Kodiak AI, Inc. (KDK) Earnings Call Transcript & Summary

Good day, and welcome to the Kodiak Second Quarter 2026 Earnings Conference Call. [Operator Instructions] Please note this event is being recorded. I would now like to turn the conference over to Mr. Steve Philistin, Senior Manager, Investor Relations. Please go ahead.

Good afternoon, and thank you for joining Kodiak's Second Quarter 2026 Earnings Call. I am Steve Philistin, Senior Manager, Investor Relations. Joining me today are Don Burnette, Founder and Chief Executive Officer; Mr. Surajit Datta, Chief Financial Officer. Earlier today, we issued our earnings release and posted our earnings presentation, both of which are available on the Investor Relations section of our website. Before we begin, please note that today's discussion will contain forward-looking statements regarding our business, financial performance, outlook and future operating plans. Actual results may differ materially from those discussed today. Please refer to our earnings materials and SEC filings for a discussion of factors that could cause actual results to differ. Any forward-looking statements that we make on this call speak only as of the date of this call, and we disclaim any obligation to update any forward-looking statements, except as required by law. We will also discuss certain non-GAAP financial measures. Information regarding our non-GAAP financial measures, including reconciliations to the most directly comparable GAAP measures can be found in our earnings release and supplemental materials available on our Investor Relations website. With that, I'll turn the call over to Don.

Good afternoon, and thank you for joining us. Q2 was a quarter of significant commercial validation and increasing scale for Kodiak as we achieved important milestones that further strengthen our leadership position in commercial driverless trucking. We expanded the world's largest customer-owned driverless truck fleet to 35 trucks, operating with no human in the cab. We made significant progress toward our targeted long-haul driverless launch by the end of the year, increasing our autonomy readiness measure to 91% as of July. We also introduced our seventh generation platform, further improving the performance and reliability of the Kodiak driver. Together, these accomplishments strengthen the foundation of our next phase of disciplined growth as we continue deploying physical AI driverless solutions at commercial scale. Turning to our second quarter results. We deployed 7 additional driverless trucks, bringing our total driverless deployment to 35 and further growing the largest customer-owned driverless truck fleet in the world. As of quarter end, those trucks have driven more than 40,000 paid driverless hours, an increase of 71% from the end of the first quarter and up more than 20x year-on-year. Over the same period, we delivered over 300,000 tons of freight, bringing cumulative loads delivered to more than 20,000, representing 32% growth from the end of the first quarter and 150% year-over-year. Together, these results demonstrate that the Kodiak driver continues to deliver increasing value to our customers while advancing commercial deployment. We are also moving at high velocity toward our targeted driverless long-haul launch by year-end. As of July, our long-haul autonomous readiness measure, which measures the percentage of safety case claims we have completed for our targeted long-haul launch surged to 91%, a 5-point improvement in just 3 months. At 91%, we have now successfully closed out the vast majority of our long-haul safety claims, including several of the most technically challenging claims that historically require the longest validation cycles. These include significant progress across our safety critical vehicle control architecture, including enhancements to our throttle, brake and steering control systems and continued advancement of our redundant steering capabilities to improve system fault tolerance. We've also made meaningful progress in our probabilistic risk assessment framework, strengthening how we model, analyze and quantify safety risk across real-world operating scenarios, which are essential for highway driving. This progress reflects the investments we've made over the past several quarters in expanding our systems engineering organization, enhancing our AI-enabled development and validation tools, growing our fleet of dedicated development trucks and increasing the scale of our structured track testing program. These investments are now translating into measurable results. One example of this improved tooling is our revolutionary AI-powered Breakpoint technology, which allows our engineers to actively seek out rare and challenging scenarios, accelerate development and close the limited remaining claims in our safety case. BreakPoint can now execute over 1 million simulations per hour and runs on a minimal compute budget of around $0.001 per sim, another example of Kodiak's capital-efficient approach in action. We combine this large-scale simulation with targeted closed course testing and real-world validation, thereby generating data more efficiently, validating safety performance faster and systematically closing out safety claims with a high degree of confidence and speed. Based on the momentum we are seeing today, we expect to continue making meaningful progress throughout the third quarter and continue to target driverless long-haul launch by year-end. We also continue to refine our operational readiness for long-haul deployment. After successfully launching service with rail transport in the first quarter, we recently expanded operations to 5 round trips per week. This expansion reflects the confidence our customers gain as they integrate the Kodiak Driver into their operations. Progress towards our long-haul launch is happening at an opportune moment where a number of broad industry trends are reinforcing the long-term trucking opportunity. Most notably, freight rates are rapidly recovering as the trucking industry faces new driver capacity constraints due to economic and regulatory changes, including a regulatory crackdown on non-domiciled drivers. We believe these trends further reinforce and even grow the long-term opportunity for the Kodiak Driver. We had another strong quarter in our industrial vertical, increasing Atlas' fleet of Kodiak-equipped trucks to 35. Atlas has selected Daimler Truck North America's Western Star platform as it continues to expand its Kodiak-powered fleet. Our upfit partner, Roush, has already begun producing these vehicles, which integrate our seventh generation autonomous platform onto the Western Star 49X for our growing deployment with Atlas. By the end of the quarter, we expect Atlas to be operating high 30s number of driverless trucks. As previously stated, as we move through the remainder of the year, we expect to ramp production and deploy a similar number of driverless trucks in the second half of 2026 as we did in the first half. This transition positions Atlas to become the first customer in the world to operate a driverless fleet across multiple truck OEM platforms. We believe the experience gained through this deployment will be invaluable as virtually every large trucking fleet operates vehicles for more than one OEM. We also recently agreed to a firm delivery timetable with Atlas that will allow us to establish a more coordinated production and delivery schedule. This new delivery timetable will better align our deployment with Atlas' growing customer demand while supporting expansion into a broader geographic footprint across the Permian Basin. The updated agreement further strengthens our partnership, improves planning and execution for both companies and reinforces our shared commitment to successfully scaling driverless operations. We continue to expect to complete the initial 100 truck commitment by the end of the first half of 2027. We recently expanded our driverless truck deployment with Atlas to service 2 of Atlas' major Dune Express load-out facilities simultaneously. These 2 depots are an approximate 90-minute drive apart and service wells spread over 3 hours apart, allowing Atlas to efficiently serve a broader range of customers over a much wider geographic area in Texas and New Mexico. In addition, Atlas now successfully serves customer well sites solely with Kodiak trucks, which requires a high level of reliability and assurance. If a well site doesn't get sand on time, it shuts down, making these deployments increasingly mission-critical. Lastly, our first international pilot kicks off in a few weeks in Canada with West Fraser, one of the world's largest wood products companies. We will haul flatbed trailers to demonstrate the Kodiak Driver's capabilities in logging operations, another of trucking's most demanding environments. We are excited to demonstrate the versatility of our system and expand the industrial use cases where the Kodiak Driver delivers value to customers today. Our defense vertical continues to build momentum. We have a growing roster of world-class defense partnerships, including our strategic collaboration with General Dynamics Land Systems, who we continue to work closely with on new opportunities. Collaborating with companies like General Dynamics puts Kodiak in a much stronger position to compete for future autonomous ground vehicle contracts, contracts that we believe could make a significant positive impact on revenue in the coming years. We were also selected for Phase 2 of the Defense Innovation Unit's Robotic Operation for Autonomous Delivery and Sustainment or ROADS program. We participated in the ROADS demo in July. ROADS aims to leverage customer-proven autonomous vehicle technologies like Kodiak's for the department's non-tactical vehicle fleet. Given our proven technology, we believe we are well suited to address this opportunity, which reflects the Department of War's ramping investment in autonomous and dual-use technologies. The depth of our defense portfolio also continues to expand. Our team is continuing to execute on the ROGUE-Fires program program with the U.S. Marine Corps. This work builds directly on Kodiak's prior work in military autonomy, including our earlier robotic combat vehicle engagement with the U.S. Army and participation in high-profile Pentagon demonstrations such as the Defense Innovation Unit's Project G.I. and the Army's xTechOverwatch program. Moving to technology. I'm excited to announce that Kodiak's Gen7 platform is debuting on the Western Star 49X. Gen7 reflects the same product philosophy that has guided Kodiak from the beginning, while incorporating everything we've learned for over 18 months of driverless operations in one of the world's most punishing driving environments. The result is a product that is more efficient, modular, reliable, serviceable and practical for commercial fleets. These improvements reflect our continued focus on building technology that addresses real customer needs. Gen7 has nearly 50% more compute power than the previous generation and features a more compact form factor that simplifies integration, enhances reliability and supports a broader range of vehicle configurations. In fact, Gen7 is the AV trucking industry's first platform compact enough to integrate into lower-cost day cap configurations, which we believe are the vehicles many of our current and target customers want to deploy. Gen7 also builds on our operational experience to increase the overall reliability of the Kodiak Driver. Initial mechanical stress testing indicates that our Gen7 sensor pods and compute enclosures will have a nearly 50% greater operational lifetime than prior models. We've paired the Gen7 hardware with the latest release of our driverless software, which includes significant improvements over the prior version and is already increasing our customers' productivity. This latest version includes new AI perception models, better handling around obstacles and debris and improved vehicle controls. Our driving in the Permian has become more confident with overall efficiency increasing more than 30% from the prior software release. The second quarter saw significant regulatory progress for autonomous trucking. At a state level, California's Department of Motor Vehicles finalized new regulations that, for the first time, extend its autonomous vehicle permitting program to include heavy-duty trucks. This regulatory structure creates a clear path to commercial driverless deployment in California for both intrastate and interstate operations and will ultimately enable us to offer coast-to-coast driverless service. We moved quickly to apply for a permit and anticipate receiving one in the coming weeks. Additionally, we continue to see significant progress at the federal level. In May, the House Transportation and Infrastructure Committee advanced the BUILD America 250 Act, the latest Federal Surface Transportation bill. Excitingly, this bill includes bipartisan language supported by the AV industry, labor and other key stakeholders that would establish a comprehensive federal policy framework specifically to regulate the safe deployment of autonomous trucks nationwide. The bill would also permanently solve the warning triangle issue without the need for an exemption from U.S. DOT. While there's still work to do for the BUILD America 250 becoming law, the bipartisan AV trucking language lays down a clear marker for the future of federal AV truck regulation. We are, therefore, increasingly confident that autonomous trucking will be included in the final surface transportation legislative package when it becomes law. Our growing driverless deployment, progress toward our long-haul driverless launch and ability to integrate onto multiple leading OEM platforms demonstrates that Kodiak is prepared to scale. Across our long-haul, industrial and defense verticals, we continued to strengthen the Kodiak Driver, expanded customer deployments and built the operational experience needed to scale efficiently. Every deployment, driverless hour and product improvement strengthens the Kodiak Driver across all verticals. With more than 18 months of commercial driverless operation under our belt and a growing ecosystem of customers and strategic partners, we believe Kodiak is well positioned for our next phase of growth as we prepare to launch long-haul driverless by the end of 2026. I want to thank the Kodiak team, our customers and partners for their continued execution and trust. Now over to you, Surajit.

Thank you, Don, and good afternoon, everyone. I'm pleased to share Kodiak's financial results for the second quarter of fiscal year 2026. We delivered another strong quarter of both operational and financial performance, successfully executing against our strategic priorities while expanding driverless deployments, growing recurring revenue and maintaining disciplined spending. We ended Q2 FY 2026 with 35 driverless trucks, in line with our expectations and up from 28 in the prior quarter as we continue to expand deployments with Atlas. These trucks continue to operate commercially without anyone in the cab and are fully integrated into our customers' day-to-day fleet operations. Revenue for the quarter was $3.5 million, representing 91% quarter-over-quarter growth. The revenue increase was primarily driven by continued expansion of Driver-as-a-Service revenue and also benefited from a onetime revenue contribution of slightly over $1 million from our autonomous trucking demonstration program with DriveOhio. Excluding this demonstration revenue, our underlying Driver-as-a-Service business continued to deliver strong sequential growth driven by growing commercial deployments. GAAP operating loss for the second quarter was $43.7 million. Non-GAAP operating loss, which excludes stock-based compensation, was $37.3 million, primarily reflecting continued investment in research and development and operational support as we scale our deployments and long-haul readiness. We incurred capital expenditures of approximately $3.9 million, primarily related to AV hardware commitments, long-haul development and continued commercial deployments. Turning to cash flow. Q2 free cash flow was negative $38 million, outperforming the low end of our guidance range. Our spending during the quarter primarily reflected continued investment in commercialization expansion, initiatives designed to reduce the cost of Kodiak driver over time and further the development of our long-haul driverless capabilities, partially offset by continued increase in revenue. We ended Q2 with cash, cash equivalents and marketable securities of $151 million, providing us with the financial flexibility to continue executing our commercialization strategy while maintaining a disciplined approach to capital allocation. Turning to guidance. For the third quarter of fiscal 2026, we expect driverless truck deployments to increase to high 30s. During the quarter, we will begin transitioning the Atlas driverless build to the new OEM truck platform, consistent with our customers' fleet strategy. The transition further demonstrates the flexibility of the Kodiak driver to operate seamlessly across OEM platforms and expands our opportunity to scale future customer deployments and increases the breadth of our customer base. As we initiate this transition, we expect more modest driverless truck additions during the third quarter while we integrate the new truck platform into commercial service with the customer. The deployments are expected to accelerate in the fourth quarter as we continue scaling deliveries. We expect third quarter free cash flow of negative $39 million to negative $41 million, primarily driven by continued investment in long-haul driverless testing and development and growing commercial deployment activities. In addition, we are narrowing our full year 2026 free cash flow guidance to negative $155 million to negative $162 million compared with our previous range of negative $155 million to negative $165 million. This updated guidance reflects our strong first half execution and our continued focus on disciplined capital allocation. We believe our continued commercial deployments, progress toward our long-haul launch and disciplined financial execution position Kodiak well for its next phase of growth. Operator, we are now ready to take questions.

[Operator Instructions] And our first question for today will come from Andres Sheppard with Cantor Fitzgerald.

Congratulations on the quarter and all the great progress. Don, I just wanted to maybe touch on the cadence for Atlas. So I think in the prepared remarks, you gave us a sense of kind of Q3 deployments. You gave us a sense that -- sorry, you reaffirmed that you're on track for the entire deployment by middle of next year. Just trying to break that out a little further. Any more cadence or granularity you can give us? How should we think about Q4 and then the cadence for next year?

Yes. Thanks, Andres. Yes. I think we said it in the remarks, right? We're steadily ramping up the new platform. There's obviously a lot of work and validation that goes into bringing the new platform up to driverless level quality, and this is something that we've been working on for the last little while, and we're excited to finally be deploying that new platform here in Q3. So as we said before, we're going to start small with the Q3 deployment on the new platform, and then we're going to quickly start to ramp up through Q4, Q1 and Q2 as we get to the final 100. I don't have anything more to add other than we plan to deploy a similar number in the second half of the year as we did in the first half of the year.

Okay. That's very helpful. Appreciate it. And maybe just a quick follow-up, maybe a 2-part question. With the launch of the Gen7 hardware, I guess, what does that do to the scale? Any opportunities to accelerate that? How are you then thinking about your fleet? Are you going to perhaps discontinue some of the previous hardwares and just move over to the Gen7? Just trying to understand how you're thinking about that. And then the last one real quick is I see that you applied for the California autonomous permit. So how are you thinking about opportunities in California as well?

Yes. So we don't have any plans to discontinue the usage of our previous generation platform. It's out there working day in, day out 24/7 for our customer today, and we built that system to be incredibly reliable, and it's really been showing the quality that we put into it. That was a program that we developed over several years, and it's still going plenty strong, and we continue to -- we will continue to support that platform for the coming years and ultimately, the lifetime of those trucks. So we don't plan to go back and retrofit, so to speak, the Gen7 on previous platforms. But all new trucks that we deliver starting in Q3 will have the new Gen7 platform with all of the upgrades and improvements in efficiency, horsepower, compute power, space savings, reliability, et cetera, that we had mentioned before. And then in terms of California, we're really excited about where California stands. We await the permit, as we said in the remarks, over coming soon. We're excited that, that opens up coast-to-coast potential operations for our long-haul release. We are focused on the Texas area for the next several quarters. But ultimately, we do believe that the longest of long-haul routes are the ones that are the most impactful for our customers. And of course, we're already talking to many customers in California who are excited to take advantage of the safety benefits that the Kodiak Driver offers.

And I'll just to add to what Don mentioned about the Gen7. In earlier conversations we have in our earlier earnings calls, we have mentioned we are taking steps to enhance the efficiencies of the AV hardware stack. This is one of those steps in that direction as we try to drive down our AV hardware costs by increasing efficiencies and productivity. So this is one of those steps in that direction as we will continue to drive down the AV hardware cost.

The next question will come from Itay Michaeli with TD Cowen.

6 First, Don, maybe just to go back to the defense opportunities you talked about in the prepared remarks. Just curious, one, what kind of milestones should we be expecting in the next maybe 6 to 12 months? And second, like how to size the opportunity for Kodiak over the next few years?

Yes. Our bullishness on defense continues to be really strong as it has been in the previous quarters. That being said, visibility into the sizing of these programs, the availability and when they ultimately come is very difficult to predict. And so we continue to be fairly cautious in giving any suggestive guidance around what the sizing of these programs look like. I think from our perspective, we see defense as a long-term strategy for the company, and we're excited to partake in programs that are available to us, the ever-evolving landscape of ground-based autonomy for military purposes. We're making great progress on the programs that we are a part of. And as I said in the remarks, with our partner, General Dynamics Land Systems and others, we're looking forward to applying to additional programs as they come down the line. But I can't give any specifics on kind of the scoping or size or expectation in terms of timing of when those will land or when those revenues will come in. We kind of look at that as purely additive to our core business in deploying our industrial and ultimately long-haul vehicles starting in 2027. And so we don't have anything more specific to add to that.

Okay. That's helpful. And then as a follow-up, going back to Gen7. One, any way to quantify the impact from cost or unit economics as you scale the new platform? And does this incorporate any of the benefits from the Bosch development that was announced earlier in the year?

Yes, it did. So this -- the Gen7 platform is part of all our initial initiatives and engineering to enhance design enhancements. And while we can't get into specifics, but we can say we're starting to cross into double digits of benefits from these pieces of the hardware as we go along and we are continuing to making those investments, right? If you think about our overall how to drive the AV hardware stack strategy, there are 3 drivers. One is obviously the engineering enhancements, including Gen7 platform, and there are a few other things we are working on. The second is how we create greater efficiencies through our global supply chain organization. That's part of the Bosch you are referring to. That's still nascent. We have started developing prototypes with them, and we'll start seeing more of those benefits into 2027 and beyond. And the last piece will be just the scale of production, which will drive more cost optimization. And just going back to defense, also, we wanted to remind, we had talked about it in the last earnings call, the Pentagon budget for the Defense Autonomous Warfare Group, I think the preliminary numbers were being talked about is $50 billion for 2027 compared to $225 million for 2026. So we believe this will benefit all players like us as we ramp into 2027 and beyond as these contracts gets allocated out.

The next question will come from Jim McIlree with Chardan.

Atlas on their call when they talked about the deployment of the 100 trucks, it kind of sounded like some of these trucks were going to be on the road or that it seemed like they were assuming that the long-haul case would be -- the safety case would be complete, and that's where some of the trucks would be deployed. Is that correct? Are some of these trucks in this 100 truck order contemplated to be used on road?

Well, I think the important thing to remember about Atlas is that they own and control these trucks. So ultimately, how and where they operate the trucks, which zones of -- and which regions of the Permian out of what loadout zones and what well sites to deliver to is ultimately their decision. We are working to close out our safety case for highway or long-haul operations by the end of this year, at which point then once we've kind of approved a given operational domain for driverless operations, then, of course, it's our customers' discretion as to whether or not they want to use the trucks in that mode or otherwise. And so we are working with them on geographic expansion. I think they talked about that. We're excited to continue geographic expansion. We've already begun expanding broadly throughout the Permian. That includes additional types of roads. But this is ultimately their decision as to how they operate the trucks and where they want to utilize them.

Got it. And on West Fraser, can you remind us again how long you think that demo or test might last and then the time frame to a decision after the demo is complete?

Yes. Jim, yes. So we expect the the pilot to start in the next few weeks, and it will last for a few weeks in Q3. And that could set up a decision -- discussions with West Fraser on a potential contractual agreement over the next few months.

The next question will come from Ryan Sigdahl with Craig-Hallum Capital Group.

In the Gen7 hardware platform, you see 50% greater operational lifetime versus prior generations. Are you willing to quantify what the previous generation was before? And then I guess, what this one implies? And then secondly, on the Gen7, Surajit, you said cross the double digits of benefits just now. I'm not sure what you're referring to there. So some clarification would be helpful.

I'll take the first part, which -- yes, so we intend the platform to outlast the useful lifetime of the truck. A typical lifetime of the truck in the Permian is around 4 years of operation, and we expect our Gen7 platform to be able to exceed the lifetime of those trucks.

Yes. And Ryan, what meant by those -- the double-digit reduction, this is our -- we're talking of our AV hardware stack, how much that costs on the truck. Obviously, we are still subscale and we are still building out a handful of trucks. But the overall Gen7, including compute redundant system, all those benefits, we expect that would result in double-digit reduction -- low double-digit reduction in AV hardware costs from the baseline cost.

Got you. Maybe just a follow-up, Don, on that. So I guess, does that imply that the 35 Atlas trucks now only have something like a 2-year useful life, half of what they kind of need?