In this latest episode, we will learn about one of the core industries that fuels Israel’s high-tech sector: The semiconductor industry. The semiconductor industry is responsible for 450 billion dollars in direct annual revenues, which is roughly ten percent higher than Israel’s entire GDP. This number still pales in comparison to the real value that this industry generates for the global economy, as semiconductors are the backbone of digitization.
This is the main reason for the concern around the significant shortage in the supply of chips across various end markets, ranging from automotive to smartphones. For the sake of illustration, let’s take an example of an automotive company that has been unable to purchase one wafer with chips on it that it needs in order to assemble cars. The inability to purchase this one wafer can be seen as $5,000-$10,000 loss in semiconductors sales, but could translate into roughly $1.7 million dollars and more in car sales. So, the value loss multiples here are in the hundreds.
To discuss the ongoing demand and supply imbalance in the semiconductor industry and understand how Israeli start-ups could capture a piece of the semiconductor value pool, we couldn’t have asked for more appropriate guests on this topic: Yaniv Garty, the vice president of Intel and the general manager (GM) of Intel in Israel, and Dr. Ondrej Burkacky, who is a McKinsey senior partner based in Germany and leads McKinsey’s global semiconductor work.
Ondrej, Yaniv, it’s a pleasure having you both today. I’m delighted and excited to have you both. And why won’t we start from a quick introduction: Yaniv, please introduce yourself and describe Intel in one sentence.
Yaniv Garty: I have one job: I’m the Intel Israel GM and I also have a global job in the field of 5G. And Intel in a sentence: we all live in a digital economy and we build the building blocks for that.
Peleg Dekalo: Ondrej, your turn. Introduce yourself and tell us what got you into semiconductors.
Ondrej Burkacky: My name is Ondrej Burkacky. I’m a senior partner at McKinsey, I lead globally our semiconductor sector. And what got me into semiconductors—I always have been a technology geek and I like speed of innovation. And what could be better as an industry than semiconductors, where you can really see innovation cycles in your lifetime—and not one of them, but many of them—happening?
Peleg Dekalo: Thank you Ondrej. To help us dive into the complex topic of semiconductors, could you get us started by quickly walking us through the value chain of the semiconductor industry?
The value chain of the semiconductors industry—brief overview
Ondrej Burkacky: It actually, in many cases, more than 90 percent of the cases, starts with sand. And with sand, what you can do is, you can basically have your kids play with it and construct castles. Or you can basically take it through an advanced-manufacturing process and grow crystals out of it, which is then the foundation for what is called a ‘wafer.’ So, a round, shiny thing made out of silicon you can then build in the structures of your semiconductor device. The first type are companies that actually [make] these silicon wafers. Then, that wafer needs to get transferred into a chip. And for that, you need design capabilities. Somebody needs to design how the logic of the chip looks like, how the structures look like. So, there are companies that do the design of it. Then, there are companies that take the design and translate it onto the wafer to start what is called front-end manufacturing—putting the structures of these chips in place. It’s a very complex manufacturing process. We see several hundreds of steps being done there. Then the next step is of this round typically, a 300-millimeter wafer cut out to individual chips. The so-called back-end processing process put the very typical wiring next to it called wire bonding. So, this is the feet of the chips that you see once you have it ready in your hands. And often enough, there is a shell put around it, which is this black substrate that encloses [the wafer] and protects it. And then it’s basically done. The so-called back-end manufacturing, assembly, and testing when the final test is done. Then the chip is delivered to the next value-chain step, which [is] typically the electronics players that put it on top of a printed circuit board, add additional chips, and out of that, we get to a system that, for example, then helps your fridge to operate, or helps your car to move. It’s basically then what we, as an end customer, realize as being a semiconductor device.
Peleg Dekalo: Thank you Ondrej for this helpful context. Now Yaniv, where does Intel come into play along that just described value chain?
Intel’s unique position in the semiconductors value chain and future plans
Yaniv Garty: There are companies who are the manufacturers themselves—we normally call them ‘fabs.’ That’s the name for the fabrication-manufacturing plant. Many companies design the logic and then deliver that file to the fab to manufacture. Those companies are normally called ‘fabless.’ And then there are a few companies who are called IDM, IDM stands for ‘integrated device manufacturer,’ meaning companies that both design the technology and the silicon and manufacture it. And that’s what we do. We basically hold a major portion of the value chain from the fabs themselves through the process that Ondrej explained, including the testing and manufacturing. We basically deliver the final goods to our partners or channels to then be put into PCs, 5G base stations, cloud servers, or you name it.
Peleg Dekalo: So, you are designing the technology and manufacture according to that design?
Yaniv Garty: Today, we manufacture the vast majority of our products in-house, so we not only design them, we design the technology, design the product, [and] also manufacture them. That’s what makes us an IDM. There’s a portion of our products [that] are manufactured externally. In that case, we play as a fabless company with other fabs.
Peleg Dekalo: OK, so this is where you play in the value chain today. Where else do you plan to play in the future?
Yaniv Garty: One of the changes we have announced is that in addition to us continuing to be an IDM, meaning we’re going to manufacture our own designs, we are also going to open our fabs to other companies and let them manufacture their solutions, their products on our process, and they have the option to use [whichever] of the many of technologies we offer from all of it to none, and still manufacture within Intel’s fabs.
Peleg Dekalo: Which is something that does not exist today in Intel.
Yaniv Garty: It doesn’t exist today, correct. We never really opened our fabs to others.
Peleg Dekalo: Now that we have a better view of the semiconductor value chain and Intel’s unique position in it, let’s talk about the ongoing global chip shortage that everyone seems to be talking about.
Ondrej, what caused the shortage, where do we stand now, and is there a prospect for recovery for when we’ll get out of it?
The global shortage in chip supply—causes and forecast for recovery
Ondrej Burkacky: First of all, I think one needs to understand that the semiconductor industry itself has been for a very long time actually for almost all of its history, a very volatile industry when it comes to supply-demand matches. And this is what typically led to these so-called cycles, right? Where there was too-low capacity, then you had a too-high capacity with the surprise as well floating around. Now what has changed in the last decade is that the industry itself has become more stable. It has grown there has been certain consolidation in the industry and with that the demand-supply match was matching on an industry level much better than previously. And with that, we had already an industry which has very high utilization of its assets [that] were planned well in respondence to the demand. Then came the COVID-19 situation in 2020, and what we have observed is digitalization on steroids. We went into home offices, used a lot more electronics equipment because we needed cameras, we needed microphones, we needed perhaps a new PC to work from home. The internet traffic increased by 40 percent in that period. We were consuming much more digital content that needed a new backbone for the internet, new gateways, new routers—all this equipment used to it. And still, we are big consumers of smartphones, tablets and so on. So that trend continued. And with this we have basically seen a demand increase in the semicon industry which was more than 50% higher than what was expected in 2020. So instead of a roughly five percentage growth we ended up seven, eight percent increase across the board in semiconductors. And it hit an industry where already the capacity planning for new semiconductor fabs is something which takes minimum three sometimes four years until you get from a starting of constructing such a fab to really at-volume products out. So, this is nothing you can react very fast to a change in your demand. Mainly—and this is the interesting thing about it—the main shortage is actually coming from not the highest and most innovative so-called leading-edge semiconductors in the five, seven, or 14-nanometer space. A lot of the constraint is coming from controllers with more mature feature technology that power our fridges, our other appliances, smart home devices, cars, and so on. And many analysts believe that the shortage is something that’s going to last until 2023.
Peleg Dekalo: You mainly talked about the demand side of things, which from my understanding, it’s at least the vast majority of this crisis is stemming from. But is there something on the supply side that we should know about?
Ondrej Burkacky: When we look at it from some of the things that happened in the supply chain, right, I would diversify it into two categories, right? One category is things that happened, like the plant in Japan that caught on fire, right? The winter storm in Texas, which everybody is talking about, right? There was definitely some impact there, but it’s not that we never had winter storms before, so I would not attribute the supply situation to this singular occurrence and just say, ‘Okay, it was a bad year.’ It was just a normal year, which perhaps a bit more here, but not much more, which was a lot of constraint on the demand side. So that’s one category of it. The second category on the supply-chain disruptions, however, comes more from regional discussions, where there [are] a lot of regional supply constraints coming up from more geopolitical situations where we talk about certain regions of the world want to make sure that they have a semiconductor sovereignty, and others might not. And a lot of constraints on a supply chain that was never designed to be regional. So, the semiconductor supply chain is a truly global supply chain, with few selective leading companies for each supply-chain step because it’s so R&D heavy, never being optimized for being able to produce the entire value chain in one region and supply to that particular region. So, if you look at it from that perspective, there is no region in the world that has it all along the supply chain. Any impediments you put on a supply chain, in terms of regionalization constraints, is a big disrupter, much bigger than a winter storm in Texas.
Yaniv Garty: We have a very similar, to say the least, view of just what Ondrej said. one of the things Intel is synonymous with is Moore’s law, which says that every 18 months, something doubles. It started with the density of silicon or chips.
Peleg Dekalo: You have to give us some background about Moore’s law.
Yaniv Garty: Robert Moore was one of the founders of Intel, and there was a law—which, basically, he’s the first to have stated—that every 18 months, the density of transistors in the chip will double, which looked at the time—that was, I believe, the early ’70s—like science fiction. Not only that, the law was proven to be very precise. It expands to other parts of the digital economy and beyond. And I’m using that reference because up until COVID-19, we, as humanity, have basically doubled the amount of data we have produced every 18 months, meaning that up until 18 months ago, what we have produced over the last 18 months is equal to all the data humanity has produced all over history, and we have used a fraction of that. So, what we have been on over the past few decades, mainly since the introduction of transistors—which I believe, Ondrej, was [what] the three Bell Labs researchers have got a Nobel Prize for in the mid-50s. It’s basically a digital transformation. Our lives are becoming more and more digital. And then COVID-19 hit, and as Ondrej said, everything has gone crazy. That digital transformation we have known and been on for a few decades now and have been growing exponentially has now [gone] crazy, and we saw that with the fact that now we need more laptops. We need more phones, screens, printers. The food that we get, we order digitally. We use more cloud services, a lot of banking, healthcare, and so on and so forth. So, the demand exploded.
Peleg Dekalo: So, your observations on the demand side are right in line with Ondrej’s. How did this then affect the supply side of things?
Yaniv Garty: We have to realize, and I think, what sometimes people kind of overlook, is the amount of silicon involved in each and every product. A very basic car today has about 70 different pieces of silicon, so 70 different chips from the brain of the car to the sensors in your tire. So, every action we have today involves a huge amount of chips, and those are [each] different. Some of those are very simple and some of those are very high end. So, when we look at the supply side of that, another trend that has happened over the past couple of decades is that the number of fabs—those manufacturing plants that can actually produce at the very cutting edge of technology—the top technology for silicon has gone from about 20 years ago to three today. Some may say four, but that’s basically the order of magnitude of number of companies that can actually manufacture the cutting edge of technology. So, you’ve got Intel, you’ve got TSMC [Taiwan Semiconductor Manufacturing Corporation] in Taiwan, you’ve got Samsung in Korea. Also, the cost of fab went from $1 billion to $10 or $15 billion or even more than that. Just to give people a sense, the size of a fab today can be two or three full football fields. Ondrej mentioned equipment. A piece of machinery in a fab cost more than a fighter jet. So, it’s very complex, and to be able in an industry that was very careful about, as Ondrej said, forecasting the demand because the investment of a fab is huge. And if you don’t utilize it, that’s a huge penalty financially to a company. You see many companies, us included, announcing that they are investing more money into building and opening new facilities, new fabs, and you see also governments stepping in. That’s the geopolitical angle of that, which also relates to the fact that if I went back 20 years ago, I think the balance was about 40 percent of silicon manufacturing was on the western hemisphere, and 60 percent on the eastern hemisphere. Today, it’s [split] 20 and 80.
Peleg Dekalo: If you had a crystal ball, what will happen to the demand/supply balance in 2022?
Yaniv Garty: We forecast that we are probably, right now, around the deepest point of the shortage, and we’ll probably start really recovering from that in 18 months or a bit more than that. So, it’s into to 2023.
Peleg Dekalo: After having talked about the semiconductor value chain, and the much talked about demand and supply imbalance, let’s talk about Intel. We’re all familiar with Intel’s strong presence in Israel. What is Intel Israel’s role within the global corporation?
Intel Israel—pillars of activity and contribution to the national economy
Yaniv Garty: Maybe I’ll start with kind of a short version of our ID. So, Intel in Israel started in 1974. We are one of the first multinational corporations that opened up a research center in Israel at the time. From five people in 1974, we grew to about 14,000 people right now in Intel in Israel. In many ways, Intel Israel is a microcosm of the bigger Intel. we have five areas of focus or pillars of activities. The first one is compute. We are focusing on developing everything from the core technology of compute, which we call Intel architecture (IA). A lot of that is developed here in Israel. Then, we use that together with other solutions and put those into a CPU, but the CPU is much more than just a computer device. It has interconnects, it has communication capabilities, it has storage inside, and in some cases, graphics. All of that is put into a package, and you start adding the software layers on top of that. So, we do all of that. The second pillar is around communications. So, it’s things like Wi-Fi, Bluetooth, Ethernet, and Thunderbolt. A lot of those are centered here in Israel. On AI, we have anything from what we call ‘AI inside.’ These are data scientists and developers focused on improving our own processes and products and capabilities at Intel through other areas like Mobileye, which is a sister company or a company that Intel has acquired here, which is focusing on autonomous driving, which is basically about AI. And, of course, Habana, which is another Israeli company we acquired lately, is building AI chips. So, these are training and inference chips, which are capable of using or applying AI capabilities in silicon and being put to play in data centers. And the last part on the design side is cyber security. That’s where we use both software and hardware solutions in order to make our products more secured. And our fifth pillar, or part of the strategy, is our manufacturing, and that’s within our center in Kiryat Gat. Today, we manufacture the latest technology of Intel products that either were developed here or developed elsewhere at Intel. Just to color it a bit with financial data, in 2020, our overall contribution here to the local economy was over $8 billion, which stands for two points of the local GDP. So, our presence here is quite big, but we are definitely not alone.
Peleg Dekalo: With contributing two percent of the GDP, you have a significant influence in continuing to shape Israel as a major innovation hub. From your perspective, Yaniv, what is it that makes Israel a leading innovator in high tech? What is our secret sauce?
Yaniv Garty: I think it’s about the balance. As long as we keep a healthy balance, which I think we do, between multinationals or big companies—because some of those multinationals are headquarters also in Israel—but as long as you keep a healthy balance between the multinationals and start-ups. Each has a different way of driving innovation with the different focus areas on what innovation is about and a different kind of end point. But what’s common, and that’s, I think, [is] one of the key characteristics of the ecosystem we have. I hope not many are going to argue with me. It’s not about IQ, because the average IQ is the same every place you go. I think we have more than a good free education system, but more than that, it’s the culture. A culture that cultivates innovation in different ways, shapes, and forms, that you then take into play, whether it’s a big company or a small company, whether it’s a company in the field of silicon or software or in services, that’s the uniqueness of the ecosystem we have here.
Peleg Dekalo: So, you see it mainly as if we put a big umbrella on it, then it’s talent.
Yaniv Garty: It’s definitely the talent. And the fact that we have this mixture, I mentioned before, takes people with knowledge, spirit, and culture, as we said, the talent, and enables them to learn the processes, how to do things best, plus them inserting their own secret sauce into that, and creating a very talked [about] culture. What a good global company does is it finds the sweet spot in combining the local culture with company culture to make the best out of it.
Peleg Dekalo: Yeah, and you basically predicted my next question because we are in the ‘start-up nation.’ So, let’s talk a little bit about the relationships between enterprises like Intel and start-ups that, as you said, bring innovations in different verticals and different shapes and forms. What do you see, Yaniv, as the different ways in which enterprises like Intel can benefit start-ups and create further value for the market as a whole?
The symbiotic relationship of Intel and startups in Israel
Yaniv Garty: There are multiple layers of that. Probably, the most visible one is things like the corporate [venture capitals]. We have Intel Capital, which has been, I believe, the most active one in Israel. Over the years, Intel has acquired companies here at a value of more than $20 billion. So, it’s 20-something different acquisitions. And Intel has invested in dozens of companies here as an investment arm. That’s the way for Intel and other companies—through their venture capital arm or through their M&A arm—to infuse the local market, definitely with capital, help company grow, and help companies have an exit. But that process doesn’t stop the minute a company enters the big corporations, and it’s always a complex challenge. Not all acquisitions end up in a positive way. By the way, I’m an outcome of an acquisition Intel made about 17 years ago.
Peleg Dekalo: Tell us about that.
Yaniv Garty: I was part of a small Israeli company that was named Envara. We were doing Wi-Fi and Intel acquired us because of our underlying technology, which was all of the MA, which we used for Wi-Fi, in order to go at a new technology at the time that was named WiMAX, which doesn’t exist much today, but it’s like a close cousin of LTE or 4G. The fact that you had many people started in start-ups, then went through the schools of how you build solutions in scale, and that view of at scale is so critical for you to actually grow companies beyond a certain stage of start-ups and actually become companies that can deliver solutions, either B2B or B2C, whatever that is, but grow into companies that actually deliver them and are just focus on technology. A lot of that—the ability of the local market to grow—went through the fact that many of the entrepreneurs had a certain period of time spent within companies that [knew] how to do it from before, which are normally the bigger companies. So that’s one angle. The other angle is, for example, we have a growth program [which is] like an accelerator. We call it Intel Ignite. We’re going to start now the fifth round, the fifth batch. Every time, we [accept] ten companies out of about 200 that apply, and we’re looking for those companies that can do a few things for them and for us—those we can actually help most and those that can help us—but this is not about us necessarily using their products. We are looking for those entrepreneurs and people which are trying to tackle big problems from a completely different point of view and are trying to apply different ways of tackling those challenges.
Why should startups be interested in the semiconductor industry?
Peleg Dekalo: Ondrej, let’s finish our discussion with the question of why start-ups would consider entering the semiconductor industry. What would be your elevator pitch to start-ups of that sort?
Ondrej Burkacky: I think that it’s actually a fascinating industry when it comes to speed of getting ideas to the market. R&D speed, culture of really understanding topics, right? Really going technologically deep. And with that, it is just from a pure ROI perspective, right? When does an idea materialize or not? You will know very fast in semiconductors whether it works right? And if it works, you will get to a significant size pretty fast, given just the product life cycles. Because just let us think about it that when you look it from the revenue perspective, more than 50 percent of the semiconductor-industry revenue comes from products that are one to two years old, which is completely different to other industries where a product life cycle is more around seven years or even ten years. So, for significant penetration, it just takes time. And here, time is often precious for start-ups, and this is why the semiconductor industry provides exactly that. When it comes to applications around data, for example, within the semiconductor industry, there is actually zillions of data available already today, so there is less groundwork compared to other industries if you want to apply your technology at a semiconductor company compared to others.
Peleg Dekalo: Gentlemen, unfortunately, like all good things, this podcast has also come to an end it was a real pleasure. I had so much fun sitting here and talking to you. So, thank you very much, Yaniv. Ondrej.
Yaniv Garty: Thank you, Ondrej. Thank you, Peleg.
Ondrej Burkacky: Thank you all. It was really fun.