Beyond The Highway: Kodiak AI’s Roadmap to Autonomous Trucking and Physical AI

Speaker 1:
Welcome to TV Cowen Insights, a space that brings leading thinkers together to share insights and ideas shaping the world around us. Join us as we converse with the top minds who are influencing our global sectors.
Itay Michaeli:
We're live at TD Cowan's 54th Annual TMT Conference. I'm Itay Michaeli, Autos and Mobility analyst. I'm delighted to have Don Burnette, Kodiak AI's founder and CEO with us today. Don, welcome to the program.
Don Burnette:
Thanks for having me.
Itay Michaeli:
Great. Maybe just to first an introduction on Kodiak AI, what you do, what you're involved with and the exciting AV and physical AI space.
Don Burnette:
Yeah, sure. Kodiak is a physical AI company. We're pushing the forefronts of AI in the physical world, specifically as it relates to the mobility sector. We've been around for about eight years and traditionally have focused on the commercial side of automation, meaning we serve long-haul trucking applications for our autonomy product. We have a single AI driver that is actually capable of driving in both structured and unstructured environments. So, unlike a lot of other autonomy companies, we don't only focus on road applications, we also have off-road applications where we serve customers. For instance, in the Permian Basin of West Texas, we announced a recent customer in the logging industry based up in Canada, and we also serve and supply autonomy to the defense industry. We've been active in the defense space across multiple programs, across multiple departments of the U.S. military, starting with the Air Force. And then we've worked with the Army. Most recently, we worked with the Marines. And so, our technology is really all about automating mobility across various environments, not constrained to specific structures or specific types of roadways, highways, or surface streets. We kind of do everything in between.
Itay Michaeli:
Terrific. So, you are live. You have driverless trucks operating in the Permian with your customer Atlas. How long have you had those? How many trucks are out there today and what are you learning as you've been able to grow this customer?
Don Burnette:
Yeah, it's been a long time coming for us. I've been working in this industry for nearly 18 years. And so, as many listeners will know, self-driving has been promised for multiple decades. And as you pointed out, it's finally here. We've actually now deployed, as of the first quarter, 28 driverless trucks. And when I say driverless, I mean nobody in the cab, no required monitoring from a remote station. These trucks are truly out there on their own. They're deployed to a company called Atlas Energy Solutions. They're out there operating on a day in and day out basis. So, it's a 24/7 operation. There's no downtime for these vehicles. And so, it's really important that they actually are up, they're ready, they're capable.
And we've been doing that over the last 18 months. We've starting to really gain some significant experience, and the efficiency of these vehicles is improving dramatically. In the first quarter alone those vehicles drove more hours of driverless operation than in the entire 2025 calendar year. So, we're really expanding the footprint. Now, in terms of what are we learning, it's really about the interactions with the customer. How do they use the vehicles? How do they use the vehicles to pick up the goods? How do they drop off the goods? Who do you need on either side and how do you handle those workflows? Those are the kinds of things you can't really learn when you're testing, until you've actually deployed a product. And now we've reached sufficient scale, still small and early days, but we've reached sufficient scale where this is a real operation. We've got so many trucks coming and going continuously throughout the day that it really needs to be a well-oiled machine. And these are lessons that we can then take to other customers as we start to bring this technology to a more broader customer set.
Itay Michaeli:
And speaking of other customers, I think you recently announced a pilot in another vertical in Canada with West Fraser. Talk to us maybe about the opportunities beyond the Permian, both with the pilot with West Fraser and maybe even other verticals around the world where your technology could be deployed.
Don Burnette:
Yeah. We're super excited to announce this customer in West Fraser. They're one of the largest wood products companies in the entire world. They operate in four countries, primarily based in Canada. So, for us, this is our first international expansion and capabilities. And at the end of the day the logging industry is one of those unique use cases where the roads and jobs tend to be very remote. They move around a lot, and they work in environments that are completely unstructured. This is an area where Kodiak's technology really shines. And so, we're super excited to go into the logging industry and demonstrate that we can really automate that entire process of transporting all of the goods, all the logs, whether those are raw materials or process materials to where they need to ultimately get. We can do that safely. We can do that efficiently. We can do that at a competitive cost point.
Beyond logging, there's also opportunities in areas like mineral and resource mining. You can find that all over the world, but one of the largest opportunities is within Western Australia. And so, we're really thinking broadly globally about this deployment and where we think we can bring this technology and provide value to trucking companies everywhere.
Itay Michaeli:
Yeah. Great to see the progress, Don. How about defense? Another area that's getting a lot of attention, of course, in physical AI. I know you've had some announcements there recently. Talk to us about the opportunity you're seeing in defense.
Don Burnette:
We just announced a strategic partnership with General Dynamics Land Systems. And they are, as everyone will know, one of the marquee prime developers of ground vehicles for the U.S. military and allied militaries around the world. And so, we're super excited to be working with them to bring the capabilities of autonomy to the future of the war fighter. And that's something that I think is really picking up a lot of steam and momentum. It's a little bit nascent over the last decade. We've seen a lot of focus on aerial autonomy, but I think ground autonomy is going to have its day. The technology is here, it's safe, it's reliable. It can be deployed and built at scale. We've built up the infrastructure and the supply chain that we need to bring that to the market. And so, partnered with General Dynamics, I think we're really going to be able to make a big impact in the defense space in the coming years, and we look forward to working with the department on future programs.
Itay Michaeli:
Terrific. And then on long haul on highway, I think you're targeting to actually go driverless by the end of the year in late 2026. What's the path to getting there? And once you get there, how should we think about the opportunity to scale driverless trucks on road over the next couple of years?
Don Burnette:
We're in the process of completing our safety case. That's what we call our exhaustive statistical data-based analysis of the capabilities of our system relative to a human driver. We're 86% of the way to 100, and we expect to hit 100 by the end of the year and start to deploying driverless in the highway environment starting in 2027 and beyond. And so, for us, this represents the largest opportunity of autonomy effectively in the world. Trucking and logistics is one of the largest industries in the world. And so we're really excited about the opportunity to expand. We already have 25,000 miles in our commercial network spanning all across the U.S. highway system predominantly in the southern states where we have regulatory approval today to go driverless and expand that fleet. And so yeah, we've signed up some of the best automotive suppliers in terms of partners, folks like Bosch. We're working with Roush as our manufacturing partner. We've put all of the pieces in place to secure the supply chain and enable us to scale once we pull the driver later this year.
Itay Michaeli:
Great. And one of the questions I often get is, you're operating long haul industrial defense. What's the tech stack angle? You have, I know the sensor pods, but a lot of AI usage. What's about the technology that makes it so transferable to all these different domains?
Don Burnette:
The key to transferability and generalizability is that it has to be an AI system. It can't be a rules-based handcrafted system, that would become too rigid as you start to change the parameters of the environment. We've been able to train our end-to-end AI solution across a very wide variety of data sets, and that's really been the key to unlocking generalized driving capabilities. We've spent years and years and years gathering, collecting and curating data from all types of environments, highway environments, surface street environments, urban environments, off-road, dirt road, in the field, through the rainforest, cross the sand dunes. You name it, we've gone out and collected that data. And we bring it into one dataset and train one system and then that system can then drive wherever it needs to go. And we don't rely on pre-mapping, high-definition mapping of the environments. This is particularly important in unstructured environments where there really isn't anything to map. So, that's how our system, it's really an end-to-end AI system. That's how it allows it to go across all of these different domains.
Itay Michaeli:
Terrific. How about the regulatory landscape? A lot has happened recently. Is it becoming more accommodative to growth? There's still challenges ahead. What are you seeing there on the regulatory side?
Don Burnette:
Regulators get it. They understand the importance of this technology to our economy, the importance of this technology to safety. We've seen broad support at every level, municipal, state and federal in support of rolling out this technology safely and responsibly. From a state perspective, there's about 29 states that have already given us the green light to operate. Now we've added California to the mix as of a couple of weeks ago, which is incredibly exciting. It's been a long time coming. We worked on that for many, many, many years, and just recently there's been a bill proposed in the federal Congress that would allow for autonomous vehicles, including trucks, to operate across all 50 states. And so, the momentum is there. We're really excited to support these initiatives. We think we have the backing of the regulatory framework, and there's really nothing blocking us from expanding and growing this business.
Itay Michaeli:
So Don, what's your vision for Kodiak over the next two or three years? And in that journey, what do you see as the biggest, maybe upside opportunities and maybe some additional risks that you have to manage through?
Don Burnette:
It's been a transition for us as a company moving from more of an R&D-centered kind of developed tech company, really an engineering powerhouse to, how do we deploy? How do we scale? What are the customer needs? What is the product we need to build? And so we've really been iterating from that perspective for the last several years, and now we're at an inflection point where this technology is going to take off and it's going to start to accelerate. And so the next couple of years is all about staying focused on scale, getting the product into the market, making sure the customers are happy with the product, and then bringing on more, adding more and just scaling up the product. Now ultimately, if you look beyond that, if you look at the next decade, I think you're going to start to see autonomy become ubiquitous across the entire transportation network, and Kodiak is going to be powering those vehicles.
Itay Michaeli:
Great. Maybe lastly, I know you have a lot on your plate, but how do you think about Kodiak's role broadly within physical AI? You've got a lot of verticals already you're growing into. Are there other opportunities you might be exploring or think about exploring in the next few years, licensing, other verticals? How do you think about the broader opportunity in physical AI?
Don Burnette:
Physical AI is really defined by two factors. There's the lack of access to data, and then it's low power. If you think about data center scale AI where these massive models with hundreds of billions of parameters are running in the data center and they require a significant power budget, we don't have that power budget in the physical world. We call that at the edge. Edge computing is very, very low power.
The second piece is, these data center scale AI systems have the internet of data, which is accessible to everyone and it's at your fingertips. That's different in the physical world. You have to go actually capture the data. Those data sets don't exist, and they're typically bespoke to the use case. And so, we already have a massive headstart and advantage in the physical AI space, because we've pioneered working on low power AI systems and we also have access to the data. And so, as I said before, our data set is generalized across all of transportation, and we definitely see a world where Kodiak driver could power passenger vehicles. It could be a driver assistance solution, it could be licensed. There's really endless possibilities for how it could be deployed in the future.
Itay Michaeli:
Exciting times, Don, we look forward to continuing to follow the progress. Thank you so much for joining us at TD Cowen's 54th Annual TMT Conference. It's great to have you.
Don Burnette:
It's my pleasure.
Speaker 1:
Thanks for joining us. Stay tuned for the next episode of TD Cowen Insights.