Afternoon, everybody. My name is Raji Gill. I head up Global Semiconductor and Automotive Technology Research here at Needham & Company. Welcome to the seventh Annual Automotive Technology Conference. We're very pleased to have GlobalFoundries presenting with us for the first time. From GlobalFoundries is Kamal Khouri, who is the Vice President and GM of GlobalFoundries Automotive End Market business. In terms of the format, this will be a fireside chat. For folks that are on the line, if you have questions, there is a chat box. Feel free to send those questions, and I'll relay those messages to
Kamal. Maybe before we could get started, Kamal, maybe a brief background on your history. It's quite impressive, and it'll be great for the audience to have some familiarity with your career.
Yeah. Thanks, Raji, and thanks to Needham for hosting this conference and inviting GlobalFoundries to share our story with you in automotive and beyond. Kamal Khouri, I run the automotive end market for GlobalFoundries. I've been with the company for 3 years and this was my first stint in semiconductor manufacturing. Prior to that, most of my career, over 20 years, has been with Chip design and semiconductor design companies.
Mm.
So typically, like GlobalFoundries customers.
Mm-hmm.
I started with Motorola, a really long time ago, as a design engineer.
Wow!
As a design engineer, and, you know, from there, I moved on to various different roles. I've done, software design, I've done hardware design. I've also moved into product marketing, product management. Shortly after Motorola, I joined Advanced Micro Devices, AMD.
Mm-hmm.
I was there for seven years, working again on many different products, from high-performance gaming machines to medical devices. After AMD, I joined NXP.
Mm
where I ran their ADAS microcontroller division. This was at the height of all the autonomous driving excitement.
Mm-hmm.
I was the general manager for that business, and then, and I joined GlobalFoundries.
Very impressive background. Thank you, Kamal. Maybe we could get started. I'd like to kind of cover 4 major themes during this fireside chat. The, the first one really is the evolving foundry relationship with the automotive OEMs directly. We see a kind of a new model evolving in the semiconductor supply chain, where essentially foundries like yourself are working more on a one-to-one basis with the automotive OEMs directly. Can you talk about why that is happening and maybe elaborate on that a little bit further?
Yeah, yeah, absolutely. It's a very exciting time, and maybe this is a bad analogy. You know, for people of my generation that grew up with hot rods and fast cars, you can think about the seventies and eighties, where car OEMs were differentiating with the technology that they had in their engines.
Mm
... and engine design, right, at the time, and suspension design. And what material you use to build your engine. Is it steel? Is it aluminum? Is it some kind of hybrid component of that? And I think at the time, the OEMs started to realize that in order to build the best, you know, engine technology in the world, they had to manage their supply chain. And I'm sure during that time, relationships were forged with aluminum and steel smelters and factories. And I think, you know, if you take that analogy and you map it to today and you say, well, what is going on in the automotive space?
There's this huge transformation where the car is changing from an electromechanical device that takes you from point A to point B, to a immersive, interactive, yet safe vehicle and experience that still gets you safely from point A to point B. That transformation, which at GlobalFoundries we call ACE, autonomous, connected, electrified, requires a significant investment in software, and that software has to run on semiconductor chips. The recognition from OEMs that if they want to deliver that differentiated experience in the car, they have to understand and manage the semiconductor that's running their
software.
Mm.
Those semiconductors, at the end of the day, are manufactured by companies like GlobalFoundries. Having that relationship with us, it becomes extremely important. Notwithstanding the chip shortage, which I'm sure we'll talk about during the session, putting that aside, I think if you look at that macro trend in automotive, it explains why it's become more and more important for them to have a relationship with Foundries.
Can you give us an example of where you're seeing most focus from the automotive OEMs across the technology nodes? Where is GF's differentiated offerings are able to support these technologies and applications?
Yeah. If you look at this transformation that we talked about, it's actually touching every point in the vehicle, from the semiconductors that control your seat, that move your seat and set the memory, to your air conditioning system, ambient lighting in the car, right? Nowadays, you can kind of set all these really cool colors and moods inside the vehicle to very smart LED lighting for your headlights as well. That's one aspect of it. There's the autonomous driving or safety features in the car, right? Warning you if you're departing from a lane, getting into a, you know, a situation where there's somebody in your blind spot
, emergency braking, all these safety applications all the way to the car driving itself. The transformation to electric drivetrain. EVs, managing the resource of the battery in the car to make sure you get the maximum range while you're running all of these applications in the car. We focus on these three pillars: processing, sensing, and power and data management in the car. We develop technology families that address each one of these sets of solutions. By looking at that holistically, you know, we're able to have a very real and tangible conversation with OEMs and Tier 1s and semiconductors about how to develop
the most differentiating features to address this transformation. Our roadmap is then developed based on that. The technologies we look at that help with those three areas of transformation, are all within the GlobalFoundries portfolio. You know, with the exception, and I think this is well known, with the exception of the one or two very high-performance central computers in the car.
Mm-hmm.
that may have, you know, some high-performance function for things like autonomous driving. But, you know, conservatively, 75%, 80%, if not 90% of other semiconductors in the car are within GlobalFoundries' portfolio.
Let's talk a little bit about General Motors specifically. You signed a kind of first of its kind, long-term agreement with GM in February of this year. Can you perhaps provide some more details around the background of this partnership and some of the key areas of focus for General Motors?
Yeah, absolutely. Like many other OEMs, this journey or transformation that I talked about is something that GM has embarked on in terms of, you know, electrifying their product portfolio. They approached us with three key objectives, right? Number one, they wanted to ensure that they could secure dedicated semiconductor manufacturing capacity for their vehicle roadmap, right? They didn't want to experience the shortage and the production halts that they experienced during the COVID pandemic.
Yeah.
That was the first key requirement. The second requirement that GM wanted was they wanted to ensure that that took place in the United States, that their semiconductors were manufactured in the United States. They had a geographical preference. Thirdly, they wanted to secure the technology corridor of their choice to give them visibility and assurance that what was being manufactured, when it was being manufactured, is making its way into GM vehicles. Those were the requirements. In effect, what we did is we built a dedicated technology corridor.
Mm-hmm.
for GM, the technology that they cared about in the location that they wanted it, and through this partnership, providing the best economics that we can by eliminating a lot of the stack margins that would have otherwise been there had we followed a, call it, traditional supply chain, where we were n degrees away from from the OEM.
Mm-hmm. Interesting. Interesting. Just to follow up on that partnership. You said you're building a dedicated technology corridor for GM. I presume this is in the upstate Malta, New York, facility. Okay.
Yes.
What does the technology corridor, what does that mean? Is there... You know, you manufacture X number of wafers a year. You're gonna allocate a certain number of wafers for General Motors across, I would presume, you know, some of those three subcategories you were talking about, whether it's processors, whether it's sensors, whether it's power management. Is that the case?
That is correct, right. We will have a dedicated manufacturing capacity.
Mm-hmm.
for GM that guarantees a certain output.
Mm-hmm
... of wafers, that will address their vehicle production needs.
Mm-hmm.
In the technologies that they care about or that they've chosen, to go into their vehicles.
Mm-hmm.
They will direct their supply chain that is going to design those semiconductors to manufacture them at GlobalFoundries using that particular corridor.
I would presume, in terms of the win-win situation, you know, from GM's perspective and from your perspective. Maybe I'll just put it to you. What are some of the advantages that you think GM will gain from this partnership? What are some of the advantages from GlobalFoundries' perspective?
Yeah, I mean, look, for GM, number one, they get greater visibility into their semiconductor production. They know exactly when a part is being produced and when it's making its way to them, so full control of the supply chain. They get to choose the technology that best fits their requirements. So again, they're not leaving it to-
Mm-hmm
others in the supply chain to make a technology development for them, right? They pick the technology that best fits their needs, in a location that they've asked for, and with economics that makes sense and makes them competitive, right? For GlobalFoundries, what this provides us is feedback from an OEM, from a technology perspective, that gives us insight into what matters, what features matter, what works better for cars, and how do we ensure that we build that technology and make it robust and long-lasting, and it gives us durability, right? The automotive market, once a particular vehicle line is being produced,
it lasts for many years, and we have, you know, guaranteed durability from that business. It's not going to perturb up and down wildly.
Are you seeing additional interest from other OEMs, for similar agreements?
Absolutely. I mean, I think OEMs are very positive on this framework, given the advantages that we just discussed, Raji, and especially in terms of supply assurance, visibility, and a choice of technology that fits their needs. So we do see a lot of interest and a lot of conversations taking place around this model.
Okay, great. Just moving on to the kind of second theme is the kind of the ICE to ACE transition that you talked about, silicon content increases that are going to be driven by electrification, that are going to be driven by advanced safety systems. At last year's Analyst Day, you showed a great slide with the ICE to ACE content estimates, and you estimated that the semi content, for instance, for a Level 2 electric vehicle, is around $1,500. That's roughly, you know, 3 times what you would get in a regular kind of ICE engine. Are these still the right numbers? How should we think about the puts and takes within that, within the pie?
Yeah, absolutely. I mean, I think when we look at the overall SAM growth over the next 10 years, it's, you know, 15% CAGR, give or take, right? That semiconductor content is really the way we look at it and the way we've developed the model is to double-click into all the components that we see changing in the car and why they're changing and kind of counting. You know, if you look at a Level 2 EV, what does that mean? A Level 2 EV means that it's providing you with safety functions like blind spot detection and emergency braking.
We kind of double-click on that and say, "Okay, well, what does that mean?" It means that there's, you know, four radars, four radar systems on each corner of the car and a radar system at the front of the car. There's, you know, perhaps five camera systems around the vehicle. All of these are interacting in some way. They're sending data to an aggregator of some sort, that then is making decisions and sending those decisions to either brake or, you know, give you a warning or, you know, move, steer the car in one direction or another.
By doing that analysis, we're actually counting every piece of semiconductor, all the way from the sensor to the brain to the power management IC, the chip that's sitting next to the microcontroller, managing the power and ensuring that battery power is reaching it at all times, as a safety function. That's how we do our analysis. Then, of course, you know, we compare that to reports. Absolutely, those are the right numbers, and we're confident in this kind of model.
When it comes to kind of battery and power management and EVs, where is GF able to support the development and address the issues associated with the range and power efficiency?
Yeah, that's, you know, that's a really great question, and it kind of ties into this, question we get all the time about, you know, leading edge versus legacy nodes.
Right.
I love the battery management question because I kind of say, you know, leading edge is in the eye of the beholder. If you are a.
Right.
If you're a battery management system design engineer, in a OEM or in a chip company, a technology that we call 130nm BCD, and that's a 25-year-old technology, is leading edge because we're investing heavily in ensuring that that technology, is able to support battery management systems. Let me just kind of walk you through an example. In an EV where you have hundreds of battery cells, how you discharge and charge those cells in a very uniform way affects the range of the car. What you don't want to happen is one of those cells discharging prematurely that brings the car to a complete stop.
Monitoring the currents in a very precise way at every single battery cell in the car is critical. We've spent a lot of R&D dollars in enhancing our 25, you know, I say 25-year-old technology, but you know, it's, it's brand new. We continue to enhance it every year to ensure that you have the highest precision in terms of measuring current and managing as many battery cells as possible. The requirement there is that this technology can handle up to 120 volts and amperes of current. You compare that to a 7nm or a 3nm node that can.
Right.
-barely handle 0.8 volts and, you know-
Right.
-microamperes of current. totally different application.
Different application.
Right. Without it, you can't do a battery management system.
Works, yeah.
Those are the technologies that we're investing in to ensure maximum range. Think about every other application in your, in the car. You know, we talked about the radar system or the infotainment system, where maybe your kids are playing a game in the back seat. All those are consuming battery power. Managing every joule of energy in the car becomes extremely critical. Ensuring we preserve every joule in the car and use it only when it's needed will help OEMs kind of extend the range of their, of their vehicles.
How does the manufacturing of those wafers help that power distribution?
I, the advantage comes in developing the actual, technology that we manufacture.
Mm-hmm.
What we do is, you know, develop the transistors and the components around that technology platform into what we call a design kit. That design kit is then handed over to the companies that are building the chips. That's really where the, call it, the differentiation is, that they use those special transistors in a way that, you know, is optimized for power consumption as well as meeting their requirements. Once they use that design kit, that's unique to GlobalFoundries and unique to our manufacturing capabilities, they get a product that is best in class in terms of, you know, power consumptions and the features that they're trying to achieve.
Of course, we manufacture it in a very robust way that guarantees that this is going to be reliable in a vehicle environment, right? cars are still going to have to face, you know, arctic temperatures in Sweden or, you know, blazing hot temperatures in Arizona, and you still want them to function and get-
Right.
you from point A to point B safely.
Yeah, no. That's imperative, that's for sure. Staying on the topic of power management and battery management, what opportunities is GF focusing on with regards to wide bandgap technologies such as GaN or silicon carbide? Obviously, silicon carbide is a major initiative in the semiconductor industry. It's powering a lot of the traction inverters for electric vehicles. How is GF positioned there? How do you see these technologies evolving in the coming years?
re absolutely right. Those technologies are critical for many, many, many of the applications that you mentioned. You know, when we look at our investment in wide bandgap, specifically in the area of GaN-
Mm-hmm.
GaN fits the application of onboard chargers.
Mm-hmm.
Every vehicle is going to have an onboard charger, and this is a very sweet spot for GaN, and this is an area of investment for us. In addition to when you think of a vehicle that has battery systems that function at 1,200, 800, 48, and 12 volts, you need things to step up and step down the voltage. GaN is also a very ideal and power-efficient technology to allow you to do what's called DC-to-DC conversion, right? As you step up or step down voltage, you don't lose a lot of energy doing that through our investments through GaN, and that's one of the areas that we're investing in for the future.
You know, as far as silicon carbide is concerned, we're definitely looking at silicon carbide as well. You know, you can hear and read about announcements every day about OEMs adopting silicon carbide for traction inverters.
Mm-hmm.
We continue to look at silicon carbide and monitor. It'd be interesting to see how that market, kind of, you know, moves forward, in terms of the supply, demand, and all the announcements that we see, coming up today.
Meaning there could be overcapacity, potentially?
You know, potentially. Yeah, potentially. It's not clear to me, in terms of short term, there's definitely going to be a need for a lot of capacity as OEMs move to a silicon carbide-based traction system. Long term, it'll be interesting to see where that goes. Yeah.
Yeah, especially with the, with the Chinese coming online as well, building a lot of capacity there. What are you seeing the most strength across technology nodes within your auto portfolio, and what's so ideal about these to the auto applications?
If we stick with the theme of, you know, processing, sensing, and power and data management, from a processing perspective, the nodes that drive all the edge components in the car. The thing that's controlling your seat, your HVAC system, your windows, your windshield wipers, you know, that technology doesn't have to change. You know.
Mm-hmm.
if you've got a windshield wiper controller, why would you wanna move to a newer, faster windshield wiper controller? as long as it's doing that job for you. So.
Yeah
we really see the majority of these type of control applications. They have a sweet spot in the, you know, call it 90, 40, you know, 20 at 2X nm. Depending on complexity. Obviously, a windshield wiper is very simple, but a 17-way seat controller with a massager is a more complex system. That's in the area of processing, and then you have this new class of devices called aggregators and zone controllers.
Yeah
... that are actually, you know, doing some localized decision-making and functions in parts of the car that may be exposed to a lot of heat.
Mm-hmm
... to a lot of extreme temperatures, you want them to be reliable. This is where, you know, 40nm nodes.
Mm-hmm
... nm nodes, 12nm nodes are extremely strong and capable. In the area of power management, our portfolio of BCD, I mentioned this 130nm BCD, 130nm technology, is the sweet spot where it's at. You know, you know, if you get too much smaller in geometry, you can't derive the high voltages that I talked about. Physics simply doesn't work there. There's a similar story in the area of millimeter wave for radar, right?
Right.
Our 22FDX platform is world-class leading for millimeter wave, and it is truly one of the leading platforms in terms of giving you the best of all worlds in radar. You know, in radar, what matters is you want a strong signal that you send out, and you're able to read that signal back when it bounces off another object with high resolution.
Mm-hmm.
You want it to be immune to noise coming from other sources. In addition to that, you want the actual system to consume as little power as possible.
Right.
You know, when you try to push that into slower, to smaller nodes, you lose one of those features. You either lose your strength, so now the range of your radar, instead of being 200 meters, it's now 100 meters.
Mm-hmm.
That's not good.
Mm-hmm.
you know, you still wanna detect cars or objects that are 200 meters ahead of the vehicle. You start to lose things when you get to smaller geometries, and that's another area of focus for us.
Okay, fantastic. Just kind of moving to the ADAS side of the equation. We talked about the electrification. You know, in our view, electrification and advanced safety go hand in hand, kind of similar to your ACE denotation. In terms of ADAS, from your perspective, what, where are we in terms of ADAS adoption? When you're working with your OEM partners, are they planning out a couple of years ahead with you? Are you getting insight to say, you know, 5, 10 years ahead when we would be looking at, you know, Level 4 type of capability? You talked a little bit about the relationship with GM, with that dedicated tech
corridor. Obviously, you're probably working with them for their long-term roadmaps, but curious on how you're thinking about ADAS adoption, say, 5, 10 years out.
Yeah, I mean, I think we have to, you know, we have to believe that, you know, with all our technology innovations, we'll be able to achieve Level 4 autonomous driving in the future, and I think that's a goal for every OEM. Not just from a, you know, from a convenience perspective, it's really from a safety perspective. I mean, imagine the number of accidents that are being reduced when you have automated systems talking to one another, communicating with one another. You know, one car telling another car
where it is and what its next action is going to be. You know, that reduces fatalities, it reduces accidents, it reduces insurance costs. It is absolutely very critical to support that. We're up front and center with our customers in terms of the sensing technologies that they need, in terms of the edge processing technologies, AI at the edge, you know, at the radar node or at the camera node, required to implement, you know, today it's Level 2, moving to Level 3 and in the future, Level 4. You know, absolutely an area of focus and investment for us.
In terms of networking connectivity, there's obviously a lot of demand for sensors, radar, cameras. There's a lot of demand for power management. One of the things that we've been hearing from this conference is, the demand for Ethernet, the demand for PCIE, in-vehicle connectivity, Bluetooth, Wi-Fi, for some more of the kind of IoT-like applications. Tell me about that. I mean, where some are talking about that opportunity for Ethernet is larger than some other opportunities in microcontrollers. Curious how you're thinking about it from the founder perspective?
Yeah. Imagine, you know, if I were to count every edge node in the car, right? Let's say there's 6 cameras, you know, 5 radars, 2 lidars that are on the exterior of the car. You have smart headlights up front. You have passenger detection technology or radar in the cabin, monitoring whether you're, you know, you're falling asleep, if there's a kid or dog in the back, in the back seat. All of those sensors are generating data that are then being routed to a zone, a node that's doing some computation or to a central computer that's doing computation. The amount of data is exploding exponentially. The nuance there. You're
absolutely right. The demand in the vehicle is skyrocketing, right? When we talk about Bluetooth and ultra-wideband, these are IoT applications, but they're also secure applications, right? Remotely opening your car with your phone, and ensuring that nobody can hack into your car or crack that code as well, that is key. There's levels of security required on top of the data that's being shared in the car, as well as the data being transmitted in and out of the car, again, ensuring that nobody's hacking into that data stream, taking control of your vehicle or taking control of your personal information as you're transacting.
A lot of things going on. The question is, it's not always about the highest bandwidth, because there's also real-time requirements, right? When you send an instruction to hit the brakes, you want that instruction to be instantaneous within a fixed amount of time because you want those brakes to go off as soon as you detect danger. Ethernet, as it is today, is not suitable. There's enhancements to things like real-time Ethernet with priority. This is where the kind of, call it, innovation starts to take place, is how do you take what we have in consumer data center, but layer on top security and safety to ensure that it's
happening real time, and that data is reaching where it needs to be in the right amount of time to, you know, to implement a safety function like braking or steering. A huge, huge opportunity in that space as well.
Got it. We have about 5 minutes left. I just wanna hit 2 quick topics. One, just your view, macro, in terms of the auto industry. From your vantage point, you know, where are we in terms of the supply chain shortages? You know, where are we in terms of, you know, more capacity coming online to support, you know, microcontrollers or power management? Any kind of sense in terms of the demand profile out there for auto. Is that slowing down at all from your vantage point, given kind of the higher interest rates for leases and things, or is that market still pretty robust?
Yeah. I think in the short term, Raji, it's a little difficult to predict, but, you know, I think we all look at the same reports. You know, we look at dealer inventory levels, car manufacturing output, and they're still at levels that are close to or below pre-COVID. I do think there's still robustness in terms of vehicle manufacturing through the end of the year and into 2024. But again, I can't predict exactly if we're gonna see a slowdown. I think the thing that is undisputed in my mind is this 15% growth over the next 10 years.
Whether, sure, there'll be a few months of slowdown or upside, overall, that trend over the next 10 years, because of this transformation that we've been talking about today, is absolutely going to happen. The demand for semiconductor is there, and the supply is not. The supply will need to grow over the next 10 years to meet the needs of the automotive makers. I think that some of them are looking at it in that holistic way and are preparing and planning for the future. Those are, you know, those folks are gonna be in good shape.
Great. We'll leave it at that. Thank you so much, Kamal. This is a great conversation. Thank you, everybody, for joining on the line as well.
Thank you so much, Raji. Appreciate the opportunity.