Ouster, Inc. (OUST) Earnings Call Transcript & Summary

Hi, everyone. This is Colin Rusch. I lead the industrial innovation research practice here at Oppenheimer. We're thrilled to have the Ouster team with us today, Ken Gianella, the CFO; and Chen Geng, the SVP of Strategic Finance and Treasurer at Ouster, a long-time veteran of the company. And we may have Angus be able to join us here as well, the CEO.

But to get started, guys, can we talk about the fact that you've got 14 consecutive quarters of revenue growth and you're sitting at a $55 million, plus or minus, run rate. As you bring REV8 into the mix, how are you thinking about the impact to revenue and that product transition from REV7 to REV8 and how that's going to start translating through the P&L?

Thank you, Colin, and thanks again for hosting us at the event today. It's been a great day. Q2 was really our first full quarter as a unified platform with the launch of REV8, and we also announced the launch of Stereolabs Nano. It's a camera specifically for the robotics industry. Really pleased with how the quarter performed. When you look ex royalties product-wise, we outperformed in Q2 than we initially anticipated. And our product revenue quarter-over-quarter for Q3 will be sequentially higher in Q3. REV8, been very, very electric, the reception that it's gotten, multimillion-dollar orders already booked, some from new, some from existing customers, a lot of brand-new use cases that it opens up. So the bookings, we've been really pleased with that. And REV8 is still in the early stages of its volume production ramps. So we said on the call that we're looking for that to ramp, really getting to scale in the September time frame as we go through Q3 and then carrying that on to Q4 and into the out years. We expect REV8 to become the core of our sensing platform for the business over the next several years. But as project -- product transitions go, this one is probably the longest tail that the company's had in its history. REV7 was a real workhorse for us. It's built into a lot of existing company production platforms already. So we expect that one to have a long tail and be part of our product suite for a while to come. The key for us looking ahead, though, it's really about continuing to grow the full product suite, continuing to grow the platform, REV8 being a piece -- big part of that. But what you're going to start seeing as that growth and potential inflection is that platform play of bringing in compute in the software for the individual applications over the next few years.

And it raises a good point around the early stages of growth that you guys are in the individual verticals that are seeing real adoption. So the industrial vertical was the biggest contributor this last quarter, was more infrastructure. How should we think about the durability, repeatability and kind of mix shift from vertical to vertical as we go forward and thinking about repeat orders and the consistency of margins as you see slightly different dynamics in each of those verticals?

Right. Well, I think the one thing we've always talked about is the adoption and cadence is very similar across almost every vertical. You start with a prototype. You move to pilot. You lock in a design win. Then you scale to production. And then once you land and you win that production, you're really expanding and building new use cases at a customer. And so we have several customers that we've been able to demonstrate that with even within the early days. I think the largest thing for us, and just focusing on smart infrastructure for a second, is having that end-to-end play, being able to be no-code turnkey solution, think easy button, for these customers is what we're seeing start to emerge, especially within the Department of Transit areas. For the industrial side, though, it's really about how can you get product to market that's durable and repeatable for the customer and a high quality. And the ruggedness that you need within these environments, especially in the outdoor environments, that's what really sets Ouster apart from the competitors. So I'm not going to make you pick which one we love the best, but if you think about these deployment schedules, you really got to love the industrial side because once they get into that logistical production mode, it really takes off and it smooths out a lot of the lumpiness.

So you signed multiple multimillion-dollar orders, right? And so I think I would love to understand the significance of that. Certainly, we've seen a lot of demonstration of the technology integrated into large capital equipment, ton of AV providers as well as some of the DOT applications. But the history of lidar, in some cases, has been around sampling and smaller order sizes and prototypes moving into these larger awards signals something different. I just want to understand that from your perspective and what that means for maturity of the physical AI space and where you sit in that ecosystem.

Well, where we sit in the ecosystem has significantly changed with the acquisition of Stereolabs and bringing the compute piece to it. So we've moved from a single point provider to expanding to a platform player within physical AI for sensing and perception. So being able to come to these companies and provide them not only just a point lidar solution but be able to offer the compute, the software and the camera, we're already seeing early indications of that really driving us, Colin. As you start looking forward with this, the lidar business is really going to be a big driver of their technology needs because the REV8 brings such a unique component to it that people are starting to build into their platforms and where before you had a lidar that was great and had the right resolution, bringing revolutionary -- speaking of -- it's perfect timing for you, Angus. Speaking of the revolutionary REV8 and the RGB that brings to the table, we're seeing these industrial customers call and want to sign on now because they want to build that as part of their core platform to get that fusion right on the chip. Anything you want to add about the REV8 launch, Angus, and how that does for industrial?

Angus, before you get started, let me refine the question a little bit. We were just talking about the significance of the larger orders that you guys are receiving. And as you get into answering kind of the significance of that for the physical AI ecosystem, the ecosystem here, I just want to understand a little bit more about why that's so important for folks, right, from a product differentiation and what that enables from a system level, if you could address both of those.

Yes, absolutely. I mean, so it's important context to remember that we've been shipping a highly competitive REV7 product line for the last 3 years. We are largely regarded as having some of the best lidar sensors, if not, in many cases, for different use cases, the best lidar sensors on the market with REV7 for the past 3 years. That's why we've built such a suite of high-end customers for our history. And so REV8 is taking something that's already really a leadership position and making it, I would say, obvious that there's no other choice by bringing even more performance capability and now RGB into the mix. And so if you've ever done sales, and I do a lot of sales and talking to customers, there's a big difference between a competitive sale and an obvious sale when customers are just asking and they're not even debating or comparing against the competitor. REV7, you compare against a competitor. You think about it. You choose the REV7 because it has superior specs. REV8, it has native color. It's functionally safe. We have long-range hypercapable OS1 Max versions. It's an obvious choice. That doesn't have a comparison at all in the market. And that makes the sale even that much more -- that much faster, that much easier for us and that much kind of -- that much more confident on the customer to actually place that order upfront. So we've seen a series of long-term REV7 customers immediately turn around. They have confidence in Ouster. They see REV8, and they've signed onboard with a major -- multimillion-dollar orders. And just this morning, the Utah DOT is such a great example of this, turned around 160 intersections with OS1 Max REV8 sensors on them. This is a brand-new product, brand-new hardware, brand-new software, and it's not normal for traffic engineers to deploy this kind of technology at scale when it's this new. But they have the confidence because it's built on the Ouster brand, the REV7 reputation, and it's got such obvious benefits with this native color and long-range capability. And that's just one example of a use case with what I would call a conservative kind of customer, not a customer that wants to take any chances on quality and capability, reliability and yet they have the confidence in Ouster. So that's playing out across all of our industrial customers, all our smart infrastructure customers. They're seeing REV8. They know it's got the capability and the quality that Ouster has become known for. And then they seek native color performance, and it just becomes an obvious decision to adopt.

So I'd love to just follow up on that RGB discussion. Why is that so hard? How long has it been in the works for you guys? I think that's something that takes some explaining for folks newer to the story, but it's a very, very big deal, and it's very, very hard from what we understand. So I would love to get your perspective on why it's so hard, what's involved and what that enables for you from a leadership and a moat perspective.

Yes, for sure. And first, I mean, the concept or the definition of what is this native color technology in the first place, so native color lidar is lidar data that has RGB or color information at every point. And the reality is you cannot -- you and I don't perceive the world in just 3D. We also don't perceive the world in just color 2D camera-like imagery. We actually combine depth sensing and color to describe the state of the physical world. You need both channels. And that's why historically, if you're a REV7 customer of Ouster, almost assuredly, you're buying a color camera in the last 3 years to pair with a REV7 lidar because you need both. You need to know that the blinkers are running on the car in front of you and the brake lights have fired and the stop light is green, not red. You get where all those things are with lidar, and you know you don't want to run into them. But you still don't know the state of a lot of those objects without color information. So our customers are already convinced of this fact, and they're already -- they've been buying REV7 lidars with color cameras. Turns out that's not an ideal way to build those systems. The calibration, the synchronization, the capability of today's cameras just isn't quite -- it's not at the -- where it needs to be for you to have hyper-safe, low-latency and incredible performance on the perception capability of these devices. It's much better if you can make one device that did all of it internally at a very, very deep architectural level that provided guarantees on the quality of the [ paired ] output, of the depth and the 3D and the color. And that's what REV8 does. And it's hard because you have to do it right. You have to do it at the lowest level possible, which is at silicon chip level. And so the REV8 has an L4 architecture silicon chip in it that has color and depth sensing combined for the very first time onto a single chip, all using the same optics, that shared optics in front of the silicon chip that's doing nanosecond-level timing, synchronization and calibration of these traditionally 2 different data streams. So it's really powerful. When you see it, it's obvious. This is a black and white TV to color TV transition. You're never going to go back from operating a color lidar once you've started using it. So it's a really big deal, and Ouster has been investing years on it. And now we have this competitive edge because we're the only company in the market actually doing it.

And what do you think that does from a time perspective for you guys to go on sale? If we're making an obvious sale, how long do you think it takes before someone comes along with something that can present itself as even in the same bucket of performance?

Well, what's interesting is there have been companies that have said that they're doing a colorized lidar in various forms over the last couple of years. And it's made -- they've kind of been these nonevents. I was just at CES in January, actually, and there was a lidar company saying they had an RGB lidar, and no one cared. And the reason no one cared is because -- the customers didn't care is because it was an afterthought. It was a poorly implemented, nonfunctional or nonuseful kind of cobbled together system, not something that was engineered at the silicon level. So I expect there to be a lot of that kind of thing that happens for the next couple of years as companies try to market the idea of color lidar, but they won't drive the benefits of it because customers are smart. I say you never trick a customer into buying a lidar sensor. All customers are going to test before they buy, and so you can't kind of fake this color capability. And so I think we have multiple years, 2 to 5 years of head start depending on which type of product and which environment we're using it in, a head start because of the silicon lead times associated with doing this correctly. In the meantime, there will be all kinds of press releases with companies claiming they've done it, and it will be irrelevant because it will be kind of a demo prototype system that customers can actually field at scale.

Awesome. So then just shifting to the sensor fusion and software piece of this, right? That's something you guys have talked about. You've made some acquisitions related to some of the software, but sensor fusion has been historically a critical part of performance for these systems. But obviously, with more information native to the sensor, you can redesign some of that sensor fusion in the computer. Can you talk a little bit about what you've done from an investment perspective around software, sensor fusion and what that means for system simplicity and performance?

Yes. So sensor fusion is a really nitty-gritty problem that every single customer has to solve. And it involves timing synchronization and understanding where different sensors are pointed and getting all of that data into one reference frame that a machine learning algorithm can parse in a GPU. You have to take all that data from many different sensors, bucket it up properly and give it to machine learning. And the quality of the machine learning output is directly related to how well you do all that fusing of data upfront. That's why native color -- native color just gets to sidestep some of that when we're talking about color camera versus lidar. But it also pertains to when you have multiple lidar sensors or maybe lidar sensors and stereo cameras or monocular cameras, stereo cameras, lidars and inertial measurement units. And so there's a broader ecosystem of sensors that all have to be calibrated, synchronized and fused before you apply algorithms to them. And I think this is a place that Ouster can provide huge value by building that software layer for customers and providing it as part of our SDK and software ecosystem, providing the -- it just works platform of sensor perception and development that no other company has ever tackled. We've done this in a couple of ways, but the big one is that we acquired a company, Stereolabs, just at the beginning of this year, which is a leading provider of stereo cameras. So that's colorized point clouds from stereo cameras and monocular cameras that we can pair with our colorized REV8 point clouds. And Stereolabs has also invested in this SDK, the ZED SDK that has over 10,000 users, something they've built over the last 1.5 decades. And what it does is it fuses the disparate data streams from all these different sensors into one coherent stream that they can then pass the machine learning algorithms. So we have 10,000 customers currently using the ZED SDK. It's doing this exact -- this really difficult technical challenge that every customer needs to solve. And now we're just investing more and more. We're getting REV8 onto the platform. We're building it out to be more robust and rigorous for production environments because I want customers -- I want the heavy equipment manufacturers of the world that are going to put 10,000 construction machines out into the world that have to last for 10 years, I want them running our software and our sensors. And so this is the start of that, that autonomy platform play, which is a major investment for us for the next decade.

Excellent. And just continuing on that, you guys have deepened the NVIDIA partnership in REV8 to DRIVE and Jetson and dedicated toolkits and drivers for Orin and Thor. How does that reference design position and the kind of the brand compatibility with NVIDIA translate into the sales process and adoption rates for the Ouster platform?