatic one of the largest industrial policy programs according to Barry Naughton one of the experts on the subject in in the recent history of uh the Chinese government so it's a clear priority and China around this issue and it's easy to understand why because today uh chip making in China is uh largely reliant on the importation of software of uh machine tools and also of uh raw materials like refined chemicals from countries uh that uh to some extent at least our geopolitical Rivals of China that includes the United States Japan Korea and Taiwan um as well and so as these countries have uh to varying degrees tightened restrictions on what types of chips they're able or willing sell to China and what types of Chip making equipment they're able to willing to sell to China China has faced uh a greater need to find alternative and ideally domestic sources of supply and so the a huge um wave of subsidies in the chinaship industry is partly uh driven by this uh this desire to have domestic sources uh in case uh International sources of tools and equipment get cut off in some cases they already have been so both from Beijing and from Washington there's been a big shift in terms of the uh the concerns about security taking a center stage and some other Supply chains rather than concerns about economic efficiency because of course none of the government programs uh to intervene in Supply chains are economically efficient uh today the chip industry uh uh made possible its immense ability to produce lots of computing power at low prices uh thanks in those small part to a very complex International supply chain and so long as governments are trying to slowly change the shape of that supply chain or pick it apart that will create more economic efficiency higher costs as well but governments are deciding that that's a price they're willing to pay as they focus more on security than they have in the past and it's worth noting I think that although the US and China are in some ways uh the two players that have seen the biggest shift in their approach towards semiconductor Supply chains they're not alone uh the Japanese government for example impose a set of export controls on the supply of certain chemicals to South Korea several years ago Russia earlier this year cut off the export of certain noble gases that are used in semiconductor production with the aim of causing supply chain issues so you can look at many countries that have a role in the semiconductor supply chain and almost all of them are emphasizing security concerns far more than they would have several years ago so how does this play out in practice well if you look at the semiconductor supply chain today what you'll find is that the United States is still the most important player when it comes to its position in the value chain across the different steps of the semiconductor production process in designing chips and the production of software tools that are needed to design chips and the production of the tools that are used to manufacture chips um all of these are places the United States maintains a major role other big players are South Korea Japan and Taiwan as well and because the chip industry uh is defined by a series of thought points where a small number of companies in some cases only a single company uh play a critical role in the supply chain this is an industry in which governments have found that it's comparatively easy to cut off their adversaries because alternative sources of Supply are limited and costly so for example um when you look at the production of machine tools needed to make chips uh the Netherlands company asml has a 100 market share on the production of uh extreme ultraviolet lithography machines that which is simply impossible to make an advanced chip so with choke points like that where there's uh market shares of 100 and certain critical tools it's perhaps not surprising that we're seeing uh governments try to take advantage of those choke points and when you look at design uh software tools chemicals fabrication tools what you'll find is that in almost every part of the supply chain there are choke points where a handful of firms or in some cases a single firm uh play really fundamental roles in the market and so long as these choke points exist governments will feel um some compulsion to use them and we're certainly seeing that play out in a much bigger way in recent years uh however the Chinese government is of course spending very heavily as I mentioned previously with the aim of trying to reduce its Reliance on some of these choke points uh whether it comes to chemicals fabrication tools software tools or chip design themselves and by any estimate China's role in the chip supply chain is going to grow over the coming decade though it's likely to be the case that the expansion of China's role is primarily in lower technology chips rather than the most um Cutting Edge but in particular in the process of making chips manufacturing ships uh most estimates have China's market share growing pretty substantially over the coming decades but the the key question I think is to what extent uh the technological tensions that we've seen over the past several years uh intensify well which they could well uh do we've seen Peck Dimensions over uh Huawei for example um over um the export of fabrication tools uh from the US uh to China and and Export controls that not only prevent the transfer of U.S technology to China but also not prevent uh other firms like taiwan's tsmc or South Korea Samsung from using U.S produced tools to uh create ships that are then sent to China so if there's an extra territorial facet uh to these export controls as well and the impact is already becoming visible on the semantic supply chain um the idea that any country can on its own produce the most advanced ships I think Still Remains a fantasy but the reality is Supply chains are shifting under the weight of this new political pressure first taiwan's tsmc is slowly beginning to diverse supplies production-based not in a dramatic fashion but nevertheless it's opening new facilities in Arizona and Japan and they also open facilities in in Singapore and in Europe down the road this is a major change for tsmc which previously has not had any uh substantial production facilities outside of Taiwan or China second China's been put off from accessing many of the most advanced Machine Tools in particular most importantly probably the extreme ultraviolet lithography tools that I mentioned earlier which it's it's impossible now for Chinese firms to buy third the U.S via new regulations appears to be dead set on stopping China from transitioning to new event new and more advanced transistor architectures which will allow the further shrinkage of transistors and as a result uh more advanced uh Computing uh Power uh fourth uh China continues to put money into its uh build out of fabrication capacity especially in less Advanced logic chips and we should expect this to have a big impact I think on the market for chips globally and on the profitability of uh of of of firms competing in this segment of the market and then finally most interestingly governments around the world are pouring money and into capital investment whether it's China or the us both which I've mentioned also Japan Europe planning 50 billion euros of new expenditure and then Taiwan Korea Singapore other players in East Asia as well almost every major government is putting new money into fabrication and a lot of this money is going to go towards lagging Edge fabrication risking substantial over capacity but governments are doing it anyway uh in no small part because they think the um the margin of security this provides is worth the cost so to sum up we're seeing a dramatic shift in the semiconductor industry away from an industry in which cost efficiency and technological advances were the two driving factors over the past several decades but a return in some ways to the earliest days of the industry when I was the demands of Defense production and concerns about military technology that were key factors in explaining how the the chip industry first emerged and how it over time and I think we should expect this Dynamic to continue because as long as the military balance is in question and indeed uh today the military balance in East Asia is Shifting as dramatically as it has at any point in the past several decades I think it's only natural and to be expected that governments will focus on Technologies where they think they might be able to achieve a military Edge and when you look at how defense planners are thinking about Advanced military systems including power signals processing Communications abilities are at the center of all these systems and for all that they'll be reliant on semiconductors so sum up we're seeing a a Chip War not only in the way we've seen in the past a battle between companies or uh or products or types of Technologies but also a return to defense and security concerns playing a central role in the chip industry and that will have a dramatic impact I think on the U.S China relationship and also the structure of semiconductor Supply chains across the the Asia Pacific region so I'll wrap up there I'm looking forward to hearing Paul's comments and then uh questions from the audience thank you so much Chris um and I would like now to invite Paul to give his feedback and also remind the audience that you can send your questions through the Q a box and we will we will talk about that after Paul give his comment ha please Angela and Chris great great uh a great summary of the book and I think um one of the things that um you know it comes through is this uh the sort of civil military fusion um and and the importance of semiconductors of course in the in the military domain which of course is is exactly the justification the US is using to restrict um uh various types of semiconductors and Equipment uh for China so it's an interesting Dynamic here I think Chris is absolutely right about the role of military funding and the priorities um uh in developing sort of semiconductor manufacturing over time but I think it's important to note too um that it's only because now the economies of scale provided by consumer uses and applications of semiconductors that have allowed the sort of the Innovation to really continue uh Moore's law to continue which Chris addresses in a book I think very well um and so if this tsmc for example the leading Foundry in the world has to invest 40 billion dollars in a new Fab um that will make sense economically they have to have a huge client base um over which they amortize the cost of that facility in r d Etc and the military portion of that of that in terms of Appliance is very very small but of course it's important um and it's as Chris notes it's it's certainly um become come back to being a more important uh uh issue which we can talk a bit about here in terms of the for example the recent export controls so I think the though that the war meeting that the chipboard meme and again I'm sort of guilty of that I use the the hashtag us China Tech Cold War um back in 2018 I think for the first time um which a lot of people have picked up um in the semiconductor industry It's Tricky I think you have to be careful because it has been well there's been a lot of competition and Chris I think does a great job of laying that out uh the sort of history there the the evolution over the last decade arguably has been you know one of more there's a lot of competition among companies um and but there's a lot of collaboration too and it's a very collaborative Global industry and so as Chris notes now as we sort of governments re-inject themselves into the picture the war that is more you know you have the sort of the great powers and um and U.S China Etc um and then there's the industry right so so I think it's always important to keep in mind the the sort of the difference between the government and its aspirations whether that's China or the US and the industry and how how those aspirations rebound on the industry I thought Chris did a great job in the book of um highlighting you know some of some of the um the the particular the people of the history um the keep the key figures in the industry but what was really interesting was there was sort of deja vu all over again so um for example he notes that Jerry Sanders the CEO of AMD in chapter 16 at war with Japan that uh he thought that the Japanese had four unfair advantages they benefit from IP theft protected markets government subsidies and cheap Capital those are exactly the criticisms of China by the US now um and so it's it's it's it's interesting to see sort of this come around full circle um and then I think it's it's fascinating that um he knows that backyard Marita the Japanese the head of Sony pointed out the problems with the U.S um too busy creating lawyers instead of Engineers too focused on this year's profit not enough training or motivation for shops or employees exactly the criticisms that Taiwan um and Morris Chang and others at tsmc are leveling against the U.S um in terms of the chips act you know whether the US is really firmly committed to to um uh long-term manufacturing which is very difficult um and requires a certain culture and and uh and sort of business approach and uh that that some think the US doesn't happen we can talk about that um so I think um the um you know the what I focused on in the book I think were that were two things was the China the China piece um and the Taiwan piece which I think are the Taiwan aspect of this whole thing I think is is really um you know the most important um part of this um and then the other thing that I think is important I think Chris does a good job of bringing this out is the importance um of the US company dominance of the tools that you know the the tools to make semiconductors those that's really and that's really where the new export control regulations for example are particularly important um and so I think that that that that can't be stressed enough in some sense that um you know the dominance of U.S companies like Applied Materials uh lamb uh Kaylee tank or in the trilogy side um and then the Eda tool players the the electronic design automation tools which are closely coupled with the manufacturing process uh and U.S companies like Cadence and synopsis and Mentor which is a Siemens company but really has a lot of usip there that's a really critical piece of this um and so for example in China there's China has sought self-reliance in some areas or sought to reduce dependence maybe a better way to say reduce dependence on foreign foreign technology but when it comes to things like Eda tools trying to reinvent a whole ecosystem around Eda tools really doesn't make sense in some sense you know to some degree because you have these these leading companies that have very sophisticated tools and are very tied into the foundries um and so why would you you know given given a choice why would you not just use those two tools um and um and you know and and use them um to design your semiconductors and manufacture them some somewhere you know as the the sort of fabulous model which Chris I think does a good job of outlining the book that's where the industry really is right so um with the dominance of a company like tsmc or other foundries like Samsung to some degree and Now intel getting into The Foundry business that chipless chip doesn't chip fabulous design chip design model is really what's made the industry go forward because tsmc has focused so much on the manufacturing side of it and becoming very good at that um and so the Chinese Chinese companies of course naturally um have have gone down that road too because um that's sort of the the industry standard and I so I think it's important that in doing that of course those those Chinese companies um like other companies are end up being independent on U.S technology and that has implications when the US chooses to use those choke points um as as Chris has noted so China I think is really important to focus on here quickly so China I think the sex on China's industry is good but I think there's there's some things that maybe could have been brought out a little bit more uh and again I've been looking at this way too long so I have I have a certain bias here um I think subsidies for example uh that Chris noted all the that government part of it you know that's not really a very good long-term business model right and neither is IP theft um then semiconductor sort of illustrate this right so despite a nearly a decade in China for example of basically Limitless subsidies China has produced really only one firm uh ymtc that's capable of competing at anywhere near The Cutting Edge uh or or globally right and smic to some degree but it was not as heavily subsidized until recently so it's not surprising that China would would pursue the Chinese government pursue subsidies um because they were way behind um and and all and many many governments pursue subsidies um but China's approach could never really for example that approach will not would not be able to compete or produce a Chinese version of tsmc for example because that's that's driven by market forces and uh the development of trust and and overall um sort of you know the the trust in tsmc and IP protection Etc so China's approach of government sort of top-down sort of subsidies which which they're going to probably do more of in the wake of some of the recent export controls is really sort of counter to intuitive to the industry because the industry globally is very Market driven um and so the more government spending and more government subsidies is not necessarily going to going to produce outcomes that are um you know that that are produced competitive companies and and up up to China's position in the global semiconductor um sort of division of labor the national IC fund which Chris mentioned uh was um was was established in 2014 lots of tons of you know tens of billions of dollars not all of which which has been invested and of course this year there's a huge investigation of the national ICI on for corruption which suggests too that the Chinese government hasn't been happy what the results produced by the by the the um by the national IC fund and at the same time we also had made in China 2025 and as part of the made in China 2025 uh sort of series of documents there was a document produced by the Chinese Academy of engineering that included specific domestic targets for for production so 70 of domestic semiconductors by 2025. now that was just done by some some very serious engineers at CAE but completely divorced from the reality of the market and so all of those targets that that are in that CAE document associated with made in China 2025 or wildly unrealistic and by 2025 China will not be producing anywhere near uh any of those targets but those targets are often cited um as an example of how China wants to dominate the semiconductor industry but really that again government sort of aspirations um and government policy preferences need to be separated from the sort of reality of the industry which is very very Market driven and then I think also Chinese companies were really far behind in 2014 um and so that this again this type of roadmap a roadmap was pretty unrealistic um and they remained pretty far behind depending on the the sub sector you're talking about because funding is really only one factor here in the industry um it can make a difference in some places but not across a complex supply chain with dozens of Technologies and the competition innovating faster I was talking to a CEO of a really Innovative us company Marvel and he said look I have thousands of Engineers great Engineers I have technology I have intellectual property and I can only move the needle a little bit in my market Niche um so again for China to to try to to reduce dependence and create dominance in any of these sectors is really is a huge Challenge and again China's Ambitions uh and the here the government in the industry are on different wavelengths um I'm on a track 2 dialogue with with China where we talk about semiconductors among other things and um one of the last dialogues we had we had weisha Jun um who's at qingua University and he briefed the China's polit Bureau standing committee when they do study sessions on technology which they do fairly often so he Brief Siege and paying on technology and he admits that everybody in the Chinese industry who understands this Global division of labor recognizes that self-reliance is a possible goal but they're under a lot of pressure from the government um their intense pressure to to reduce dependence um and U.S actions um like these export controls that we'll get to give you know give China no choice but to but to pursue sort of double down on the self-reliance Meme and so you see a lot of that even in the last year coming from the senior Chinese leadership um but again as I noted given their preferences Chinese industry and Chinese companies that were going to do chipless Fab design what would reach for you for the Avail the best available tools just like anybody in the world would do um and they wouldn't seek you know to to pursue alternative development but again the geopolitical landscape is very complicated now and so uh Chinese the Chinese industry is under a lot of pressure to develop Alternatives almost in every area uh to to U.S technology um and then I think Taiwan is really important Chris has a good chapter on Taiwan but I think it's really it really is the sort of Crux of the issue going forward as we look forward the status of Taiwan sort of looms large um the largest really issue in the industry right because um as the U.S woke up to the danger of this dependence on on Taiwan and tsmc and launch the chips act um in in the discussion started in 2020 around 2020 and began luring uh companies like tsmc and Samsung uh and others Intel to to build new Fabs in the U.S um this is just at the time that the US and Taiwan relation us and China relationship was was becoming much more tense technology competition at the center as secretary blinken noted in his me speech and so I think Taiwan then now has become sort of the focal point of all this and Taiwan has dragged into this in many levels because of those extra territorial export controls that Chris mentioned um so while the chip sack was designed to to you know to restore Advanced manufacturing the question there is going to be how much is enough uh tsmc given its druthers would probably not want to build a five in the U.S because they're very good at building incremental new Fabs in Taiwan they understand the market conditions they understand the government support um and the culture importantly right um and so they're very skeptical that they can that they about the US and its long-term commitment to manufacturing into those subsidies the chips Runs Out in 2026 for example so will the US continue to fund Advanced manufacturing Beyond 2026. um and then the question still is how much is enough so some projections at 20 by 2030 suggests the us could have you know five or six Advanced fives um and some number of advanced packaging facilities but the other pieces of the supply chain still not clear Upstream how much upstream and downstream capacity will be in the US but there will be some reduction of that dependence on tsmc but probably not all that much right because um the that picture that Chris showed of that the the Phoenix or the the Arizona Fab uh they're going to start with 10 000 wafer wafer starts per month there tsmc in Taiwan is well over a million and moving north um so in terms of of that facility putting a Big Bend in in the in you know making a difference in terms of the geographic concentration in Taiwan right it's going to take some time they will expand that Fab but it's not gonna it's not gonna be anywhere near um you know a significant reduction in that dependence for some time so Taiwan again for the next 10 years is going to be Central uh to the industry uh in a way that um I think that people now appreciate but it's really hard to change that because these are five and ten year Horizons for companies uh to build Fabs and to amortize the costs et cetera et cetera so the other piece of this though then is that as this is all happening um this this there's a there's a there's a focus on Taiwan in in the wake of particularly the Russia Ukraine uh situation people are sort of saying well China and Taiwan you know is similar I think that's a really bad analogy frankly I mean I think Chris mentions it in the book but I do not agree with the statement that Chris made in the book that what happened in Ukraine could not happen in Asia I think the two situations are very different and so I really disagree that that with with the sort of a lot of people make this analogy but again I think that the you know China and C are not Russia and Putin and in fact you know in contrast you could say that Russia and Putin uh make C in the Chinese leadership seem very very sane um it's it's inconceivable to me for example that Chinese leadership would threaten to use nuclear weapons here right um so the issue is really under what circumstances the U.S would accept a unification of China and Taiwan um and U.S the US policy right now is to continue for example relying on these 40 year old documents that say and say that nothing has changed that the U.S still supports the one China policy Etc but Beijing recently doesn't believe this doesn't believe anything bidenery U.S officials say on this they're looking at actions military sales the Pelosi visit um the indo-pacific economic framework and things like the chips for Alliance which which includes Taiwan um and you know other statements by U.S officials that essentially characterize Taiwan as part of the U.S defensive perimeter in Asia um you know as a bulwark against China which of course doesn't go down very well in Beijing and then Congress in the U.S is also weighing in for example um with all sorts of crazy bills on Taiwan which are again leading China to believe that the US does that no longer supports unification um but is really pushing for Taiwan Independence and so there then the question is what's the um you know would the US accept the peaceful reunification of Taiwan for example and this is where the semiconductor industry comes in because I think the answer is now increasingly no because of those facilities at tsmc would U.S officials accept China being in any position to control policy related to tsmc for example and particularly when you see the new export control package which is essentially cutting off Chinese companies from using tsmc as a manufacturing base so the problem is um you know how to how do you solve this this problem right and preserve the unique nature of taiwan's position uh mtsmc's position in the global semiconductor supply chain and so my argument is that you know there needs to be some really creative thinking here to to to to get Beyond this idea of arming Taiwan to solve the China problem right I mean I think um the again those documents that were negotiated in the 70s you know when China was didn't have a military Taiwan didn't have a semiconductor industry um and Taiwan didn't have such a vibrant democracy um you know don't really hold up right now out and I think that's the that's the that's the problem that we're facing going forward is how to solve the the Taiwan problem and the semiconductor piece of it now is probably the most complicated piece because um you know tsmc for example couldn't survive uh for very long without complete contact with the outside world um Taiwan and tsmc tysms in particular is in this real time the position where Upstream materials are flowed again Wafers and and materials and tools and and services from the US and Japan and and the Netherlands um and then semiconductors are being shipped out by air so any kind of an air blockade for example on Taiwan would immediately disrupt the global semiconductor industry in in a fundamental way um so I think that's what that's an area where some creative thinking needs to happen here because you know if we continue down this path of of U.S China great power competition with Taiwan caught in the middle of it um and semiconductors sort of uh you know also also uh part of the equation I think um you know the the results here um could be pretty pretty serious so finally I think the um you know the the issue of Chinese companies and the semiconductors Are Being Framed is Being Framed I think in general in in the wrong ways um yes the Chinese government for example is is um has done subsidies but all developed countries subsidize Advanced manufacturing so the question really going forward is what are the good subsidies and what are the bad subsidies um the oecd did a really good study last year on this and I think that's something that needs more more discussion but you know again the the approach of sort of cutting off China um and Chinese companies from different pieces of the of Islamic electric supply chain I think um rather than engaging in trade negotiations and discussing things like subsidies and and trying to really solve those problems I think is a really you know perilous path which is going to have unintended consequences these latest export controls for example which attempt to freeze China at at different nodes in logic and memory um have really created chaos in the industry and we have clients that are that have complained to me about them they were dropped sort of suddenly and um and those kinds of controls I think are going to have huge unintended consequences uh they're going to have an impact on those U.S tool makers and their profitability I was on a call last week with lamb um and they uh their their earnings call and they they're looking at a two to 2.5 billion dollar hit in 2023 depending on how those those x-port controls are are implemented um and other companies like Nvidia which which is now restricted from selling its most advanced gpus to China also looking at similar levels uh of of hit on this so I think um over the the the the short term there's going to be some big the impact on on both U.S companies of course and Chinese companies um ymtc uh and smic their leading Foundry are really in a in a in a difficult position now if the U.S enforces all the restrictions for example on those semiconductor manufacturing tools um that that Chris mentioned um and then finally again this Taiwan problem I think is something that needs to be addressed in a more creative way because further dragging Taiwan into the into this great power competition I think again is dangerous I've talked about uh unknown red lines for example so to the extent that these export controls cut off Chinese companies from being able to use tsmc you know this started with Huawei in 20 in 2020 um and now the latest controls add 28 more Chinese companies all their leading AI companies for example um you know this is Beijing is looking at this and saying hey you know Taiwan we think Taiwan is part of China but our leading tech companies can't use Taiwan as a manufacturing base so we don't really know what what how Beijing will will react to that I think they will react to these latest export controls um cautiously but at some point strongly because China remember controls some some Supply chains of its own and has some choke points of its own uh on things like rare Earth's magnets and um and EV battery supply chains so I think you know rather than getting into this this sort of supply chain War which this could evolve into um I think that that there you know that would be prudent for both sides to sort of step back from this but the problem is the overarching U.S China relations are on such a powerless state that is hard to see um either side sort of backing down and taking a look a hard look at Taiwan and how to preserve sort of the that these Global and and pretty sensitive semi-meter Supply chains that I think Chris really laid out well in his in his discussion so it's it's it's certainly true that the military and in the the driving force behind this are these military concerns that Chris mentioned in the case of china the use of gpus for example Advanced gpus to do modeling for Hypersonic Glide vehicles that might be might come back and haunt the US and Taiwan in the military contingency but I think the the bigger overarching issues should be trying to figure out a more creative way to preserve Taiwan um as as this and it's its position in the global supply chain because the alternative is really of conflict would immediately um disrupt not just the the the the semiconductor industry but the but the global economy given the importance um of Taiwan and all the Ripple effects that would have if even one missile landed in science park for example so I think that that's that's where we are now and I think Chris's book um really lays out uh in a fantastic way the history of the industry and then sort of how we got here um I think the going forward part um given the recent events um and just sort of the coming to head of the bigger you know great power competition um and Taiwan I think are really the sort of focal point um of of of the of the industry going forward we have clients who are asking us constantly for risk scenarios over Taiwan um and most of those are tech companies in the supply chains of of of of the Taiwan companies but also any company operating in Asia or in China is not worried about Taiwan um and that's sort of that's sort of where we are and I think the timing of this book couldn't be better because it really does lay out um the um you know all the issues at stake here um and I think um the the the those risks I think are real um that going forward and I think we need some we need people to be informed about how the industry Works which again I think the book is great at doing that um but I think we need a new way to think about how the industry fits into great power competition um and in particular um the Taiwan issue let me stop it okay thank you Paul um I saw there's a couple questions in the chat box already and before we we move on to the audience question I actually uh well you want to take advantage of the opportunity to ask you guys some questions myself um feel free to jump in I I one of either of the speaker so my first question is really a Layman's question um what I wanted to understand you know a couple years ago I speak to an industry expert in China and then he told me which I'm a little bit skeptical that um look I mean China can produce chips in fact China produce a lot of trips and what we are lacking behind is the most cutting-edge technology like those smallest tiniest chips that only tsmc can produce like so but it looks like now we're in a very serious uh position like I mean it so the first thing I want to understand is like what exactly is the technological gap between China and the U.S or other Western countries are we only talking about the size of the chip or there's many other stuff that we're missing here well I think the the first party answer the question is to remember Moore's Law which has defined Improvement in the summer industry for the past half century and so what Moore's law says roughly speaking is that the computing power produced by the most advanced ship doubles every two years so if if you're two years behind you're only 50 was good and if you're 10 years behind you're really far behind um and so that that's that's why being one or two or three generations in advance it's more advanced than uh your competitors is a huge huge difference because no one's going to buy a computer uh with a third or a quarter or a tenth of the computing power um as we can get in The Cutting Edge and especially when it comes to Data Centers and artificial intelligence applications having access to more computing power is absolutely critical um and so if we project forward in 10 years time if if China is held constant in terms of the quality of advanced gpus and the rest of the world has not held constant and it's growing in Mexico rate it'll be a large gap opened up now as as Paul mentioned the the tools needed to reduce Advanced chips are really complicated need software any materials you need Machine Tools um and in most of these spheres China's quite behind in terms of what's domestically available um it's reliant on tools from Japan tunnels from Japan rules from the US and the Netherlands and so the reality is that the most Cutting Edge you can produce at scale in China is uh three or four generations behind what you can do in Taiwan and I think we ought to assume that over the next uh at least five years and probably decade that number will grow to five six seven generations behind what you can produce in Taiwan and so consider the doubling involved meets generation shift that's a really vast Gap that's in the process of being opened up yeah Angela and I think it's a great question I would just add that you know that China was set to move up the that that scale but then when smic was going to buy that euv tool the the extreme ultraviolet lithography tool from asml but there was a control put in place in 2019 time frame by the vasenauer agreement which is the multilateral uh control regime that was really set up to to do weapons of mass destruction and other military end use um and so the U.S government pushing that control basically has essentially prevented smic which I think actually had a contract and apparently every time Chinese officials go to the Netherlands they ask where is our euv machine um but that's that's really a bottom line because as um Christopher noted um it's a monopoly only one country in the world produces that a piece of equipment it's incredibly complicated it's probably the most complicated piece of equipment known to man there's like 5 000 part suppliers to to the to SML for that piece of equipment it's you know 250 million dollars I think the latest one might be in 400 million dollars the latest version of that it's really incredibly hard to to develop and maintain it requires a huge sort of you know service Trail lots of 30 phds to come in and help you set it up and so China can't really for example create that technology um out of whole cloth they can't steal the IP they can't you know they can't sort of invent a new system that's better than that right um in in the in the near term now maybe over a 10-year period there will be they can come up with a new advances or other ways to do this but it's it's sort of a bottleneck that can't really be overcome so that means China's smic is really Frozen at around 14 nanometers they've used their existing ultraviolet lithography equipment to get down to some level at seven nanometers but it's really tricky to do that and get commercial yields um and in fact the buy demonstration in part was reacting to that the media reporting around smic getting to some level at seven nanometers in terms of these export controls because there was a feeling because there's no companies in the US that are producing at 70 meters right Intel is is is you know is is in the game there but most of their production is not in the US um and so the uh the the issue of sort of moving up that that technology scale is really tricky and even if they even if smic had that equipment it would take them a number of years to master and then as I say with these really high-end and expensive Fabs you need to have a big customer base over which to amortize the cost of that and it's not clear that that the economics even in China would work out um without some change in the way the industry you know operates and they would have to get a lot more clients and customers who would trust them and their technology um so that so China is sort of you know there's this sort of screwed here because of this this this export control and then the latest export controls set the bar at logic which is what we're talking about here at 16 and 14 nanometers and the the problem there is that the industry the rest of the industry doesn't like that and so it's possible for example that the tool makers in Japan and even asml will not respect that that control because right now it's a unilateral U.S control it's not been taken to Boston hour yet um and so it's possible that the Chinese could and I think that this will happen will be able to continue to to to function and and have Fabs operate at those at the 14 nanometer level because it's it's such a lucrative market for for companies that they don't want to give that up um and so the the result of this one result of this export control ban for example will be to basically freeze out U.S companies and Japanese and asml and other smaller tool makers could take over the market in China and keep those Chinese fives running but they're still going to be stuck somewhere between 14 and 10 nanometers or seven nanometers and they're not going to be able to to go to the really Advanced chips that are in your iPhone um or in your 5G base station that really require that that fight to be five five nanometers are less because of things like power consumption and just you know sort of the overall you know Computing capacity that Chris mentioned that you need for certain applications but China again will continue to produce at some of those Legacy nodes which are are still very important so it's not like they can't they can't they can't produce the problem is if you cut off that advanced that advanced piece the business model becomes tricky because those those Advanced parts of the of the are much more high margin and so what you do is you you generate a lot of Revenue at lower margins with your legacy nodes and then and then you generate newer uh the newer nodes you can generate a lot more Revenue which you then plow back into r d so if you take that away the business models become a little problematic in China and so that's what people are grappling with now um as the US government has tried to freeze them at this this node for logic and and nodes for memory too so us has dragged memory into this too um and that's been complicated because there are companies like skynax and Samsung that are doing a significant amount of memory Manufacturing in China and so those facilities are now in Jeopardy because they have no road map for example going forward right great great insights thank you both um as I was reading Chris book um I particularly focus on the the second half because talk more about China I care care about what is the the severe situation that China is facing right now I I increasingly become quite pessimistic about the Outlook and looks like Chinese government have wasted a lot of resources I'm passing along on a lot of companies and that doesn't seem to go seems to have gone nowhere so I mean one big one quick question I have because we have quite a number of questions from the audience already is do you think China the Chinese government is doing the right thing at the moment I mean it's is it is it pursuing the most optimal approach uh in order to address um you know these these urgent demand from its industry and to deal with the geopolitics I mean I mean is or should China do something different I mean what are the missteps that the government have have made I mean I welcome inputs from either of both you both of you I think the the economically optimal strategy for China would be to not do what it's doing and instead try to integrate more deeply to International Supply chains and if you look at um what made it possible for first Japan and South Korea then Taiwan to win important shares of my electric supply chain they did so not primarily by focusing on self-sufficiency or independent development they did so by integrating really deeply uh with uh with the rest of the International Supply chains that's true whether you're looking at Japanese firms in the 60s 70s and 80s or South Korea's emergence as a memory player in the 80s and 90s or best example is tsmc which on the one hand is the world's most important ship maker the other hand as we've discussed couldn't exist without tools and software and customers around the world and they've become successful because they've been more integrated than anyone and they're the center of the ecosystem the challenge that Chinese industry faces is that because of China's military buildup the U.S has concluded it has no choice but to focus on computing power as key to what remains of its military Edge in Asia and so the U.S strategy is basically to say that you know the U.S knows the Pentagon knows it's going to be facing China will produce more ships and more planes and more drones the more missiles than the US will especially if you focus on just Assets in in East Asia and so the US has got to try to it wants to keep its military Edge I try to keep a edge in quality because it's going to lose out quantity and so the Pentagon believes that its Primary differentiator in terms of quality of Defense systems is computing power signals processing sensors and that's all about chips and so the US strategy is basically to double down on its Central role in the chip supply chain and put off China from accessing these systems with the goal of opening up the Gap that I mentioned in Computing capabilities over the next decade and hoping that the US will be able to apply Advanced Computing to its military systems so that China will fail to do so in its Computing systems and thereby the US hopes to keep what remains of its military Edge now given the current geopolitical landscape I think um that Dynamic is really difficult to change and so uh I think it would require a change in China's foreign policy and defense policy ultimately to open up a situation in which Chinese firms could integrate in the way that was necessary to produce another tsmc for example yeah I I totally agree with everything Chris said I mean the problem is China the Chinese government has you know very few choices here I mean as I noted earlier uh the the idea of sort of government subsidized and top-down development of the industry is just at some level doesn't compute um because that because of this Global division of labor and the sort of you know the need for companies to be able to sort of um specialize and focus on where they have comparative advantage so for example the Japanese government right now they have a very well planned semiconductor strategy they lost a lot of them they lost the manufacturing War if you will so they're not they're not focusing on trying to out-compete tsmc they're focusing on Next Generation packaging um and they want to do r d with the US and they want to sort of they want to be a player they want their companies to be a player in sort of 3D packaging and the next chiplets which is going to be the next sort of uh a way to extend Moore's law for example so they're that's what governments governments should be helping their companies to to to focus on Niche areas in the industry where they have an advantage or whether they what they can play but the problem in China is that because of export controls and because of the the Geo the geopolitics that Chris mentioned there they're sort of stuck with trying to figure out how to how to subsidize and and do it do every part of the supply chain virtually every part of the supply chain and again that's just you can't really do that you don't have enough people you don't have enough intellectual property I mean essentially right now China is Awash in capital for semiconductors not only do you have the national IC fund you have local government guidance funds you have the Star Market in Shanghai so the Chinese government has introduced sort of Market mechanisms where promising Chinese Summit can enter startups for example can go public and raise raise capital on on domestic markets but at the end of the day you know that's that's a problem because that's that that approach is not going to produce a tsmc or a really big company that that can compete or even the niche players that and there will be some Chinese companies that that will move up the the chain you know both in terms of the design and production and um and things like manufacturing equipment but the the scope of U.S export controls is really a a huge impediment to this because the latest package for example not only includes specific gpus and Manufacturing equipment it includes the semiconductor input to semiconductor manufacturing equipment right so it's it's been sort of designed as this interlocking thing to really freeze China um and really make it difficult even if the government pours more money in for that for those companies to to innovate and develop and move up the chain because they're cut off um from a lot of of key inputs and so they have to develop almost the whole you know restart and start start at zero and invent the whole thing from from the beginning right and there's just you know that we're talking about you know an impossible task not because of the money but because of the people trained engineers and also you know know business managers uh for example one of the big things lacking in Chinese companies is just business managers who know how to run a business somebody came up your business and know how to make that work over time um and so that you can continue to to you know innovate and reinvest and and get have a client base and that's really what's been lacking also so China is is in a really difficult position so the government will double down but there's no easy answers to any of this right they're not going to be able to to just spend their way to you know back into the game here for example so they're going to have to focus domestically and hope that they can still access some of these other Technologies from non-us sources um like Japan and the Netherlands and other other parts of the world that may not may not go along fully with the US um the policies and and restrictions right I mean so I mean these are a great great points um we are now a little bit out of time and now we have four questions from the audience let me try to summarize the audience question in in one big question for you you guys so most of the questions revolve around the the consequences of these restrictions and obviously vast majority were imposed from the US side and do do you think that those restrictions uh really achieving the goals uh of of the US government or do they have some unintended consequences I I believe that Paul may have mentioned some of them in his earlier remarks um but generally speaking I mean are these Tech embargo on polls on China right now uh are they doing what they're intended to do or are there any unintended consequences I think there will definitely be unintended consequences of the deck that Paul mentioned the companies will be harmed uh because sales will be lower than expected things like that um I think those were expected unintended consequences I guess I would say from the perspective of of the US government I think I I'm in agreement with all of the controls will work in terms of their intent of causing uh dramatic delays in China's ability to advance semiconductor production um so in that sense they will work I think it's pretty plausible they will work in their broader military goal of opening up a gap between the Technologies available to the Chinese military versus technology is available to the US military over the next decade that seems pretty plausible I think the biggest unintended consequence that is hard to answer right now is what uh if any way the Chinese government will retaliate um in the past the Chinese government is actually not retaliated uh in against prior rounds of export control so when the US put Huawei for example the energy list trying to talk about values and didn't do so and so it'll be interesting I think to to watch what kind of decides to do this time whether it's it's rhetorical uh whether it's economic uh or whether it's something else in the Diplomatic or military sphere as well and I think the greatest risk of identity consequence is if in fact China does retaliate setting off a spiral of of a further either economic tension or political military attention um that's that's the risk that I see is is most worrisome even though if I had to guess my guess would be that the Chinese government retaliation States primarily in the rhetorical and maybe a bit in the economic sphere um judging by by past responses to comfortable export controls yeah totally agree I mean I think that um we're every every every one of my clients is asking what's the Chinese response going to be um and there's a whole range of things there and I think that the the strength of that response will be that will be really important to watch I think that though the independent consequences are potentially really unraveling the Indus the overall Global semiconductor industry in in unexpected ways so for example in the in in 2020 when Huawei in 2019 when Huawei was put on The Entity list and then later um was the the foreign direct product rule cut them off from from Taiwan um that Ripple through the industry and probably contributed a big chunk to the global semiconductor shortage because it changed the way procurate managers thought about buying semiconductor so they stockpiled a bunch of stuff and they they upped their orders and so that that sort of rippled through the system and you know one one executive in the industry told me that they thought that that the Huawei any list caused the whole Global semiconductor supply chain orders and I think that might be an exaggeration because they were other factors there but that just shows you how the industry is very delicately you know delicately sort of balanced between these Global players and so when you when you disrupt something over here the the effects can Ripple through the system in unexpected ways and I think here we're probably going to have this happen um because um the the as I said the the US tool makers probably are probably going to have to pull back a lot from China we're going to have some other players move in but that but it will be it will disrupt the cycle of innovation because losing for example a big chunk of the China market for the US tool makers will mean that their r d budgets are are impacted pretty heavily right um and then that will Ripple back into their ability to innovate which some some have argued is a national security concern for the U.S because that could undercut their Global Leadership because the problem is if you take China out of the equation it's not clear that there's not enough of a market of for example globally to make up for the loss of a a big chunk of the China Market um and so so you know maybe you can as I said the chips act you know there'll be Fabs built in the U.S but probably that will not make up if those tool makers for example lose even more of the China Market than they've already lost um over time as the Chinese government you know subsidizes other players uh you know the Japanese and and other companies move in so there's there's a lot of unexpected sort of second and third order collateral effects that we haven't we don't really know yet but I think you can probably bet they're going to be bad for Innovation and bad for sort of the industry going moving forward in the way it has before with you know Market driven uh economies of scale uh you know capital and people flowing into the most optimal kinds of uh positions which again arguably produced Taiwan and tsmc being the world being too dependent on them but it also has allowed the Innovation that has allowed us to cram this much computing power into into this small space and put it on a missile right and so the question is you know is that Dynamic fundamentally changed or is does this does the does the industry sort of adapt and and restructure its Supply chains in a way that can help you know maintain that development and maintain Moore's Law and I'm I'm not sure that I'm a little more pessimistic on that score um because I think the industry is is very complicated and messing with it too much uh May produce you know really bad outcomes wonderful I'm so glad that even within an hour we were able to go so deep into this question and I really want to thank um Paul and Chris uh for joining the discussion today and there's so much more in this book um so I really highly recommend you get a copy it's it's quite cheap I think for a book of this size um and you got a lot out of it um and um I I personally learn a great deal from today's discussion and again I want to thank the speakers and thank the audience for joining us I'm sorry to uh those audience who raised question we are not able to go through all of them um but um uh we hope that we have the opportunity to invite you guys back next time uh for for longer and more part of uh discussion so thank you all and goodbye thank you

Channel: Philip K.H. Wong Centre for Chinese Law

Views: 10,515

Keywords:

Id: nd9JVM7H9LE

Length: 69min 37sec (4177 seconds)

Published: Fri Oct 28 2022