Exhibit 99.2
The following is a transcript of Cowen Inc.’s 14th Annual Global Transportation & Sustainable Mobility Conference held on September 10, 2021.
<<Jeffrey Osborne, Analyst, Cowen Inc.>>
Hi, good afternoon everybody. It's Jeff Osborne, the mobility technology analyst at Cowen, pleased to be joined by my colleague, Josh Buchalter, who's been working on the semiconductor team for several years and is doing quite a bit of work with me on the sensor side. Very pleased to have the team at Cepton joining us Jun, Winston and Hull. Thanks for hopping on with us. I'm sure you've got an incredibly busy day, but I appreciate you giving us 30 minutes for the session here.
For those that might not be familiar with Cepton or the recent SPAC announcement, do you mind just spending a few minutes giving us a quick overview of what Cepton is up to, and we'll take it away from there with some particular questions that Josh and I had.
<<Jun Pei, Chief Executive Officer and Co-Founder>>
Great. Hello everyone. Good morning, good afternoon, good evening, wherever you are. And we're Cepton Technologies, a startup company in Silicon Valley. We're a lidar company. So as you're probably aware, there are many lidar companies out there are already public and many more are private. We actually is one company that differentiated ourselves from five years ago when we started focusing on ADAS industries and that is the biggest pie right now. It is a public knowledge that we have the largest design win in ADAS industry after five years of hard work and that is from General Motors and will be proliferated into multiple models of their vehicles. And we're really glad to participate in this conference and talk about this design win, which is really the validation for our technology and also for those people interested in the technology itself. I'm really happy to talk about the MMT lidar that's unique to Cepton. So let's get started.
<<Jeffrey Osborne, Analyst, Cowen Inc.>>
All right. Perfect, Jun. Yes. And it's finally nice to see now that you're public, what efforts you've had out there. It's so great to hear about GM. Maybe before we dive into the specifics of Cepton, just given you've been working on this for five years, have a differentiated approach from a tech perspective. And there is so many different companies out there in the broader lidar space. I was wondering if you could just give the investor audience sort of a framework on how we should think about lidar companies in general and comparing and contrasting their approach. There is certainly a lot of parallels to cameras in terms of range, field to view and other variables. But if you were to think about analyzing an approach to assessing the winners and losers for investors in the space, what would that framework be?
<<Jun Pei, Chief Executive Officer and Co-Founder>>
Well, there is a lot of loaded question there. It's really coming from multiple angles. But perhaps let me just start from is there a need for lidar out there. And that actually has been answered in many ways very notably by GM's awarding us that was that’s largest design win. But really from the technical perspective, you look at yourself, you have your nose, you have your ears, you have your eyes, and these are sensory systems for you. And lidar is a sensor for cars. The first application or first purpose of having lidar is for safety. It's not the bandwagon of level four, level five autonomies from years ago, a lidar will actually make your drive safer.
So there is no ending in people's desire for additional level of safety. Now if you're 90% safer, you want 99% and after 99% you want 99.9%. So this desire for additional level of safety is never ending. As long as the cost is justified, so additional sensor modality into your safety suite is a sure trend. You considered cameras many years ago as a luxury item, now it's really a regulated item and we see lidar in the same path.
Now, talking about the technology, which technology will be winning? Well, I'm an optics engineer from training in the study, in this field for decades. Even for me looking at all the available technologies out there is a confusing world. So for me to actually talk about technology, the only validation of a technology being viable or not is actually whether there's commercial design win.
And, of course, I'm talking about ourselves quite a bit that the MMT technology that we have invented strikes a critical balance between performance, cost and reliability. And there are many companies that would actually stretch out in one of these areas. They would work on high performing lidars or earlier on people thought that the solid-state lidar could be super reliable. So these are only one pillar or one factors of the overall package. And what we have done? The MMT technology is striking the balance between performance costs and reliability and that's a huge enabler to have the commercial design win. So I'm happy to share more details, but I hope that I covered a lot of aspect of your perhaps simple question.
<<Jeffrey Osborne, Analyst, Cowen Inc.>>
Absolutely. And maybe two more on my end and I'll turn it over to Josh. But as an optics engineer, as you were starting Cepton, I'm sure you've tried a lot of things that you then canceled and put on the back shelf and never revisited. And so maybe you could just walk through the journey of where – how you come, where you are, did you start ever pursue 1550 versus 905 time of flight versus FMCW maybe some of the challenges that you had with those different technologies. And then following up on that, maybe just define what MMT is, that seems to be your secret sauce.
<<Jun Pei, Chief Executive Officer and Co-Founder>>
Yes.
<<Jeffrey Osborne, Analyst, Cowen Inc.>>
But I would love to hear about the history and why you chose the path you did, and then define what MMT is and how that's moving forward?
<<Jun Pei, Chief Executive Officer and Co-Founder>>
Yes. Thank you for that, asking that question, you're far right into my backyard there. Indeed at the beginning of the company, unlike many other Silicon Valley startups that you have a technology looking for application, or you have an invention looking for a place to use, we started with a market focus. We want to make a ADAS lidar for everyday consumer vehicles. So that's actually a market focus driving all the subsequent activities. And with that thought, we actually indeed spent a lot of times like how we're going to make the lidar. And lots of them are luckily thought experiments, not actually spending money, doing the experiments and very structure thinking what actually put the three decisions ahead of us. You need to decide, which illumination that you want to use for lidar 905 or 1550. You want to decide what kind of detection you want to use, time of flight or FMCW. And lastly, you want to actually decide how you make a one point measurement into an image of the world. So the imaging or proliferation of your lidar measurement around environment is the third column that you have to decide.
We really have considered the first two columns, illumination and detection. These two wheels, if I may call them, already invented. So for us, we chose 905 as the illumination, because the – it is a silicon-based technology, extremely mature, decades old, and it cost a little over a dollar, a piece for the laser diode and it is already automotive grade.
The second column, the detection, the time of flight was even invented, even earlier, right after laser was actually discovered in the 1960s. And now the avalanche photo diode that we use cost less than a dollar automotive grade, very mature technology. So these two wheels have been invented and they solve the problem already. So, if – of course, there are fancy words, like 1550, FMCW, you always ask yourself, well, whenever you're facing a new technology, what problem it solves. Okay, now comes down to last portion, which is the imaging. That's where most of the lidar companies should and are actually investing and trying to come up with the most efficient way of doing imaging.
A dozen years ago, Velodyne started the first imaging lidar that is actually spinning lidar. So spinning mechanism itself is an imaging mechanism that is very, very crude, but it was effective as a hobby project. When it comes to automobile applications, quite a few years back people talk about solid-state scanning, people talk about flash lidar. And there are a number of notable companies using MEMs mirror as the scanner to scan the beam. So all of these are imaging mechanisms that Cepton Technologies, we ourselves have considered very, very carefully and quite a number of them are thought experiments. As I mentioned, we really considered the flash lidar being – not being able to see very far just from all – back of the envelope calculation, unless the cost is escalated to astronomical high for space station. Docking is a flash lidar, but that's a few million dollars, a piece. And then the MEMs mirror has a fatal reliability issues that we could not resolve with our ability.
So after all, it came down to this one mechanism that we eventually invented the Micro Motion Technology, MMT, which utilizes a very old technique, voice coil. So it's just like you are loud speaker. It has a voice coil inside and loud speaker, your flow current through the voice coil, it actually moves slightly. And this motion, if you attach a paper cone, it will generate sound and that becomes your loud speaker. It is a very old technique, very reliable. In your cars, they are plenty of loud speakers and they all work and they outlast their car. So we adopted that. So instead of putting a paper cone on top of the voice coil, we put an optical array, and then we flow current through this voice coil and it moves, moves at microscopic rate. Micro Motion Technology, MMT, that's where it came from. And it moves in optical array and this motion of the array translating into a beam steering after lensing going out.
So this type of micro scanning does not involve any mirrors. So it's very, very simple, highly efficient mechanism. And it does not have any friction like the motor and shaft, that will work here. So all of these are coupled together. It became an elegant solution that is low cost, highly reliable and preserved the performance of 905 laser and detector. So that's the core of MMT and that's really the enabler from the technology perspective that give us the largest design win in this business.
<<Jeffrey Osborne, Analyst, Cowen Inc.>>
Well, excellent. Josh, did you have anything on your side, on the ASIC or any of the other technologies?
<<Joshua Buchalter, Analyst, Cowen Inc.>>
Yes, sure. I guess one quick clarification on MMT. In our past conversations, you've described it as when I asked whether it was solid state or non-solid state, you explained, I think very elegantly that it's not as binary as that. I think it would be helpful for investors if you walk them through that conversation, because some do get – in many of the – guys – we form our framework get caught up in solid state or mechanical. Yes, thank you.
<<Jun Pei, Chief Executive Officer and Co-Founder>>
Okay, great. I'm glad you remind me of that conversation that solid state was a big word. And then I asked you what exactly – how do you define solid state. And people's conception of solid state is there is no moving parts. That is more or less true. However, as a scientist or technologist looking at the world, I don't view things black and white. It is – if you imagine yourself, if you have a microscopic eye that you can see motion in nanometers or micrometers, everything around you is actually moving, you puff in the air, your desk will actually move a little bit, but you just cannot see because it's in nanometer scale. And the same thing with solid state versus non-solid state. It's really a definition of what the motion scale is.
So a natural question is do you consider loudspeaker a solid-state device? Some say – would say, yeah. If I look at it, it doesn't move. Well, that's a tweeter. Really you cannot see any motion even when it's making sound. But when you look at a subwoofer, it actually moves and moves by quite a bit when it generates loud sound. So there's a gradual scale from your subwoofer to your mid-range to your tweeter. And that's how a non-solid state evolves into a solid state. As I mentioned, our MMT falls right in between this. It's actually just like a mid-range loudspeaker. So I can ask everyone here to make your own judgment, whether you call a mid-range loudspeaker, a solid-state device or not. That's how your view MMT.
<<Joshua Buchalter, Analyst, Cowen Inc.>>
Thank you. And then switching gears, how – can you talk about your ASIC, what design implementations went into that? And how important is that to the perception when combined with your software? Are your key customers using you primarily to gather the data in a cost effective way? Or are they utilizing your solution because of its ability to perform perception on the device?
<<Jun Pei, Chief Executive Officer and Co-Founder>>
Yes. Again, thank you. That's a wonderful question. When I actually talked about MMT, I talked about the voice coil moving, I talked about the transmitter sending out laser and the scanner actually scans the beam, all of these seemingly rather straightforward and simple, but the actual coordination of all these activities are rather complicated, just taking the receiver end when the light pulse comes back to you, you have to amplify it, you have to digitize it, you have to calibrate it, and then you have to pump out the last bit of data with the correct point called information. So all of these are actually carried out with our ASIC chip that we call a single chip lidar engine that we spent the last three years to develop.
This is an automotive grade chip that actually can take all of these tasks. And it is a sesame-sized chip that costs a little over a dollar, another great enabler for us to have the largest design win is this chip. Only with this type of cost we can get down to the level that a mass deployment is possible.
Now, further to this as a company, we started with a lidar building block, the MMT. We also actually constructed multiple variations of MMT lidar and one of which is going to be in the consumer vehicles at General Motors down the road. But we do have a plan and execution working not only the raw point cloud, but also turning these raw point cloud or the raw data, I may say, into a more perceptive – perceived data, more abstract data, when you drive out there, I can tell you, X, Y Z position has a dot. You have to know there is an open road, there is a tree, there is a dog. All of these are abstract perceived information that we are working on to turn them into some more useful information to our customers.
So we're moving up in this value chain, adding additional software content to our ASIC chip to start with and follow the additional processing platforms to our customers. Well, depending on who the customer is and their level of interest, we are actually very flexible in terms of – you can just take our raw point cloud and go with our sensor only, all the way that you can take our perceived data and use only the abstract format. So we're – we already have a perception software suite called Helius. That's used for non-auto applications. We're working on the auto applications and it is a long journey but we're on our way.
<<Joshua Buchalter, Analyst, Cowen Inc.>>
That's helpful. I guess, shifting gears to the commercial aspect, can you talk about your partnership with KOITO? How did that come about? And what does that bring to the table as you go talk to your end customers?
<<Jun Pei, Chief Executive Officer and Co-Founder>>
Yes. Well, first of all, KOITO and us actually are together that captures the General Motors design win. So we actually observed a very strict supply chain relationship that we're the tier 2 partners. So we supply KOITO with our technology and with critical components like our ASIC and some MMT critical components. And KOITO would actually manufacture these by assembling them together and vouch for their quality and reliability, and eventually vouch for the liability to the end customer. So this is a classical supply chain. And I consider lidar being such a safety critical component must observe this strict supply chain rule.
How did we and KOITO come together? Well, KOITO is this 100-year-old company that they're really successful being the largest headlamp supplier of the world and that's their past 100 years. And now, well, a few years ago, they're looking to their future by establishing field offices in Silicon Valley, in Europe doing technology scouting. This is actually looking at a technology that can be related to headlamps, or can be not related to headlamps. It's just what the future of KOITO is. They have the aspiration of extending the lifetime and also a dominate the auto supply chain in many other areas. Long story short, we came together because of their desire to advance their technology. We came together because we want a tier 1 to support us in terms of deploying our technology to everyday consumer vehicles. So that's how we came together. Well, our relationship is not mutually exclusive. We're non-exclusive to each other. So we actually have a multiple OEM engagement and multiple tier 1 engagement, but the most notable one obviously is KOITO and the General Motors design win.
<<Jeffrey Osborne, Analyst, Cowen Inc.>>
Just…
<<Joshua Buchalter, Analyst, Cowen Inc.>>
Oh, go ahead, Jeff.
<<Jeffrey Osborne, Analyst, Cowen Inc.>>
I am just going to ask a quick follow-up and then I'll turn it back to you, Josh. Is the GM design win, will the lidar unit not be in the front grill, but it will be in the headlamp or just has nothing to do with their headlamp business?
<<Jun Pei, Chief Executive Officer and Co-Founder>>
Okay. So our lidar is one of the most compact lidar in terms of this form factor and does have low power consumption that enables it to be either inside the headlamps or in the grill or behind the windshield. The General Motors implementation is something that I cannot disclose, but I've already…
<<Jeffrey Osborne, Analyst, Cowen Inc.>>
That's right.
<<Jun Pei, Chief Executive Officer and Co-Founder>>
Yes.
<<Jeffrey Osborne, Analyst, Cowen Inc.>>
That's correct. Thanks much.
<<Joshua Buchalter, Analyst, Cowen Inc.>>
Hey, I'm going to try a different way. Are you able to share what types of applications its enabling for the GM vehicle or anything else you can share? And correct me if I'm wrong, but I believe it is sole-sourced, is it correct?
<<Jun Pei, Chief Executive Officer and Co-Founder>>
Yes, it is. Okay. So I will say more generically about what the lidar is used for the ADAS function. And the first and foremost is actually additional level of safety. Once you have people's hands off the steering wheel, once you have possible pedestrians walking in front of the car that you need the automatic emergency breaking, the AEB system to function correctly with confidence with no false positives and all of those things the lidar is serving as a critical function in providing reliable data that – a noise free data that you can actually execute. Of course, this is – General Motors already has a Super Cruise ADAS function. And obviously, this deployment, there will be a next generation at a much higher level and it will enable the cars to go into the roads that could not go before with the ADAS function enabled. So it does not limit it to only the highways or with clearly markings under good sunlight, under good lighting conditions. It actually just opens up the horizon for the ADAS application to other areas. So safety…
<<Winston Fu, Chief Financial Officer>>
So actually sorry to interject, Jun. I mean, I think that's a question you should direct at GM. But I think Jun's answer was kind of – just want to make clear, Jun's answer was kind of more general in terms of what our lidar enables. So we don't want to get anything specific into GM's intent.
<<Jun Pei, Chief Executive Officer and Co-Founder>>
Yes.
<<Joshua Buchalter, Analyst, Cowen Inc.>>
Okay, understood. Last one from me, I'll turn it back to Jeff. You've mentioned a couple times that your system is designed explicitly for ADAS and I think implied in that is that some of your peers are going after more full autonomy. What is there to be glean from that on the cost side? I guess, is your solution priced below many of your peers? And if so, how are you able to do that? What about your architecture?
<<Jun Pei, Chief Executive Officer and Co-Founder>>
Yes, obviously, as I mentioned, the three pillars, performance, cost, reliability has to be balanced, and you cannot just emphasize on one and forget the other. And whoever worked with OEMs auto manufacturer or tier 1s will know the importance of cost. So, we're looking at below $1,000 lidar at the moment with the goal of reaching below 500 in large volume. So these are the ballpark numbers that I get to say and certainly is – these are very, very attractive good numbers. Just ask yourself if you can promote your safety from 90% to 99% with extra $500, would you do it? The answer is yes for many. And how did we do it? Yes. Okay, go ahead. Sorry, Winston.
<<Winston Fu, Chief Financial Officer>>
Yes, I just want to touch on something to make clear. Our view and we have a rich product roadmap. Our view is the road to autonomy is first through ADAS and Level 2+. And you deploy it and you go down the cost curve and the volume curve with ADAS and then you add additional lidar content to get the 29s, 39s, 49s of reliability. And so you're able to add more LIDAR content because the cost – you've gone down the cost curve and continue to innovate on the technology front.
<<Jun Pei, Chief Executive Officer and Co-Founder>>
Let me just wrap it up by saying how we got there. And you probably already sensed when I talk about our technology, how we selected the illumination, the detection, and how did we arrive at MMT. Cost is in everywhere. We chose the simplest laser diode that we chose the simplest APD. And then we actually used a loudspeaker technology, ask yourself how much is a loudspeaker and the voice call is such a mature thing that the cost is all over the place, unlike the expensive MEMS device or mirrors or God forbid, flash lidar with special semiconductor processes. We just use off-the-shelf components mature technologies, and put them together in an elegant way that it becomes a low cost and effective solution.
<<Joshua Buchalter, Analyst, Cowen Inc.>>
Thank you. I'll take it back to you Jeff.
<<Jeffrey Osborne, Analyst, Cowen Inc.>>
Yes, thanks, Josh. And for the investors tuning in, I would definitely encourage you to check out the Cepton Twitter feed from early April, great post on Tesla. I'm not a fan of Tesla's approach and their anti-lidar stance, but I think you guys had a great rebuttal on what cameras are good at, what lidar is good at and why both can co-exist. So congrats on that post. Very detailed. Maybe a couple other things I was hoping to get at the final three minutes here.
So as you've got now the SPAC merger deck out, your S-4 is filed which is great to see. How would you answer the question that we get almost about every SPAC as it relates to revenue visibility? Some – the perception is a lot of these SPACs are sort of sticking their finger up and saying up into the red as fast as possible. So, obviously, GM now is out in the public domain with the S-4, but how do you answer what's visible? What sort of near-term possibility versus maybe more of an ambitious still on the come type of thing? I don't know Hull if that's in your camp or whoever wants to address that?
<<Winston Fu, Chief Financial Officer>>
Let me start. And so our, forecast is bottoms up and as you alluded to, the foundation of our forecast is the large production award from GM. And there is information in the S-4 that talks about how we've looked at the forecasts. We use third-party data to look at the exact models that we've been awarded to look at the overall volumes and then we judge that down and estimate an attach rate for lidar in those models. And so some luxury models are going to have 100% percent attach rate, some of the mid-runners, but very high volume runners are going to have a smaller attach rate based on option packages.
And so, we take a balanced approach there. And then we – there is upside if there is of course more aggressive marketing by our OEM, and then there is another set of – so that's the awarded part. And then the next part of our sales funnel, we are engaged with all the top 10 OEMs worldwide and then also trucking OEMs. And so, they include affiliates and other OEMs that we've been engaged with for many years. So we give an estimate based on kind of what's OEM design win cycles are, but we certainly feel like GM is a leader and GM is a leader in ADAS. And the fast followers, the cycle time could be faster and theoretically should be faster.
<<Jeffrey Osborne, Analyst, Cowen Inc.>>
Absolutely. Maybe one, I know we're out of time, but one quick clarification on GM and it might be in the S-4 in the deck, but I forget. Do we have a sense of what model year or what the SOP is of when that starts? Is that something you can share or no?
<<Jun Pei, Chief Executive Officer and Co-Founder>>
2023.
<<Jeffrey Osborne, Analyst, Cowen Inc.>>
2023.
<<Jun Pei, Chief Executive Officer and Co-Founder>>
Starting…
<<Jeffrey Osborne, Analyst, Cowen Inc.>>
Starting the production. Okay.
<<Jun Pei, Chief Executive Officer and Co-Founder>>
Yes.
<<Jeffrey Osborne, Analyst, Cowen Inc.>>
Perfect.
<<Hull Xu, Vice President of Finance and Strategy>>
Yes, there is a lot more information in S-4 that was filed a couple of days ago. So I encourage people to take a look at that in terms of the award and the platform and the models. It'll be on.
<<Jeffrey Osborne, Analyst, Cowen Inc.>>
Excellent. Well, gentlemen, thanks so much for taking time. I appreciate your insights.
<<Joshua Buchalter, Analyst, Cowen Inc.>>
Thank you very much.
<<Jun Pei, Chief Executive Officer and Co-Founder>>
Thanks, Josh.
