Conference Transcript - KDK 2026-05-27

Etai McKellie, Analyst — TD Cowen

Okay, great. Good morning, everybody. Very pleased to have our next guest at the conference, Kodiak AI. First time here at the TD Cowan TMT conference. Very pleased to have Kodiak's founder and CEO, Don Burnett, for a fireside chat. I'm Etai McKellie, I cover autos and mobility. So we'll get right into it. Don, great to see you. And if anybody has questions, feel free just to raise your hands throughout the session and we'll get a mic over to you. So, Don, welcome. Great to see you. Thanks for being here. Maybe just for those who are newer to the Kodiak story, a brief overview of the company, some of your verticals and differentiation. We can maybe just start a little bit there.

Don Burnette, CEO

At a high level, Kodiak is at the forefront of the physical AI revolution that we're seeing play out in front of our eyes. We're focusing on the mobility vertical of physical AI. And within that category, we focus on three specific applications or industries. The first one is probably the largest opportunity, certainly the largest TAM with a massive need, which is long-haul trucking. We focus generally on more commercial-like vehicles with our AI technology. And so we serve the long-haul trucking market. We have trucks deployed based out of Dallas throughout the southern United States. We have over 25,000 miles in our commercial network that we have the capability to autonomously operate in. In addition to highway driving and what we call more structured autonomy, we also do unstructured autonomy. This is one of the things that makes Kodiak somewhat unique within the autonomy space, is that we also have the capability with the exact same AI system to go into off-road and unstructured environments. And that opens up new TAMs, new opportunities, and new businesses. So the first real application that we've deployed is with a company called Atlas Energy Solutions. They're based out of Texas. They operate trucks in the Permian Basin of West Texas. And that's an industrial application moving sand in off-road settings. And we have 28 trucks now in the hands of Atlas, owned and operated by Atlas in a driver-as-a-service model where those trucks are out there running 24 7 using our technology providing value and we've been scaling that business over the last year and a half and then the third vertical is in defense so defense tech is becoming one of the one of the hot topics of the day and kodiak's been working in defense for several years now we've worked with multiple branches of the u.s military starting with the marines we've done a lot of work with the army and more recently we announced a project with the um excuse me we started with the air force then did a lot of work with the army and most recently we we announced a program with the marines uh we're we we just announced a strategic partnership with general dynamics land systems where we're going to be working with them on a future ground autonomy based platforms for all branches of the government and and allied nations And we're really excited about the capabilities that we bring to the defense space with commercially mature AI technology, which, again, is, I think, a competitive advantage and somewhat unique within our space.

Etai McKellie, Analyst — TD Cowen

Perfect. Todd, you mentioned the unstructured nature, being able to go from different verticals. And maybe it's a good segue into the tech stack. There's a lot of conversation with investors on AV 2.0, 1.0, the differentiation. Kind of where do you land there, and how do you leverage your tech stack to be able to serve various end markets?

Don Burnette, CEO

People love their taglines and their marketing labels. But there's no real good definition of the different AV 1.0, 2.0 things. But the way that Kodiak operates is really in the AV 2.0 or even 3.0 nowadays kind of fashion, where we have an end-to-end AI system, which is operating in a real-time capacity. We don't depend on legacy approaches that may have been used in decades past, such as high-definition maps, for example. And this is precisely what allows us to go into environments that effectively can't be mapped, either because they're technically impossible to map or because you don't have the opportunity to map. So, for instance, in defense, you don't get to go into a conflict or war zone with your robot and say, don't mind me, I'm just in here, I'm mapping out the environment, I'll be back one day, right? That's not how it works. You need to go into areas where you've never been before, there are what it's never seen before. You might get aerial imagery, and that's the best that you get. With a real-time end-to-end AI system, you can serve that use case. And that's, again, one of the big differentiators with Kodiak. Also, it's allowed us to go into the industrial trucking sector. So I mentioned it briefly in the intro, but we work in the oil and gas industry within industrial or off-road trucking, but we also announced a partnership, our new customer, in the logging and wood products industry. There's also opportunities in mineral and mining resources, and this is a global opportunity. All of this is unstructured, and all of this really requires a modern end-to-end AI system in order to navigate confidently and perform safely.

Etai McKellie, Analyst — TD Cowen

Terrific. And still sticking with a bit of an overview for investors who are newer to the autonomous vehicle trucking space in general, how do you characterize the opportunity in trucking industrial defense versus like a robo-taxi, right, which is getting a lot of attention today as well? How do you kind of think about the most attractive verticals within broader

Don Burnette, CEO

physical AI and autonomy? Yeah, I mean, look, I'll say something more broadly to start. I think we're going to see a complete transformation of the transportation network over the next decade, AI is advancing at such an amazingly rapid clip. And we're at the forefront of bringing that AI technology into the physical world. Of course, we're starting with long haul. We're starting with industrial. We're working on defense. But you're going to see this permeate into all aspects of the transportation space. So the ones that people focus on today are mainly the robo-taxi sector and then primarily the freight sector, the large vehicles. You know, the interesting thing about the taxi market is that it's not actually a – well, it's not one of the larger transportation markets, right? Here in New York, it's a little bit lopsided, but if you go to other places in the country, it's not as big of an opportunity. So we see autonomous trucking as roughly an order of a magnitude larger than what the robo-taxi market represents today. Now, if you think that everybody's going to get rid of their cars and start driving robo-taxis full-time, you know, countrywide, that could change. But ultimately, freight is the backbone of our economy. And everybody in this room, everybody listening to this talk depends on freight moving reliably with resilience on time and efficiently to bring goods, services to people who need it, where they need it, and at a cost point that they can afford. And so autonomy is going to play a massive role in enabling that economy to continue to grow. And so not only is it roughly an order of magnitude larger than I think the next biggest opportunity, but it's something that the economy really depends on, and autonomy is a necessity.

Etai McKellie, Analyst — TD Cowen

Awesome, absolutely. We do have some spaces here up front as well. So you're doing it. You've got your deployed in the Permian. I think you mentioned 28 trucks, and you're learning a lot. So what are you learning? How has it been progressing? and how useful are those learnings into other verticals like defense and, of course, long-haul?

Don Burnette, CEO

Sure. So the learnings are everything. And I talk about three pillars that go into building an autonomous product. And this is uncharted territory, right? There haven't really been autonomous vehicles in the past, and so we're all figuring this out for the first time. Those pillars are technology, safety, and product. And usually the conversation is focused around technology, where, you know, what kind of AI do you use, what's your tech stack look like, what's your capabilities, how do you handle edge cases, and all that great stuff. But ultimately, it misses the broader point of once you have a self-driving vehicle, what do you do with it, and how do people use it? That's really where the implementation and the product pillar comes into play, where you hand it to a customer. Like, let's say I gave gave you an autonomous truck, and you were like, OK, how do I turn it on? How do I use it? Is there an app? Is there a website? Do I talk to it? How do I get it to do what I want, pick up here, drive there? What if I have to make a change? How do I have visibility into the product as it moves? How does it interact with my existing infrastructure? You have freight carriers and industrial trucking companies have all their own existing infrastructure. So there's a lot of these questions that go well beyond the technology that you need to address. And unfortunately, there's no easy shortcut to get those learnings and understand what you don't know until you've actually deployed it. So one of the advantages Kodiak has is now for over 18 months, we've actually had a true driverless product in the hands of a customer. And when I say driverless, it's sad that I have to caveat this because there's so many different definitions of driverless. I'm talking about empty cap. Nobody in the vehicle, no required remote monitoring of the vehicle. These vehicles are truly out there on their own, operating with no supervision, nobody watching, until they may ask for assistance or something like that. And so you have to go through all of the pickup, all of the drop-off integrations, and that's hard. It turns out it's challenging when you don't have a safety driver, a safety observer, to kind of cut the corners for you. And this is something that I think most autonomous companies don't quite realize is that when you mock it with a safety driver, inevitably that driver, that person will do things for you, even if you tell them not to, right? They will make eye contact. Well, there's no eye contact once the person's gone, right? They could give social cues. They could help out with turning a wrench. They could, you know, hit the tires, do some inspection work. There's lots of little subtle things that they will inevitably do for you. and it's not until you truly pull the driver and put it in the hands of a customer that's operating a real business, so not as a demo. And this is something I think is also really important here. The truth of the matter is when we work with companies like J.B. Hahn and Werner and other large trucking companies in the highway sector, we're moving freight for them as are our competitors, but it's pretty much a one-off relative to their scale. It's not their true operation. I mean, we try, we really try to make it the true operation, but their business does not depend on those trucks working, right? If something happens, you know, we may pay a penalty, we may have a conversation with them, we may figure it out and debug it, but ultimately their business doesn't depend on autonomy today, right? With Atlas, that's not necessarily the case. They're serving well sites all over the Permian, and we now have 28 trucks in their fleet, which is, you know, a decent percentage of their fleet. They still have many, many more manual trucks operating. But their business really depends on us delivering. OK, so the SLAs are real. The business and value proposition is real. And it's got to work. In addition, there are 20. Sorry, I'm going on. But I think this is important to kind of convey. It's a 24-7 business. So there's no downtime. There's no reset time. And you asked about learnings. And I'm getting there. The people who actually have to utilize these vehicles, they aren't the executives that strike the deal and write the contracts, right? They aren't the people that I interface with and talk about over a dinner of how exciting the future of autonomy is, right? These are folks who are just trying to do their job day in and day out, run an efficient ship, get the product from the well site or from the drop zone to the well site, excuse me, from the extraction site to the well site and back, And they just want to do it on time so that they can complete their job. And if it's clunky, if it's slow, if it's inefficient, they get really frustrated. They don't care. And so we've had to work through these challenges. And those challenges are real. And they're something that you cannot learn any other way. That was the whole point that I was getting to with this story, is that you really have to be in the trenches with the customer in a real environment where it really matters in order to know if your product actually works. And that is something that Kodiak has really uniquely been exposed to, particularly in the trucking space or in the commercial space. And those learnings are, I call it a flywheel, that we can then take to other customers. We've been there. We've understood the challenges. We know what the questions are. We know where the hiccups are going to be. And we can be much, much more efficient as we bring this to our second, third, fourth, and fifth customers. And that applies across both defense and over the road.

Etai McKellie, Analyst — TD Cowen

Yeah, that's great. Maybe just to continue on that, you know, in Q1, I think your paid hours of driver's operation up 120%, customer truck deliveries up only 40%, so implying significantly higher utilization. How much do we read into that in terms of success, learnings? What does that mean going forward? And does something like that even drive new customer interest as they see your initial customer just utilizing the trucks a lot more?

Don Burnette, CEO

Yeah, absolutely. So, you know, they're a 24-7 business, right? And in the early days, we delivered the first two trucks. They, you know, they were not being operated continuously because of these challenges that we ran into, right? Not from necessarily a hardware reliability perspective, but just a usability perspective. And again, there were lots of things that we didn't know we didn't know, and they didn't know that they didn't know. And we had to kind of learn this together. and so gradually over the last year and a half the efficiency of our fleet has been going up not necessarily because it's a combination right we have implemented tools that make the customer more efficient and effective i think it's important to remember that that atlas owns these vehicles they are they are owned and operated by atlas we don't control them as far as we're concerned if they want to park them in the parking lot that's fine with us uh because we we you know we have a recurring revenue model where they pay us for basically the opportunity to use the asset. And then it's on them to actually utilize the asset. But it goes both ways, right? They try to do something, it's got to work. And if it doesn't work or if it's a little bit slow, then they have to work with us to make it better. So yes, we have been continuously increasing the efficiency of the vehicles, but also enabling them to be more efficient with the fleet. And as you said, we've now gone up over 120% in Q1. We drove more hours of operation in Q1 than all of 2025 combined. So the fleet is not only growing in scale, but it is absolutely growing in efficiency

Etai McKellie, Analyst — TD Cowen

as well. That's great. And maybe to double-click on West Fraser, your first international expansion, your first new industrial and market outside of the Permian, what led to that pilot? I know it's kicking off in Q3. Walk us through like maybe that in a typical customer conversation of kind of how long does it take to, you know, enter a pilot? How should we think about West Fraser and that

Don Burnette, CEO

relationship in the next 12 months or so? Well, I can give you a reference point. When we started talking to Atlas, between when we started having the conversations and when we deployed our first driverless vehicle, it was just over a year. So that, you know, kind of gives you a sense of what the timeline was from a first customer perspective. The sales lead time and convincing companies that this is something that they should adopt, it varies depending on the sophistication of the customer. So I wouldn't say there's one size fits all, but the skepticism that we face is definitely dropping and having evidence of an existing fleet operating for a real customer on a daily basis is certainly a very compelling proposition that is bringing a lot more customers to the table. Now, we're a small company. We run a really lean operational budget, and so we can't serve all customers everywhere. We really need to pick and choose where we think we can have the highest impact, where we can provide the most value, and where it makes sense for us to gather more learnings. And so we ideally wouldn't want to expand into other opportunities. I mentioned oil and gas. Logging was always on the table, and we definitely wanted to go find kind of a marquee logging company. Wes Frazier, we just announced as our next customer in the industrial application. They're one of the largest wood products companies in the entire world. They operate in four countries. They have over 50 sites. And this is something that we're really, really excited to bring to the table. And then, of course, there's mineral and resource mining, predominantly in Western Australia. We've talked about in the past having a strong customer pipeline of interest in that region. It's a little bit farther away to get into the Western Australia market, but we think there's broad application for customers worldwide.

Etai McKellie, Analyst — TD Cowen

Terrific. And then switching now to long haul, a lot of focus on your target to kind of remove the driver and commence driverless long haul operations in late 2026. How are we on that? Any update you can share in terms of what goes right? What are some of the risks left to retire in that journey?

Don Burnette, CEO

Sure. So we've talked about our safety case completion process in the past. Effectively, our safety case consists of a series of claims about safety. So the truck will be safe in this scenario, in that scenario, and if this happens or if that happens. And it's a very long list. It's organized in a tree. And ultimately, you've got to check off all the boxes. And checking off the box really consists of three things. Gathering data in the real world to understand your exposure to that type of scenario, how often does it happen? That's a question you have to answer for everything that can happen. It's not enough to say, will this happen? You also have to say, this will happen at some frequency, you know, some number of miles. And then you have to evaluate your performance. We predominantly use simulation for this. So for all the possible scenarios and situations that can occur, we have a massive simulation database that we build, and we execute the system on that simulation database to test every, you know, pocket of corner case that we can possibly think of. And then the third component is what we call structured testing. You can think of this as track testing. So we take a sampling of those simulation situations, we go to the test track, and we set it up in the real world with real vehicles, with real actors, sometimes dummy actors, but with real physical things, and we evaluate that the track actually performs and matches as the performance we saw in simulation. Once we've done all of that for a given claim, we can check the claim off and say, OK, that claim is done. And so we've reached about 86% of the claims that are finished. We've done some of the long lead time ones that are a little bit more risky. And we've kind of front loaded those. And we do expect that kind of pace of closing safety cases to claims to accelerate towards the end of the year. And we're targeting hitting 100% and pulling the driver for highway operation at the end of this year.

Etai McKellie, Analyst — TD Cowen

Terrific. And not to skip too far ahead, Don, but we think about once you remove the driver, let's say we're in 2027, how should we think about the ODD migration from there? You know, one of the appeals of the trucking industry is, you know, could there be faster scaling and expansion of ODDs relative to like RoboTaxi? How are you thinking about that at a high level into 2027 and beyond?

Don Burnette, CEO

So you're definitely, you know, a lot of people like to look at the RoboTaxi deployment as a comparison. and there's a couple challenges with that. One, we're years farther advanced than they were when we first unveiled RoboTaxis as an industry, or we, Waymo, did many, many years ago. And so you're going to see natural acceleration of expansion just because the processes, the systems, the technology, everything is more mature than it was back then. And secondly, as you mentioned, the highway environment is very uniform. And so that will play into our advantage where most interstate highway routes look like every other interstate highway route. That's not true when you go into the cities, right? New York is much different than Boston, is much different than San Francisco, is much different than Dallas. So highways are much, much easier to expand. And then I would say the third thing, going back to the tech stack, we don't rely on heavy mapping in our stack. And that allows us to bring up new routes much, much more quickly. And we do expect expansion to be quick. I'll give you one metric on that. We are now bringing up new routes for Atlas. Now, the Permian is a 75,000 square mile area. So it's a massive, massive place, right? We don't drive, they don't drive everywhere within the region. They drive on specific routes and serve customers. Well sites tend to move around every three to four weeks. So you drill a well, you extract oil for three, four weeks at a very high rate, and then you put kind of a maintenance system there. Those are the pump jacks that you typically see when you look, you know, you see the pictures in the shows. and then they move on. So Atlas is constantly reconfiguring where the trucks are driving. We can bring up new routes within a matter of days. So completely from scratch, places we've never seen before, and that goes from nothing to full driverless validation within a week. So we think we can bring that rapid expansion to the highway environment and we think we should be able to expand very quickly once we actually hit go.

Etai McKellie, Analyst — TD Cowen

Interesting. So one week for new OD and Permian, and maybe not, you think a similar timetable for a while?

Don Burnette, CEO

I don't know if it's going to be, it took us a while to get to, you know, a matter of days. I mean, you are going to see an ODD expansion route by route, right? It's not going to be, but we should be able to bring up routes. I mean, we've been working on building this commercial network, which is now over 25,000 miles within the U.S. over many, many years. And so we do expect it to happen pretty quickly. I can't yet give a specific time frame.

Etai McKellie, Analyst — TD Cowen

So it sounds like you're on track for long haul, a lot of momentum in industrial and defense. Maybe high level, a question just on the numbers, how should we think about the path to positive gross margin, EBITDA, in terms of building blocks to get there over the next few several

Don Burnette, CEO

Well, it really comes down to scale. So this is a capex-intensive business, right? We're an asset-light business, but we ultimately do have to kind of fund the capital up front to deploy the trucks. And of course, we make revenue over a multi-year period for every truck that we deploy. And so the revenue is somewhat back-ended relative to the cost. And so there's a transition point where, on a kind of per-truck basis, we move to gross margin positive as the truck provides value over time to the customers. And so for us, we're not planning to expand our OPEX significantly over the next several years. We like to say, you know, you'll see our CapEx rise. We call that success-based capex, right? The more we deploy, the more people want our product, the more it's going to cost. And you're going to see some increased costs in the short term, but then the revenue is going to start to ramp, and there's going to be a crossover point sometime in the future. We haven't guided to any specific time frame for which that's going to happen, but we are always looking at this very responsibly. We're working to bring down the cost of the bomb, the bill of materials, which are one of the two main driving factors behind our gross margin. You have the actual hardware itself, and then you have the service and support that you pay on the back end. And we're always looking to make the trucks more efficient and more capable so that they require less support, less maintenance, less cost as we move forward.

Etai McKellie, Analyst — TD Cowen

Absolutely. And then how do we think about the regulatory environment within this? I think there's actually some recent developments. So I think we're pretty positive. Let me just talk a little bit closer, yeah.

Don Burnette, CEO

So this is evolving by the weeks, which is amazing. It used to be the same answer for like several years, right? So right now, it's a state patchwork of regulations. We feel very, very confident that within at least the Sun Belt, the southern portion of the United States, we have the freedom to operate driverlessly without any questions. And so our business is really not super concerned about the regulatory environment. That being said, a couple new announcements. Just a few weeks ago, California finally passed regulations for the allowance of self-driving trucks, which we are going to be taking advantage of and submitting our application to start driving in California. That completes the coast-to-coast network, which is a big kind of black spot previously. And then just this week, just last week, they announced in the transportation bill in Federal Congress, a preemption law that would effectively allow and permit self-driving trucks to operate across all 50 states with preemption over state regulations. So that hasn't been passed yet. Obviously, we're really excited about that. Whether or not that law goes into place this year, it's kind of inevitable. It will come, and I think that's going to put to bed a lot of the questions and concerns around regulatory. So all in all, regulatory is moving in the right direction. People understand how valuable this technology is to our economy, the necessity to continue to grow this industry in a place where drivers are depleting as opposed to coming into the business. And so, yeah, we feel really good about the regulatory situation. Terrific.

Etai McKellie, Analyst — TD Cowen

Maybe kind of thinking about the long-haul opportunity over the next five years, what do you think are the top two or three factors that will drive success for you over the next few years? Is it sort of, you know, the ODD expansion, pricing, customer relationships? What are the top things you must provide?

Don Burnette, CEO

I would start with commitment to safety, right? Everything hinges on safety in our industry. If it's not safe, nobody's going to adopt. From a customer demand perspective, we feel very confident that demand is there, right? There is over 3 million trucks driving on U.S. highways on any given day. And so the demand is clearly there. I think this is a business where those who adopt autonomy early will have significant advantage from a pricing perspective over competitors. And so I think there's just not going to be any risk of demand, provided that we continue to operate safely. And then for us, it's really about access to scale, ensuring that we have the supply chain, ensuring that all the components and pieces that are necessary to bring that safe product to the market are in place to really start turning the crank and just produce the vehicles. I think supply is going to be the limiting factor for the market, and that's a great problem to have.

Etai McKellie, Analyst — TD Cowen

And maybe speaking on supply, how important is it to sort of get technology fitted directly on OEM trucks as opposed to an up-fit model? How far can the up-fit model actually carry the industry?

Don Burnette, CEO

Yeah, I think you'll see a transition from the UpFit model, which is what companies are currently operating with today, versus the kind of OEM integrated model. The OEM integrated model will eventually come. What form it takes still remains to be seen, and what timeline that is still remains to be seen. But when you say how important it is, I would say it's not very important, because in the time frame that I think the OEMs will be able to deploy this technology, UpFit can absolutely handle any amount of scaling that we will be able to produce in that time frame. We could upfit thousands and thousands of trucks per year if that was the demand, and we were ready to do that. So by the time we kind of get to that scale and saturate the sort of upfit model, I think the OEMs will all be in a place where they can offer autonomy directly on the line. Terrific.

Etai McKellie, Analyst — TD Cowen

And what other maybe opportunities within the kind of physical AI landscape might you pursue in the next several years? Could there be licensing opportunities? Yeah, I mean, several years is a tough one, right?

Don Burnette, CEO

We already have three verticals that we're focusing on, so I would say we have our handful, and we're pretty excited about going after those markets. But this technology absolutely is generalizing, right? It's no longer specific to a given use case the way it used to be, the way we used to think about it. And so I would say that the entire mobility space is fair game for Kodiak's technology. There's no reason that our AI can't be in robo-taxis. There's no reason that our technology cannot be licensed for driver assistance technology or deployed in medium-duty trucks or last-mile delivery. Like, all of those are fair game. I think the mistake that a lot of companies make is to kind of cast too broad of a net before they've actually established themselves in any one market. And so we really want to focus on the three verticals that we've been pushing hard toward. And after we've kind of got a stronghold in those in those areas Then we'll start to look at what are the other tangential areas we could potentially expand to great

Etai McKellie, Analyst — TD Cowen

And we have 40 seconds. We must sneak one more in Don the one vertical we didn't cover in detail

Don Burnette, CEO

But just a couple words on defense. Yeah, look, I mean defense defense is an exciting opportunity for Kodiak One maybe in the 30 seconds I have left the one thing I'll just mention to folks is that in the in the federal budget the 2026 budget for for the Department of War's defense autonomy group was about $225 million. The proposed budget for that same group for FY27 is over $50 billion. So it just goes to show you where people are starting to direct funds when it comes to the defense budget, and autonomy is at the absolute top of the list, and we think there's going to be significant tailwinds, not only for Kodiak, but the entire autonomy industry as we move into the next couple of years. And so we absolutely plan to, with our partners, General Dynamics Land Systems, take advantage of that budget increase and go after some significant contracts in the near future. Terrific.

Etai McKellie, Analyst — TD Cowen

Don, thanks so much. Please join me in thanking Don for the conversation.