Silicon Triangle: H.-S. Philip Wong on the Implications of Technology Trends in the Semiconductor Industry

>> Kharis Templeman: What I'm getting from you is that no one company dominates the entire supply chain. It's just impossible for that to be the case.

>> Philip Wong: No one country dominates either.

>> Kharis Templeman: Yeah, well, that was my other question. So would it be, in theory, possible for one country to develop control of the entire supply chain from start to finish and produce a cutting edge chip?

 

>> Kharis Templeman: Well, hello everyone, I am Kharis Templeman, I'm a research fellow here at the Hoover Institution at Stanford University. And I'm joined today by Professor Philip Wong. Professor Wong is the Willard R and Inez Kerr Bell professor in the School of Engineering at Stanford University. He has been here since 2004 as a professor of electrical engineering, prior to that, from 1988 to 2004, he was with IBM in the TJ Watson Research Center.

And most important for today's conversation, from 2018 to 2020, he was on leave from Stanford. And was the vice president of corporate research at TSMC, the largest semiconductor pure fab foundry in the world. And, since 2020, he remains in a consulting advisory role with TSMC. Today we're gonna talk about a chapter that he and another engineering professor, Jim Plummer, have co-authored.

That's part of our semiconductor working group report. The title of that chapter is Implications of Technology Trends in the Semiconductor Industry. Thanks again for joining me today, Phil.

>> Philip Wong: Happy to be here.

>> Kharis Templeman: Yeah, so I just wanted to start off today's conversation by noting that you're probably the foremost expert in our group on the actual engineering process of semiconductor manufacturing.

And so I wanted to start today by talking a little bit about what semiconductors are. I'd imagine not all of our audience is an expert in this area, I'm certainly not. And so when we think about semiconductor chips, they're not all just one thing. The report chapter that you wrote lays out that there's logic, memory, and discrete, analog, and other, or DAOs.

So how do these chips differ, what are the different things that they do?

>> Philip Wong: Okay, well, thank you, Kharis, first of all, thank you for the opportunity to chat with you. The first question, what is semiconductors? First, and recently I saw some comments about why semiconductor, why not a full conductor?

 

>> Philip Wong: Isn't a full conductor better than semiconductor? Okay, so semiconductors refers to a class of material that is somewhat conducting, that's what it was semi. And you can actually use electrical means to turn that into a conducting state, or you can use the same means to turn this into a non-conducting state.

And that is similar to flipping a switch in, let's say, your household light switch, you can turn it on, turn it off. And the switch is, basically, the heart of all the computation that's happening in computers, in your cell phones, in all kinds of things. Or even, you wanna roll up the windows of a car, you turn on a switch, not a physical switch, but an electrical switch.

So that's, basically, what semiconductors are about, it's talking about a shorthand for electronic devices that are really tiny and that could serve as a switch. And that could also become turned into devices that can store information, store ones and zeros, that's memory devices. They can also be used for many other things, and I'll go on to talk about that.

So as far as where semiconductors are used, basically, anything that requires electricity, your things that you plug into the wall with a plug in a socket in the wall. Or some thing they used a battery to power it up, batteries for cell phones, and batteries for cars, right, cars run on batteries these days.

So anything that uses electricity probably has a chip in there. So that means that, well, we use electricity everywhere, from things like refrigerators, or the things that move, like cars. Or things you put in the pocket, so basically, everything that we use, so semiconductors are basically everywhere. And to take it into the semiconductor products, you can roughly divide them into several types.

One is what we call logic devices, namely, things that you do to compute. And the other one is memory devices, things that you use to store information. And there are actually two classes, one that stores information somewhat temporarily, and the other one that stores information for a longer time.

You can think of it as long-term memory or short-term memory, right? And the third kind is a little bit diverse, and it covers a wide range of things, from devices that can handle very high power or high voltages. Those that you use, for example, in power systems that transmit and store power, energy.

Or the sensors and actuators, sensors that senses either light or temperature or your motion, those are sensors. And also actuators that actuate things that turn, things like motors and things like that. And of course, image sensor is one prime example of sensors. We're doing this video conferencing here that uses an image sensor, which, basically, is a semiconductor chip.

And from the industry point of view, we kind of lump all these eclectic chips together into what we call specialty technologies, specialty semiconductors. So basically, three kinds, logic that do computation, and memory that stores information, because you need information to compute. And the specialty, which is an eclectic collection of everything else, basically.

 

>> Kharis Templeman: Right, okay, so thanks for that primer, let me now ask. Are there different companies that have the lead or that have dominant market share in those different types of chips, or is TSMC the dominant producer of all of those?

>> Philip Wong: Yes, so let me just preface it by saying that I'm speaking today as a professor from Stanford, not endorsing any company in particular.

But since you asked about is there any companies that is kind of dominant in those field, and then I don't think I'm seeing anything confidential or anything, you can find them in public available documents. So in terms of logic, clearly probably see in the news that companies like Foundries.

Foundries are basically companies that produce these semiconductor chips, actually manufacture those semiconductor chips that TSMC in Taiwan is the market leader right now with a very large market share. And the other leading companies are Intel and Samsung. Intel is based in the US, Samsung is based in South Korea.

In terms of memory chips, there's a variety of companies. Korean companies like SK Hynix and Samsung have a lot of market shares. And as well as Micron technologies in the US and also in Japan, this Kioxia, which formerly was from Toshiba and Kioxia, collaborates jointly with Western Digital in the US.

So those are kind of major companies that are in memory and logic space. In terms of specialty, because it's a really plastic collection of companies. There's just a variety of companies everywhere. For example, for analog kind of devices, and there is Analog Devices in Massachusetts, there is Texas Instruments in Texas.

And of course, TSMC produce a lot of these specialty technologies as well. And of course, I forgot about in the logic space, of course, GlobalFoundries in the US also is a player, and GlobalFoundries also play a lot into these specialty technologies as well. So it's kind of roughly the landscape, if I miss any particular companies, my apologies ahead of time.

Roughly this landscape.

>> Kharis Templeman: Okay, great. If I understand you correctly, you were describing companies that manufacture the chips. But as our audience may or may not know, much of the work of designing chips is done by different firms. So could you talk a little bit about the difference between design houses and fabs or foundries?

 

>> Philip Wong: Yes, Kharis, I'm really glad you point this out because this is one of the major source of confusion in the media and also in public discussions. When I read a news article or listen to news forum, they often say chip companies, and they'll say, okay, chip companies XYZ to produces chips.

But oftentimes they didn't make a clear distinction about the design of the chip versus the manufacturing of the chip. Now, oftentimes in the media, people will say, chip companies such as Nvidia and Qualcomm and AMD and so on make these chips. Now, I would actually argue with them, they did not make the chips they designed the chips.

So oftentimes you read, chip makers like Nvidia and Qualcomm. And those are a great source of confusion because they are not chip makers, they are chip designers, they don't make the chips. It is Intel, GlobalFoundry, Samsung Foundry, TSMC, those make the chips. So we have to make that decision very clear, because that creates an impression that American companies such as Nvidia, AMD and Qualcomm and so on are leaders in chip making, which is a little bit of confusion, and that is not appropriate.

So to answer your question, the chip designers, there are actually two kinds of chip designers. One is the chip designing companies that make market-facing, customer-facing products. And you can include in them Apple, for example, they design chips that go into cell phones, computers and so on. And as a consumer, you could buy them and so on.

And same thing with Nvidia, they make graphics chips, you can plug them into your computer and play games with it and fancy video games. And those are products that you can buy directly from them. And there are fabulous chip makers, well, not chip makers, chip designers that design chips, but then they go into someone else's system.

For example, Qualcomm, they design chips that goes into the cell phone. Qualcomm does not make the cell phone themselves. They provide chips for cell phone manufacturers to use the chip, and then they go into the cell phone that people buy. So there's this little bit of distinction between two kinds of chip designers also.

And the companies are very like Apple and even Tesla, they are making cars, and Tesla designed their own chips. And so those are companies, will designed their own chips for their own system, this we call industry, they call them system companies, they built their systems. And then there are the other chip, fabulous chip designing companies that design chips that go into systems.

Companies such as Qualcomm, Broadcom, many of those companies that are designing these chips for people to put into a actual system.

>> Kharis Templeman: It didn't always used to be the case that these two activities were separate. And in fact, Intel, from what I understand, still designed some of the chips that they manufacture.

But the large majority now of chips are designed by one company and manufactured by another, so why the separation? What are the business advantages of doing the designing in one place and just specializing in that versus the manufacturing and specializing in that?

>> Philip Wong: Absolutely, going back in time, at the beginning of the semiconductor industry, everybody is doing everything in the same place.

Including at the earlier times, before my time, companies like IBM would make their own semiconductor manufacturing machines. And then they would make the machines, the machines would make the chips. The chips would go into the computers, the big ion computers that system 360 in the 60s and 70s.

And so companies would do everything from top to bottom. As time goes on, companies realize that they are not very good at doing everything, they probably are very good at doing something, but not everything. So then the industry started to differentiate. There will be companies who are very good at making equipment for making chips, there are companies who are good at making chips itself.

There are companies who design the chips, and there are companies who actually put the chips into our systems. And various companies would pick various combinations that they are good at to do that. And so for companies that they would design and make the chips themselves, they would call them.

Sometimes you see something called integrated device manufacturer, the IDMs. And those classes of companies like intel, and previously, IBM does that, and test its instruments for them. For example, for their analog products, they do that themselves, and analog devices do it the same way. So those are the IDMs, Integrated Device Manufacturers.

And then somewhere in the 80s, people realized that from running the business point of view, more efficient to separate the design and the manufacturing of the chip. And there rise what they call the foundry model, mainly, your companies will be specializing on the design. There's somebody who specializes on the manufacturing and therefore deriving larger efficiency in running the companies and running the business.

So that's the foundry fabulous splits that occurred around early to mid 80s. And a prime example of that is the founding of the TSMC in Taiwan by Morris Chang. Who used to work at Texas Instruments, which is an integrated device manufacturer.

>> Kharis Templeman: Right, thanks for that explanation, I hope our audience could follow that.

I think you did a very good job explaining the difference there. I wanna touch on one more distinction before I shift direction. And that is the difference between what are sometimes referred to as leading edge chips or kind of the newest design for chips and mature or legacy chips.

Can you tell us, or walk us through a little bit what the difference is between those two categories?

>> Philip Wong: Yeah, very good, and let me explain the difference between leading edge chip and legacy chip. First, for the legacy chip, there are two branches that I need to explain because it often get conflated.

And therefore the discussion about legacy good chip get mixed up and that's not good. The difference is, for leading edge chip, that means that you are in the leading edge of the technology development and research. So this is the newest product, the fastest, the most energy efficient computing chips that you can find.

For example, you buy a phone, and this year you have a phone. And next year you have a newer model of the phone that has longer battery life, can do more things because they have the newest generation of chips. So those system products will often drive the development of these leading edge products.

And those are the most advanced, most expensive products, semiconductor chip products that you can find. And so that's the leading edge chip. And then you can oftentimes hear kind of names such as 5 nanometer chips, 3 nanometer chips, and so on, so forth. And those are just labels for these technology products, such as you've got a car of model five, model three, model two, and so on and so forth.

So you can think of these names as kind of models, right? And generally speaking, the smaller the number, the better. And that's a long history about that, and we'll go into that sometime later when people are interested. To understand that they are identified by numbers, numbers of nanometers, and then this model seems generally the better, okay, so that's the leading edge chip.

So the legacy chip means that, okay, so now you have a semiconductor foundry or some integrated device manufacturer that produce a generation of products. And of course then time moves on. A few years later, that particular product that you develop is getting old because there are other newer generations that performs better and higher energy efficiency and so on.

But the flip side is these older products are cheaper to manufacture, now, because we already know how to do this. All the equipment that you use to fabricate the chip or manufacture the chip are all depreciated, so you don't have to pay for depreciation. You're just printing money right now, so those are what we call the legacy chip.

The older generations of technology, the mature what sometimes you can see mature node. Node is a name, a technical name, for these technology generations, so you can say a mature node or legacy chips. Now, among those legacy chips, there are actually two very important distinctions. One is simply older generations of logic chips or memory chips that is just past their prime but still useful they're still useful.

And you use it because you don't need the performance, you need to only the energy efficiency and you just want lower cost. And that's one of the main reasons why they're used. Second, would be that oftentimes manufacturer will say, okay, now I know how to make these chips.

I can use a similar technology as a base and develop other functionality that add on to the original purpose, which is computing. And we have many examples of that, and that gives them special capability, such as the image sensor that we have. We're recording our video on the image sensor.

They are done on this mature, what they call mature nodes, but they are not just simple, no change technology. One actually spend a lot of time developing new, adding on new things to it to make it, to be able to capture image much better than before. Or handle higher voltages or make a response to high frequency signals for RF or communications and so on.

So people spend a lot of time and resources and energy and spend a lot of research dollars to make that, do these other things. But they are based on a previous generation of technology, why are they based on a previous generation of technology? Because then you're starting from a cheaper base, you don't start with really a fancy, costly platform and start with lower cost platform.

And so that's kind of two different when people refer to mature node or legacy nodes, oftentimes they don't make a distinction like that. But this distinction is really important because you do spend a lot of time and energy to develop these other special capabilities.

>> Kharis Templeman: Right, so I wanna shift direction a little bit and talk about the kind of evolution of the market for semiconductor chips.

And in particular how TSMC, as a pure Fab foundry company, developed what is now a dominance in the leading edge chips, as opposed to legacy chips. Intel, from what I understand, used to hold this position until not that long ago. We're talking, within the last decade, and then they fell back.

So what happened? What was the secret sauce to TSMC's success, or what kind of missteps? Flipping the question around, did intel take to lose this race over the last decade?

>> Philip Wong: Yeah, okay, great question. I don't have a lot of insight into how companies actually do their work and so on.

But what I could say is that, which is the kind of technology leader or market leader at any one point, is a very temporary situation. Basically, the leadership is fleeting, basically. We have right now in the logic space, three major companies competing for the technology leadership in the logic, advanced logic that namely Intel, Samsung and TSMC.

And I would characterize it as running neck to neck, and one of them is slightly ahead of the other ones. And I would say at any one point in time, if any company do something really fantastic, they could move forward and become the leader. Or flip side, somebody who makes some mistakes or makes some wrong calculations, they could fall behind.

So this is not, I would say, it's not a permanent situation, it is a temporary situation. And just like in a horse race, you can be ahead, you could be slower at any point in time. There's no end point in this race either, I should say that.

>> Kharis Templeman: Right, so that actually raises an interesting question about China, which we haven't talked about yet.

So, as you know, the PRC is pouring a lot of resources into trying to develop its own domestic semiconductor manufacturing capabilities. On the order of hundreds of billions of dollars of subsidies, with the hope that maybe they can not only join this race, but eventually pull ahead in the race.

What do you think of the prospects of that approach?

>> Philip Wong: Well, first of all, China has already joined the race many years ago, right? They started their own logic foundry, specifically SMIC and several other smaller foundries as well. They also started some memory companies, like CXMT for DRAM for making memory devices, and also YMTC to make NAND flash, the long term data source devices.

They're already in the race, and they started a little bit later than the rest of the world. And so as homing, kind of staying on the analogy of a race, you have a marathon, and somebody started first and somebody started later. So I would imagine it would take a little bit of time for somebody that started later to catch up to the rest of the world and how much, as far as how much time it would take.

It's really a strong function of a number of a variety of functions of parameters, availability of talent, availability of capital, management systems and so on. It's really hard to say. And in this race, they are clearly behind the rest of the world right now, as far as what kind of things they could do to accelerate in their progress is really a variety of things.

Putting in capital or investment is one thing, but it's not the only thing.

>> Kharis Templeman: Okay, do you think, going back to this definite, this distinction between legacy and cutting edge chips. What we've heard and seen a lot about is China's potential for dominating some of the older technology, some of the legacy chips.

In part because that's less profitable and the kind of leading horses in the race, to continue this analogy, aren't paying too much attention to it or aren't devoting a lot of resources to that. But those chips are still very, very important, they're included in many of the electronics, the products that we use on a daily basis.

So do you share that concern? Do you think there's a chance that PRC companies could move in and dominate a particular part of the chip space, even if it's not a leading edge chip?

>> Philip Wong: I think it is very well possible, and this scenario is basically a business decision.

It's not so much a technology, not gated by technology, because those older technologies, the ways to manufacture them, and the way to design and manufacture are pretty well known. And so there's not really anything that prevents any particular country or regions, in fact, is not limited to China.

It could be in one day, maybe Vietnam or whatever, Brazil could one day do this, and as well. So it is not a matter of technology capability, is a matter of business situation. So if they could, if one country or one region could create a business environment that allow them to dominate the market, I can see, totally see that happening.

 

>> Kharis Templeman: Let's talk a little bit about the supply chain that goes into manufacturing chips. We sometimes see in the papers or in comments from people who don't know this industry well, that chips are the new oil. They're geopolitically important. You can seize chips, and then you dominate the leading technology that will build the economy of the future.

What do you think of that analogy? Are chips the new oil, or is there something more to this?

>> Philip Wong: There's definitely a lot more to that. When people say chips are new oil, I agree with that notion that it is as important as oil. In the sense of oil is a source of energy that the entire world depends on and shifts, runs basically the entire world.

From your rolling up your windows in the car to doing the most advanced computation for AI and for chips for 5G and those in that sense, yes. At the same time, chips are very different from oil. And you can maybe I talk about in some of my advocacy for chips also, which is chips needs to continually advance in order to show its value, in order to provide its value to society and oil.

They have been in the ground for millions of years and they never change. Although one could argue that the technology to explore the oil do change, but the oil itself does not change. And so the chips, for example, you can stockpile a lot of oil, put a stockpile in there, and in case of emergency, you can just draw upon the stockpile, which the US does also and many countries do.

But to stockpile chips is only effective to a certain extent because technology will get old. Would you like to stockpile your cell phone so that you don't have to buy cell phone for another ten years? You will not, because the technology advances, these chips can do more things into the future.

So you cannot stockpile chips to a certain extent. You can stockpile for a little while, that works, which is what China is doing right now. They're stockpiling a lot of chips in anticipation of more restrictions and export control and so on. But it could only work for us to a certain extent.

So that's the real difference, clear difference between oil and chips. You need to constantly renew it and therefore, the research and development of chips is just as important as the manufacturing of the chip because of this particular attributes. Because if you move a fabrication facility, let's say in the US, you only have the capability to run to fabricate this particular kinds of chip, and you have no way to provide future chips.

 

>> Kharis Templeman: Right, so I'd like to segue into talking a little bit about the parts of the manufacturing process. You've noted there's design, development, fabrication, and then testing and assembly, and all of those may be done in different places by different companies. So are there particular parts of that supply chain that are especially vulnerable to disruption or that are, especially on the other hand, resilient and very diffused around the world that are easy to enter versus not?

Could you walk us through a little bit kind of the economics of the chip supply chain?

>> Philip Wong: Yeah, so let's talk about supply chain itself first. The chain consists of a long, really long chain. Starting from the basic material you need to mine the material, you need to purify it, and the materials itself, and also the materials that you use to fabricate the chips, chemicals and gases and so on.

So starting from that, you got materials, and then you have manufacturing equipment that you use to produce a chip. Taking analogy of a kitchen, you need to go to the market to buy the material. You need to have a kitchen that have a stove. There's oven and pots and pans, those are the equipment.

And then you need to have companies that develop the technology. They need a chef that could come up with a recipe, and then you need to manufacture those chips. And that's the manufacturing aspects of it. And keeping on the analogy of a restaurant or a kitchen, you have a chef who come up with a recipe and nice dishes, but then you have a franchise restaurant.

Then you need to have the manufacturer of those food, a new burger or whatever. And then after the chips, and then you have the designing of the chip. How do you configure the chip? You wanna have a burger that has a tomato on the top or tomato on the bottom.

That's different. So the design of the chip, and then you have the system companies who use the chips do something useful. So this is an entire chain. And along the chain, there are many companies that provides the tools to do things, tools to manufacture chips, tools to design the chips.

And those are all part of the chain. So now when we talk about a supply chain, that means that if one link of the chain is broken, you already have a broken supply chain. Doesn't matter if the rest of the chain is intact or continuous. You just need one point, it's broken, then you cannot make anything.

So in other words, any particular part of the supply chain is broken, you don't have a complete supply chain, and you have a big problem. So that's part, as far as the various pieces of the supply chain. And you probably see in many discussions in the public media, that some countries or regions are really good at doing certain things.

For example, in Japan, they're very good at materials, and also semiconductor equipment. In the US, the major semiconductor, there are many major semiconductor equipment, manufacturing equipment companies, also electronic design automation tools to help design chips in a very efficient way. They're all kind of headquarters and based in the US.

And also many of the system companies who put together chips and design chips and also put together systems are based in the US. Many of them are in Bay Area here, right near Stanford. So those are the kind of strength that various regions would have. For example, in the Netherlands, they have a very dominant company that makes what they call the thography machine, who is part of the manufacturing process of the chips.

And in Korea and in Taiwan, they are very strong in manufacturing those chips. In Korea, they manufacture memory chips. In Taiwan, they manufacture logic chips. There's this supply chain and globally, there are regions that are good at it. But it doesn't mean that once you deleted that region out of the picture, then you're quickly in trouble.

You could be in trouble temporarily on a short term basis. But I think if any supply chain is disrupted, then in most cases, the rest of the business community will come together and fill in those spaces, because then if somebody is deleted out of it, you created a vacuum.

And over time, now the question is over what time? So some will take longer, some will take shorter, and over time, that space will be filled in.

>> Kharis Templeman: All right, so what I'm getting from you is that no one company dominates the entire supply chain.

>> Philip Wong: No.

>> Kharis Templeman: It's just impossible for that to be the case.

 

>> Philip Wong: No one country dominates either.

>> Kharis Templeman: Yeah, well, that was my other question. So, would it be, in theory, possible for one country to develop control of the entire supply chain from start to finish and produce a cutting edge chip?

>> Philip Wong: I would argue that, no, the supply chain is so diverse and so complicated that no one country or region is big enough and robust enough to be completely self contained.

 

>> Kharis Templeman: Okay, so Follow up question on that, then we sometimes see in the reporting on Taiwan cross strait relations with the PRC. And TSMC's central position in the chip supply chain, that after PRC were ever to take over Taiwan and control TSMC itself, that would give them dominance in the manufacturer, at least, of chips.

Does that follow, or does the supply chain that TSMC relies on, is that too complex to be replicated just within a PRC Taiwan kind of ecosystem?

>> Philip Wong: I think that scenario that you described, which probably is described by many in the discussion, is entirely impossible, because, for several reasons, one is for the reason that you just described.

The supply chain so complex that there's any particular company would rely on the rest of the supply chain to be completely operational. And so if the rest of the supply chain is absent or not cooperating with that particular company or that region, the company's not gonna be able to operate.

That's the first question, that's the first kind of degree. Now, a further consideration, which most discussions like that did not go into, is that is the level of trust that is required in this business. For example, in a Foundry, you would have many, many thousands of customers, and many of the customers are competing against each other.

And the way a Foundry work is that they need to work with all the customers, many of them compete with each other. So what does that mean? That means that the customers have to really trust the Foundry in collaborating with them in terms of product roadmap. In terms of new inventions that maybe, they are still cooking within the company, but they're not willing to disclose to other people, especially their competitors.

So Foundry, they really needs to be have this really high degree of trust between the foundry and the people who use the foundry. And so if an entity is not no longer trusted, it would not allow them to be a viable Foundry.

>> Kharis Templeman: I'm really glad you said that word, and it's something that we've been emphasizing a lot in our discussions around this report, that a trust is such a key part of this industry.

And I'm tempted to quote from the late George Shultz, the former secretary of state, who was a distinguished senior fellow here at Hoover, his favorite catchphrase was, trust is the coin of the realm. And so I wanna pivot a little bit in the last couple of minutes we have here to talk about the future of the semiconductor industry.

And to talk about how this future might be developed, who's going to be involved in it? And I wanna ask you, first off, is Moore's law dead? So the idea that we're gonna have a doubling of computing power every two years or so, we're gonna see this continual shrinking of what you can put on what can be fit on a single chip.

Is that process played out, and if so, where do we go now from here on out?

>> Philip Wong: Great question. Well, this is a cliche now, but it is really a wonderful question, because this topic has been discussed and there has been so much confusion about what actually people mean by Moore's law.

And whether Moore's law is dead or lie, to me it doesn't matter, and it is really irrelevant. Okay, let me explain. In the past, maybe like 50 years or so, we have seen a tremendous improvement in semiconductor technologies, both in terms of the capability and the amount of things that it does.

And also energy efficiency and speed and so on, that we have reaped a lot of benefits from these advances in technology. Mostly driven by the need from society, because new capability and new energy efficiencies are desired in many of the products that companies produce. So I expect that to continue to happen, and in fact, not only to continue, but even accelerate, because we have been so now so dependent on continual advances of our technology.

And many of the technologies dependent are founded upon semiconductor technology, we talk about quantum computing, we talk about 5G, 6G. We talk about AI, those all need semiconductor technologies to make their advances. So I expect the applications and the uses of semiconductor technology and therefore the impact of semiconductor technology in society to actually grow faster than before.

And so there's a great need, and as you know, when there's a great need, innovation comes, when there's a great need, innovation comes, and there will be business opportunities. So I think those will not only continue on a path before, but actually accelerate because of the great need and great benefit that comes from it.

Now to the next question of how do you get those advances? In the past 50 years, we would maybe draw an analogy, we would like walking inside a tunnel. The way that simulator technology achieved its advances was by shrinking the transistors or the semiconductors from bigger ones to smaller ones.

And that path has been so successful in the past 50 years that we are like walking inside a tunnel. That's the only thing you need to do, going forward, we know exactly what to do, no other methods or innovations required, just over there. Now, we are at the exit of the tunnel with this path, just not everything that we do, any tool, any techniques, always saturate.

And we reach kind of the end point of using this methodology where at the exit of the tunnel, now, there are two ways to interpret this. At the exit of the tunnel, you fall off the cliff and you're dead, you make no progress. Another interpretation is, at the exit of the tunnel, you're not bound by the tunnel, you can go anywhere you want.

And I think that's the excitement in semiconductor technology right now, we are at the exit of the tunnel, there are many ways to go because there is a strong need and innovation will come. And those innovation are not confined by what you can do in the past 50 years.

 

>> Kharis Templeman: Okay, so I'm gonna hope, I'm gonna assume we're in the second of those scenarios where you can go any direction and we've got this boundless set of possibilities. But one of the things we also know about innovation is that there are benefits to cooperation, right? To working with partners across the world, and the more minds looking at a difficult problem, the better.

Especially the more diversity of backgrounds and approaches that you bring into looking at a problem. So, is there any possibility that we could build a consortium or a collective that works on this problem that crosses a lot of these national boundaries? But that still maintains the basic level of trust that you mentioned before?

 

>> Philip Wong: Absolutely, I'm glad you mentioned that, because as I said before, manufacture of chips depends on continual improvement. And continual improvement requires research and development. Now, what makes good research and development, and that is the collaboration across different regions, different expertise, is clearly important. No one university is best in research, and you need many other universities to advance knowledge.

And that's the same way with semiconductors, and you need many participants across the world, across the region. And we have seen many good examples of such collaborations happens really well. For example, within the US, actually, this is actually outside the US as well. Somewhere like the Research Corporation, which pull money from industry and fund university research, has been operating for the past, maybe 20, 30 years.

And as a model for a well defined objective and making a lot of progress in semiconductors, so those are one example. Another example, the Inter University Microelectronics Center, or IMEC in Belgium. Those are a big consortium of industry that do advance development together as a collection as a consortium.

So those are very successful examples of international collaborations. And I believe that if the world will come together and coordinate and collaborate on the further R&D of semiconductors, we will be advancing this technology much faster than before.

>> Kharis Templeman: Great, on that note, Philip, I think we're just about out of time, so I'll bring our conversation to a close.

Thanks again for speaking with me, to our audience, I'll just repeat, I'm Kharis Templeman. I'm a research fellow here at the Hoover Institution at Stanford University, and I've been speaking with Philip Wong, a professor of engineering at Stanford as well. And we've been talking about his co authored chapter with Jim Plummer entitled Implications of Technology Trends in the semiconductor industry.

Thanks for listening, and we encourage you to check out the report of which this chapter is a part, entitled The Silicon Triangle, the United States, Taiwan, China, and Semiconductor Security.

>> Philip Wong: Thank you for the opportunity, thank you.

>> Kharis Templeman: I'm Kharis Templeman, Silicon Triangle is a special podcast series of Matters of Policy and Politics.

 

>> Female Speaker: This podcast is a production of the Hoover Institution, where we generate and promote ideas advancing freedom. For more information about our work, to hear more of our podcasts, or view our video content, please visit hoover.org