Chris Miller, Author of Chip War, at CIC

foreign [Music] again good afternoon welcome to the third Chancellor's lecture series event for this Academic Year this event titled Chip War computing power and the U.S China competition today the today Chancellor Lewis the wdt teaching team and I have the pleasure of welcoming Dr Chris Miller I'll formally introduce him in just a few minutes the content of his recent book chipwar and today's discussion supports multiple ndu cic and wdt learning outcomes to include the evaluation of cyberspace capabilities in the achievement of National Security goals and the identification of implications from emerging and disruptive Technologies on the changing character of War so if you've had an opportunity to browse Chip War you probably found several aspects quite interesting a couple things that stuck out to me is the reoccurring character of William Perry long before he served as Secretary of Defense Perry was instrumental in detecting and advancing missile technology with with chips it just so happens that ndu is also the home of the William J Perry Center for hemispheric defense studies the book also includes a short chapter on how a potato executive helped to revive America's computer chip industry in the 1980s and he did that by investing in an Idaho company that specialized in dram chips so as a nod to that chapter I brought some potato chips here feel free to grab some on your way out but please no eating in the auditorium and the New York Times review that you probably read describes Chip War as a non-fiction Thriller one that makes a whale of a case that the Silicon chip industry now determines both the structure of the global economy and the balance of geopolitical power so with that I'll now I'll now give you the standard introduction Chris Miller holds multiple academic positions to include associate professor of international history at the Fletcher School of Law and diplomacy at Tufts University he is the gene Kirkpatrick visiting fellow at the American Enterprise Institute and he is the Eurasia director at the foreign policy Research Institute at the Fletcher school he teaches courses on U.S Russian relations Russian foreign policy and the history of U.S foreign relations and without further delay thank you for your patience please join me in extending a warm welcome to Dr Chris Miller thank you very much for uh the invitation to join you all today uh delighted uh to be here to share some conclusions uh from my book and what it means for U.S China competition across multiple different spheres I hope you'll come away at least understanding some of the differences between potato chips and computer chips and perhaps something beyond that as well during the Reagan Administration there was an economist who worked in the White House who was alleged to have asked once potato chips computer chips what's the difference and so I hope I can convince you that in fact there is some pretty crucial differences between the production of both of those types of chips important though both of them are I I began uh this project not intending to write a history of the semiconductor industry I'm a russianist by training my previous three books I have been on different aspects of Russian history politics foreign policy and economics and I started eight years ago planning to write a history of the development of military technology during the Cold War question that motivated me at the outset is why was it that during the Cold War the Soviet Union could produce all of the key military Technologies of the early Cold War like Atomic weapons and long-range Delivery Systems but by the end of the Cold War had struggled to keep Pace with the United States and its allies and as I dug into that question it seemed to me that Computing Technologies were at the core of the answer uh not just any specific type of military system but Computing that was distributed across different military systems that helped them communicate together to acquire new sensing capabilities to understand all the new information that defense systems were getting which then raised another puzzle well why couldn't the Soviet Union do that too they had lots of smart physicists lots of investment in their military r d programs and yet the United States jumped far ahead in terms of its Computing capabilities by the end of the Cold War such that when the U.S military arrived in the Persian Gulf War in 1991 it stunned both itself and the rest of the world with its ability to deliver weapons with extraordinary Precision in a way that ended the war far more rapidly than almost everyone in Washington and also in Moscow thought possible and it seemed to me that semiconductors were at the core of this shift in military power because semiconductors are the device that makes all modern computing possible and as I was doing this research in the evolution of military technology we were learning in real time the extent to which the rest of the economy was dependent on semiconductors too the last couple of years of semiconductor shortages have Illustrated that ships are not only important for smartphones or for PCs or for data centers but also for dishwashers and coffee makers and Automobiles indeed the entire economy today really can't function without access to some of the conductors which provide the computing power the memory capabilities the sensors and the communications that all manner of devices rely on today and so what I'd like to suggest to you is that you can't understand the modern world without putting semiconductors at the core of your story and you can't understand either the trajectory of the global economy or the balance of military power without looking at the role that semiconductors have played in the past and will increasingly play in the future now now what is a semiconductor perhaps to start for those of you who haven't uh done phds in electrical engineering or Material Sciences I indeed had not done when I started this book it's a piece of silicon often the size of your fingernail with thousands or often millions or even billions of tiny transistors carved into it a transistor is just a switch that turns a circuit on or off when the circuit's turned on it produces a one when the circuit's turned off it produces a zero and all of the ones and zeros undergirding modern digital Computing all the data processed on your smartphone or PC all of the data remembered on your devices is all just ones and zeros produced by circuits flipping on and off on Tiny silicon chips and in order to fit a billion or in the case of a Cutting Edge smartphone 15 billion tiny transistors on a small piece of silicon these transistors need to be extraordinarily small and today the most advanced transistors of the type that are in your smartphone are measured just by a number of nanometers they'll often have features just a dozen or so nanometers wide or tall which means that each of the transistors on your smartphone is smaller than the size of a virus and they're produced with almost perfect accuracy billion by billion powering the smartphones and PCs and data centers that we all rely on the process of manufacturing transistors by the billions with basically perfect accuracy is the hardest manufacturing process humans have ever undertaken there's nothing we've produced more of in the world than transistors in fact we've produced more transistors just in the last year than the combined quantity of all other Goods produced in all of human history nothing else comes close and to produce billions of transistors at nanometer scale that's billions of a meter in terms of scale requires a complex supply chain that stretches from Europe to the US to Japan to Taiwan where today Taiwan produces over 90 percent of the world's most advanced processor chips including chips in your smartphone likely chips in your PC and chips in all of the data center and Telecom infrastructure we rely on and in addition to producing 90 of the most advanced processor chips Taiwan also produces over one-third of the new computing power the world adds globally which means that Taiwan isn't simply important for its just strategic implications is also important because the world economy would freeze if we lost access to Computing computer chips produced in Taiwan where one company the Taiwan semiconductor manufacturing company is both the world's largest and the world's most advanced chip maker and today almost all of their chip making capacity is on a small island just off the shores of China which makes the chip industry important not only because it will Define the future of U.S China competition in terms of Chip's relevance for military uses for intelligence uses for the broader economy but also because the production of chips is deeply intertwined with the geography of U.S China competition and indeed the island that is most likely to be the source of a potential future crisis between the United States and China is also the island that produces all of our most advanced processor chips which of course is a situation of extraordinary risk that we and the rest of the world find ourselves in today so Taiwan produces the most advanced chips in the world today but it does so taking advantage of a supply chain that stretches across many different countries and if you want to manufacture and Advance semiconductor today you need to start before the manufacturing process with designing what your Chip is going to look like and a chip that has 15 billion transistors on it is extraordinarily difficult design you can't design that by hand they're just too many components you need to access specialized software tools that are produced just by a couple of companies all in one country the United States which has almost a monopoly on the production of the software tools needed to design semiconductors once you've got access to design tools you need to actually undertake the design of semiconductors and although semiconductors have the same basic structure the designs are very diffuse a chip designed for communications purposes will be very different from a chip designed for sensors a chip in a smartphone will be very different from a chip in a Data Center and so there's a lot of unique specialized knowledge that the companies that design ships I have inside of their inside of their companies and their employee base and most of the most advanced chip design firms are based in the United States companies like apple which is one of the world's most advanced designer of chips and smartphones and in PCS companies like Nvidia based in California which designs almost all of the chips used for training AI systems and data centers companies like AMD Qualcomm and others play a leading role in the chip design business even though most of those firms do all of their Manufacturing in Taiwan once you've got your chip designed you need to acquire the machine tools and the materials needed to manufacture Advanced semiconductors and because of the extraordinary Precision involved in manufacturing billions of components at virus size scale the machine tools needed to make chips are the most complex machine tools ever invented in human history with really nothing else that comes close to make an advanced chip you need lithography tools that shine raise of light through very precise patterns to carve shapes into your silicon chip you need deposition tools capable of depositing thin films of materials just a couple of atoms thick on your piece of silicon you need etching tools that can etch tiny Canyons just a couple of atoms wide to create the shapes of the 15 billion transistors that make your smartphone work and these machines are extraordinarily complex to make themselves drawing on very sophisticated International Supply chains I'll just describe one of these machines to you to illustrate the amount of precision involved the first and most important step in ship making is called photolithography carving or printing shapes with light and today if you want to make an advanced semiconductor you need to not just use visible light because visible light is far too broad a brush with which to paint the tiny shapes on Advanced ships you need light with a wavelength of 13.5 nanometers the most precise precise wavelength of light we can produce at scale which is used to carve shapes into advanced ships today and producing light at a wavelength of 13.5 nanometers is extraordinarily complicated there's just one company in the world that knows how to do it asml based another the Netherlands the most uh the largest tech company in Europe by market capitalization which has unique capabilities to produce light with the wavelength and here's how you do it you have a ball of 10 30 microns millions of meter wide falling through a back a vacuum you pulverize it twice with one of the most powerful lasers ever deployed in a commercial device the light explodes into a plasma measuring 40 times hotter than the surface of a sun the plasma emits photons at just the right wavelength which are then collected by the flattest mirrors humans have ever made and then directed towards the Silicon on which your chip is being carved undertaking this process of engineering is brutally expensive each one of these machines costs 150 million dollars a piece making the most expensive Machine Tools in human history that require multiple 747s to transport they require Personnel from asml the company that makes them to be on site for the life cycle of the machines and there's only one company in the world that knows how to do it and that's just one of the machines you need to produce Advanced semiconductors the other machines that I've described that deposit or that etch shapes on semiconductors are scarcely less complex and because it's so complex because the engineering is so precise and because the skills needed to build these machines are so unique the companies that know how to do it have almost monopolistic positions in the supply chain you can't make an advance semiconductor without acquiring machines from five companies one base in the Netherlands three in California one in Japan without which is simply impossible to make advanced chips and these companies have been in their Market positions for decades and cases even for half a century and although all of the most advanced chips are made in Taiwan today the tools that make those ships are imported into Taiwan from Japan the United States and the Netherlands so the chip supply chain stretches from the Netherlands to the United States Japan all the way into Taiwan and that's not even discussing the ultra purified materials the Silicon Wafers and and gases that are needed to manufacture semiconductors there's really nothing else in manufacturing that comes close to the complexity involved because there's no other type of manufacturing that is anywhere close to the Precision and ultra low fault tolerance that making semiconductors with billions of components necessitates now the international supply chain that I've just described to you is a Triumph of human efficiency it's uh it has produced advances that have made possible the distribution of computing power across Society you can't live an hour of your life unless you're asleep without touching multiple semiconductors usually hundreds or thousands of semiconductors and although almost none of us have ever bought a semiconductor all of the devices we rely on have dozens or hundreds or thousands of semiconductors embedded inside of them and that's because chips today are both more powerful and cheaper than ever before and in 1965 in the early days of the chip industry Gordon Moore who's a an inventor a chemist and one of the founders of Intel one of the leadership firms in the US noticed that the amount of computing power on each chip was doubling every single year and that relationship came to be known as Moore's Law and more predicted would last for at least a decade all the way through 1975 and and in fact it's lasted all the way up to the present roughly every two years the amount of computing power per chip doubles we find ways to fit twice as many transistors on chips because we've shrunken down transistors from the point when they where they were visible which is what they were in the early days of the chip industry to the virus sizes transistors that make possible Computing on smartphones and PCs today and that's only been possible because we've built these extraordinarily complex and precise Supply chains that have both produced more advanced ships and ships that have lower cost for the computing power they provide than ever before there's no other segment of the economy you know the segment of technology has progressed with anything close to the rate of Moore's Law nothing else doubles every two years I like to think about what life would look like if airplanes flew twice as fast every two years it's it's sort of impossible to conceive how fast they've been going I think be near the speed of light by this point um but semiconductors have done it year after year they've delivered on the promise of exponential growth which is why the first commercially valuable chip sold in the early 1960s had four transistors on it and today we're at 15 billion for the chip inside of your smartphone that growth rate is simply unparalleled by any other technological trend line any other part of the economy so in some ways the chip industry is an extraordinary success a Triumph of human Ingenuity a a benefit of globalization that's allowed the efficiencies that have made this industry and its extraordinary advances possible but in other ways as I as I've alluded to uh already there's deep vulnerabilities embedded in the way the supply chain is structured today and in particular our Reliance on semiconductors that are manufactured in Taiwan is an extraordinary vulnerability particularly given the growing Chinese threat to the island in addition to that there's a growing vulnerability in the fact that America's advantages relative with adversaries have been declining in recent years it used to be the case even a decade ago that the US and its allies were far far ahead of any potential adversaries when it came to the ability to produce the components that were necessary to develop computing power both chips used for computing chips used for memory chips used for communications and sensors the US used to have a last lead over adversaries like Russia and China but since 2014 the Chinese government has identified semiconductors as an extraordinary vulnerability that they face and it's understandable why the Chinese think they're vulnerable when it comes to semiconductors because today this year like every year for the past decade the Chinese government has spent as much money importing chips each year as it's spent importing oil there's no product on which China is more Reliant and that's because China can't currently make the most advanced semiconductors at home it has to import them almost exclusively from China's adversaries Taiwan Korea Japan and the United States and so since 2014 Xi Jinping and other Chinese leaders have identified as a priority finding ways to build up China's ability capabilities to domestically produce chip building up Chinese Champions that can produce chips comparable to what are designed by U.S firms and manufactured in Taiwan and as part of this industrial policy program China has been spending tens of billions of dollars a year in subsidies at the national level as well as at provincial and local levels to build up China's ship industry acquire new technological capabilities and establish firms that can compete globally in the production of advanced semiconductors and over the past decade China has made some real strides when it comes to its chip industry certain types of memory chips for example uh China's reached The Cutting Edge in terms of production when it comes to manufacturing processor chips China is still meaningfully behind Taiwan but it's much less far behind than it was a decade ago when it comes to designing certain types of chips like the chips that go into smartphones Chinese companies have proven themselves just as capable as American firms like apple or Huawei and so they Gap in aggregate between U.S capabilities and Chinese capabilities has narrowed substantially over the past decade and that's important not only to smartphone makers and PC manufacturers although it's very important to them it's also important because of the the Strategic and the military ramifications because as I suggested at the outset there was a deep relationship during the Cold war between America's Computing capabilities and its intelligence and Military capabilities that flowed from that it's always been the case that intelligence has uh has depended in no small part on computing power whether it's the British computers during World War II crack Nazi codes at Bletchley Park or American anti-submarine Warfare efforts in the Cold War which funneled undersea sound data into the most advanced supercomputers that the United States was able to deploy or the fact that the National Security Agency and the Soviet KGB both had privileged access to their country's most advanced supercomputers there's a direct relationship between intelligence and computing power because Computing is all about acquiring data and making sense of it and that's exactly what intelligence agencies do too and so if you accept my claim that historically Computing has been at the core of the race to develop and deploy intelligence capabilities I think it only holds it over the next several decades that relationship will be remain important and probably even become more important in determining which countries have the most advanced intelligence collection and processing capabilities and similar for military systems I think there's no doubt that Computing will be at the center of Next Generation military Technologies too it's not just the question of can we deploy Computing to specific military systems it's also can we collect data make sense of data and network more and more devices together and use computers to design more effective systems this has been something that US military and the defense industrial base have been doing to a large degree over the past several decades but its importance is only growing and especially as we begin to think about the application of semi-autonomous systems into military environments the role of computing only grows if you want to train a system to act autonomously or semi-autonomously you do so in a data center data centers are where autonomous systems are created and just like you need a very Advanced Data Center to teach a car to drive semi-autonomously you also need a very Advanced Data Center to teach a drone to fly semi-autonomously and so in no small part I think the capability of future semi-autonomous military systems will depend on the quality of the data centers in which they're trained and data center quality is basic a function of how advanced are the chips that you have inside of them which is why last year the Biden Administration released a sweeping new set of export controls that limit the ability of China to acquire the most advanced chips for training AI systems and data centers which are designed by U.S firms and manufactured almost exclusively in Taiwan up until last year it was legal for Chinese firms to import AI chips from the United States so long as they were supposed to be used for civilian purposes there was plenty of Open Source reporting of ways in which ostensibly civilian ships were being deployed to military uses but the core problem is that the exact same chips that can train cars to drive can train drones to fly there's no way of differentiating between the different types of chips it's just a matter of knowing how they'll be used in the U.S concluded I think justifi I believe it had no idea how tips were being used once they were entering China and so now it's illegal to import the most advanced chips between AI systems into Data Centers to China precisely with the goal in mind of limiting China's ability to train more advanced systems for use and intelligence and in military systems so the the bite Administration is betting that computing power will be at the center of strategic competition given its use for intelligence and Next Generation military systems and the Chinese government is making the exact same that the Chinese military has been open in its belief that the application of intelligence to military systems will be Central to the future of warfare we know Less in open source discussions of how the Chinese intelligence agencies think about this question but I see no reason to doubt that they've drawn similar conclusions about the necessity of acquiring lots of data and therefore the means to process that data so there's a an arms race going on right now in Computing Technologies which means that there's an arms race in semiconductors and if you think like I do that a key key component to the quality of Next Generation systems will be the Computing capabilities we've used to train them the arms race and semiconductors May well be as important as the race to produce any given type of system because chips will go into training all types of systems it's a critical component across all the different types of capabilities we'd like to harness and deploy over the coming decades and because it cuts across all of these different spheres semiconductors are are at the core of both how China and the United States perceive the competition now the U.S benefits from the fact that today although the US is far from capable of producing Advanced ships on its own it relies on machines with Netherlands in Japan Manufacturing in Taiwan almost all of the process steps for making an advanced semiconductor are in the hands of either U.S firms or firms and close allies or Partners whereas China can do almost nothing in the sphere of semiconductors that can't also be done by other countries so in terms of of pure capabilities the US and its allies are meaningfully ahead of China even though the Gap has narrowed but in terms of where the manufacturing and assembly happens much of it happens in East Asia and especially in Taiwan and in China and and here I think is uh is the source of much of the risk we face today our Computing capabilities both for civilian uses but also to no small extent for government uses are produced in East Asia in Taiwan and often assembled and packaged in China which both presents security risks do we have the knowledge we need of how they're manufactured whether they're manufactured the way we believe that they're being manufactured and it presents risks that in case of a crisis we might lose access to the chips that we require in economic terms it would be a disaster if we were to lose access to chips made in Taiwan it'd be very difficult to produce a smartphone anywhere in the world the next year if we didn't have access to chips made in Taiwan PC production would fall easily by a third perhaps by half data center roll out would grind to a halt we wouldn't have any AI chips it would take months if not years to find alternative sources of Supply modern cars have a thousand or so semi-conductors inside of them many of which are made in Taiwan it would be a disastrously messy process to find ways to Source the chips needed for auto production from non-taiwanese sources and that's to say nothing of the dishwashers and the microwaves and the coffee makers that also rely on semiconductors that are simple but many simple semiconductors today are made in Taiwan and there's not much spare capacity if we were to lose access to Taiwan given the extraordinary amount of capacity that exists in Taiwan so there's a huge risk hanging over the world economy the World Tech sector and also the US government's ability to access the chips it needs if we were to lose access to taiwan's chip production and so the U.S government and U.S companies are beginning to think about ways to disentangle themselves from this dilemma the chips Act was funded last year with the aim of boosting ship making the United States which is a uh a worthy goal it's something that I think is important but I think we should be realistic uh about uh how difficult it will be to begin to disentangle ourselves from Manufacturing in Taiwan and assembly in China the reality is that the US economy and aggregate including microelectronics at the US government relies on is deeply deeply intertwined with both Taiwan and China and the process of disentangling will cost many billions of dollars over I think at least a decade to find ways to do that and it's not so simple to say let's decouple as voices in Congress occasionally advise you can do that but the price tag will be in hundreds of billions of dollars over the next decade a price tag that so far as I can tell American consumers thus far are unwilling to bear so the challenge we face right now is to find ways to to address the security risks in microelectronics both Economic Security the prosperity that ships make possible and also more traditional forms of security the chips that our government systems intelligence and Military systems rely on but to do so in a way that is actually viable in terms of how much it will cost and that requires I think a strategy for Selective decoupling electronics that we're just beginning to understand right now we've seen U.S firms in the US government begin to think through what would it require to have non-china Supply chains for critical microelectronics and what we find in return is that it's far more difficult than we would like it's far more difficult than we would like because there are chinese-based components in almost every type of system and as the manufactured goods we acquire especially Electronics get more and more complex it's harder and harder to guarantee that one component won't be sourced from China and particularly given China's Central role in the production of the world's Electronics today there are many types of goods where it's hard to find non-china sources of supplies so simply advising decoupling is uh without being willing to pay a very expensive price tag I think is not a viable strategy and instead we need much more targeted approaches to identify what's the threat that we face and what's a realistic Pathway to finding ways around a Reliance on on Taiwan and China there are alternatives there's certainly chip making in the United States and in more secure locations like Korea Japan and Europe there's assembly facilities outside of China although I'm not sure Vietnam is that much more secure if you're worried about military escalation scenarios and so the fact that the world's smartphone manufacturers are moving their smartphone production from China to Vietnam strikes me as perhaps missing a glance at the map but I think the the the reality of of this assembly dilemma show is just how complex any sort of decoupling strategy will be and so as we try to think our way through this challenge the reality we've become far too dependent on ship making in Taiwan Factory dependent on assembly of electronics Goods in China and that the price tag for restructuring Supply chains will reach to the many tens of billions of dollars and probably hundreds of billions of dollars over the coming years uh clear strategic thinking about how we do this in a cost-effective and a politically viable way I think is absolutely Central to addressing the semiconductor challenges that we face in addition to that I think there's one final point to underscore certainly there are extraordinary risks involved and we've seen the US government focus on risks a lot over the past several years risks that have been addressed by export controls risks that are being addressed by efforts to restructure Supply chains but next to the risk we've also got to think about the reality that American prosperity and security have been primarily guaranteed not by risk mitigation but by the fact that Moore's law has raced ahead and given us a tremendous advantage over our adversaries the reason that the US is so far ahead in semiconductor Technologies in collaboration with Taiwan and Japan and Europe is not primarily because we've been really good at stopping adversaries from learning our secrets but because The Cutting Edge races forward so rapidly that it's just extraordinarily difficult for adversaries to catch up the fact that Moore's law has continued uh with very few changes over uh Now 60 years is uh is a great benefit not only the US economy but also to the U.S strategic position and the fact that the innovations that have driven more as law have happened largely in the United States have have made possible many of the advantages and intelligence capabilities and Military capabilities that I've described and so next to the agenda of risk mitigation we've also got to add a second more important point which is that fundamentally we need to make sure that Moore's Law survives and continues uh well uh into this Century because U.S Prosperity over the last half century has depended on it U.S security has depended not and I think looking forward U.S strategy depends not only on us staying ahead of China but of us staying ahead of China because we're able to harness the advances that Moore's Law produces and so as we think about the the risks and the opportunities associated with microelectronics and some my conductors I think next to the desire to hold back Chinese advances we've got to focus just as much and probably far more on the effort to make sure we've got an environment that is conducive to ensuring that the next generation of some other Technologies continue to race ahead in the United States so with that I'll wrap up my comments but we welcome any thoughts or questions that you have thank you

Channel: NDU CIC

Views: 29,625

Keywords: cyber, information, dod, education, ndu, cic, chip war, chris miller, microchip, china

Id: JrOiI2b-mA4

Length: 33min 52sec (2032 seconds)

Published: Tue Feb 28 2023