Good morning, everyone. I'm Greg Konrad with the Jefferies Aerospace, Defense, and Airlines Equity Research Team. Welcome to the second annual Jefferies Defense Tech Summit. Today, we are very lucky to have Ben Wolff, CEO of Palladyne AI, joining us today. With that, we'll kind of get right into questions. Maybe for some of those who aren't as familiar, how do you define Palladyne AI's core market? I think that's probably shifted over the past six months, but what problem are you trying to solve, and how do you think about just the company overall?
Thanks for having me today, Greg Konrad. First and foremost, Palladyne AI is an embodied artificial intelligence company. We are focused on software that makes machines smarter, so that machines can operate in the real world without humans having to direct every single action or reaction, and without having to pre-program what the machines do. Think about everything, from drones in the air to industrial robots in a factory, and everything in between. We make the software that makes those machines able to think, reason, and act in real time. Importantly, we do that on a decentralized basis, so that the machines don't have to be continuously connected to the cloud. All of the intelligence actually lives on the machine. That's our primary mission and focus. Now, having said that, in the defense sector, we are also vertically integrated.
We concluded that having the greatest autonomy software in the world could only get you so far if you didn't also have hardware platforms that were optimized to be able to leverage what that software can do. We now are in the business of designing, developing, and producing missiles, UAVs, uniquely mission-capable UAVs for the defense world. We also make components for some of those UAVs and for other big military programs. For example, in our factories in Michigan, we manufacture components for the F-35, and for the Bradley, and for the F-22, and propulsion systems for missile programs, a wide variety of manufacturing. We're really focused on trying to do what the Department of War has said it's focused on, which is, number one, collaborative autonomy for weapons systems, and number two, sovereign manufacturing in the U.S., so that we reduce supply chain risk.
That is how we're vertically integrated on the defense side. Again, all of that is in support of our embodied artificial intelligence software.
Maybe just kind of putting some numbers, or no numbers at least, just directionally, how large is the TAM for AI-enabled robotics across defense or industrial, and where do you see the market maybe adopting autonomy the fastest?
I think that the market for embodied AI covers the universe of machines that humans would like to be able to become fully autonomous, not just remote controlled, not pre-programmed. There's some great analogies that I've lived through over the course of my career, and I know, Greg, you have as well. If you take a look at the early days of Microsoft, when Microsoft was first bringing its operating system software to market, if you look at what the early TAM estimates were for PCs and for their software, it was infinitesimal compared to the billion-plus desktop PCs that exist today. You can take a more recent example. I come out of the wireless sector. You know that I was the co-founder and CEO of Clearwire, which was a nationwide wireless carrier that now its assets serve as the backbone for T-Mobile's wireless network.
I saw a transition happen in the wireless space. If you took a look at basic flip phones that Motorola had that were the market leader, and then you see what happened when Steve Jobs introduced the software for the iPhone and the whole iPhone package. The early estimates for what Apple's share of the phone business would be, the wireless phone business, it was infinitesimal compared to what it is now. What do those things have in common? It's that software enabled an entirely different user experience, and the whole TAM for the entire industry, in fact, the definition of the industry changed as a result of what Microsoft did with PCs and what Apple did with phones. I think we're at that same kind of inflection point with embodied AI.
We can take machines that have previously been highly pre-programmed or that are remote controlled by humans, or even piloted by humans, and we can make them autonomous. What is that TAM? I think over the next 10 years, that's hundreds of billions or trillions of dollars of TAM, but we've got a lot to do to prove that case.
Maybe as part of that, how important is defense? I know you have industrial access along with defense. How do you think about defense's importance to the near-term growth strategy?
Defense is incredibly important to us, and defense is active with us on both the industrial robotics side, because you realize that the Department of War is very active in manufacturing and sustaining all manner of different types of machines. That's where our industrial strength AI comes in for industrial machines. Obviously it's much more visible what's happening on the weapons systems side. Defense is very important to us. We have contracts with the Department of War on both sides of the house, both the industrial machines and on things that fly, UAVs specifically. Far more volume anticipated in the early years on the drone side.
When we can make drones fly autonomously, be able to respond in real time to what's going on in the environment that they're operating in, on the ground, in the air, and then more importantly, be able to act as a collaborative swarm. There's a huge opportunity. It's no secret that the Department of War has talked about massive investments in UAVs and smart missiles. They talk about collaborative autonomy. That's exactly our swim lane. To answer your question in a nutshell, defense is incredibly important to our company going forward.
Just two follow-up questions to that. I think you've put out some white papers on the first. What are the core capability of Palladyne AI's software platform? Which I guess is also AI, but, and how does that software enable robots to perceive, learn, adapt, and collaborate?
Let's talk about it specifically in the case of drones, because that's what people are really focused on today. Historically, a UAV or a drone gets flown on the battlefield with a soldier holding a remote control device and managing that drone, where it's flying, what speed, what it's looking at with its camera, direction of flight, speed of flight, altitude, all of those things. It puts a lot of cognitive load on the human operator. There are different layers of automation or autonomy that are being introduced, and a lot of people in the sector use the same words to define a variety of different capabilities.
At its most simple category or most simple capability, you can see a drone that has way points automatically inserted into the control algorithms, and the drone will take off and fly to those way points on its own without requiring any soldier or operator intervention. That's great. That's one form of automation. It's not what we call autonomy. What we do with our software is we enable a drone to automatically take off, but with a very simple set of instructions from the operator. The drone can go to a particular geographic region. It can identify objects of interest, targets, ISR assets, whatever they might be, and be able to autonomously decide what that individual drone should do in response to what it's seeing going on on the ground. It doesn't need to follow a convoy.
Does the convoy fit the definition of what a target is as defined when the drone first took off? More importantly, the drone can operate in tandem with other drones that share the same software. It does not have to be from the same OEM. You could have drones from four or five different OEMs, which is what we just demonstrated in an active engagement with the Department of War on a battlefield, where you had multiple drones from multiple OEMs all collaborating with one another so that you had a better sight picture of what was going on on the ground, and able to share highly relevant and valuable information so that targets of interest could more easily be identified. That's something that is all done with the compute that's on the drones themselves.
It does not require any ongoing interaction with a central node or a server that's on the ground somewhere. In comms denied environments and/or GPS denied environments or both, our system can work quite well. That's what makes us very unique. You can analogize it to the self-driving car industry, which we're all familiar with. At one end of the scale, you've got cars that are being talked about being able to drive from Los Angeles to New York just with the entry of a single instruction. That's full Level 5 autonomy. That's comparable to what we do with drones. Up the stack, when you start talking about level 1 or level 2, you can be talking about something as simple as a car changing lanes on its own, and that's all it can do. Change a lane without collision.
That's also what people talk about in terms of autonomous operation of a drone, but it's far more limited in its capability. The key point to take away from us is we're at what we call Level 5 autonomy, fully capable, autonomous response to what's going on on the battlefield in real time, and to do it collaboratively with other drones.
I think that kind of touched on the next question, which was about SwarmOS. Maybe thinking about the system itself. We talked about drones, but how portable is the platform across drones or ground robots, industrial systems? Kind of with that, you talked about different OEMs systems working together. How much hardware integration is required for each customer deployment?
A few years ago, we split or tiered our AI development for fixed in place robots versus robots and machines that are mobile. When we talk about our IQ product, that's for robots that are fixed in place, bolted to the floor, manufacturing type robots. That's one set of capabilities now that we have commercially available. The other is for things that are mobile with an initial focus on UAVs, things that fly. The architecture is intended to be able to work with any kind of mobile or fixed machine for that matter. It's just there's another layer of complexity when you have to take into account a machine that is moving. We are capable of being extensible across all different domains, whether it floats, flies, drives, or even assets in space.
We announced recently that we have a contract with the military, to be able to integrate satellites as part of the knowledge and sensor base that we integrate into our swarming software. It is extensible from an architecture perspective. In terms of the complexity of integration, we do have to integrate the cameras or the other sensors that are involved. We have to be able to operate with the navigation system and the autopilot on drones, for example. It's not a heavy lift, though. As we have brought in new drone OEMs to work on our platform, we generally are able to get a new platform up and going within two weeks. It's a relatively short, light touch integration effort.
I think with kind of the new formed business, there's been a lot of activity, but maybe if there's some recent contract wins you can point to. How do you think about validation of that AI autonomy capabilities and maybe adding to that kind of in baseball terms, like where are you in terms of that spectrum?
Baseball. I would say that we are in the top of the third in terms of realizing the opportunity, not necessarily in terms of validation. I think in terms of validation of our SwarmOS, maybe we're at bottom of the seventh right now because of what's just transpired over the last 60 days. I said at the beginning of this year in my public comments that our biggest task for 2026 was to educate the Department of War and other customers on the fact that our capabilities actually exist. It's not science fiction. By some accounts, the Department of War had anticipated our capability that we're bringing to market not being available until the 2030s. Our job is to educate the Department of War and other customers that it's available today. We started that path in January.
I'm delighted to be able to tell you that we just completed a major joint exercise with the Department of War. I can't get into too many details about it, but it was a very large three-week exercise, and we integrated our software on four different drone OEMs. The feedback, again, I can't get into details, but the high-level feedback was we exceeded expectations and delivered an autonomy solution that they had never seen from anybody else, not even close. That is great news for us from a validation perspective. This was not a small test. This was an actual battlefield simulation, real world, thousands of troops on the ground. We've now been invited to, I think it's at least five more real-world joint exercises over the remainder of this year. That means we're checking all the boxes that we intend to do for 2026.
We're getting the exposure we need. Soldiers real-time, getting their hands on the technology, seeing what it can do to lighten the cognitive load and overall enhance mission effectiveness.
Just in terms of the business model, and I know it shifted a little bit, or it depends on the product, but what are current customers buying? Is it software licenses? Is it engineering services, integrated solutions? How do you think about that transition from some of these large-scale small development contracts to more scalable software revenue?
I'd say the answer is yes, all of the above. We sell a combination of software on a licensed basis, either a one-time, fully paid upfront, perpetual software license in the case of drones, which we price, by the way, at about 10% of the cost of the total drone. That's the value add that we bring to any drone that our software is on, to an annual software license fee per robot on the industrial robotics side. Those two businesses we've been completing and commercializing the software for the past two years, and it was just early this year that we launched those products into the market, so we're growing that business.
On our avionics components business, we have some great activity on the revenue generation side relating to both our BRAIN guidance and navigation board that gets sold to other OEMs, other UAV and missile OEMs. We've got a lot of activity on our precision components business. I alluded to that earlier in manufacturing. Yes, we also provide some engineering services, what we think of as best-in-class engineering services that assist other defense companies and aerospace companies with the design all the way through manufacture of new aviation platforms. We're hitting on all cylinders across all of those different categories.
I guess the answer to this question could vary depending on the product. You talked a little bit about demoing products, but what customer metrics matter most? Is it labor reduction, mission success, safety, uptime? Is it the autonomy performance? How do you think about what measures success in your customer's eyes?
I think from our customer's perspective on the mobile AI side, the answer to the question is, are we enhancing mission effectiveness? That all gets to the single point of reducing the cognitive load on soldiers so that they can do more, be more productive, get their head in the fight, focus on what really matters on the battlefield. I think we answered that over the last 30 days resoundingly, yes. I think we have exceeded both their expectations and ours when it comes to real-world battlefield performance. On the industrial AI side, manufacturing, that's about uptime and throughput relative to cost. Is the customer going to get a real-world return on investment based on investing in the software and whatever hardware they need to go along with it, to be able to increase productivity, quality, throughput, overall ROI, and reduce cost? That's the software side.
In terms of the other products that we sell, on precision components, can we make it better, faster, cheaper than our competitors? I think the answer is yes as we win more and more business. I think our state-of-the-art manufacturing facilities that are highly automated are proving out that we can reshore manufacturing, enhance the industrial base, do it the way the Pentagon wants to see it done, and increase overall availability of weapon systems in the U.S.. I think we're well-positioned on that front as well.
Then maybe just a little bit more on the competitive environment. Who do you think about as your most relevant competitors? Is it the OEMs that you're selling the software to? Is it AI software firms? Is it more startups? How do you think about the competitive environment?
Yes, it's everybody. It all depends on the sector that we're talking about and the particular product. On the AI side, there's no question there are some venture capital funded startups that are focused increasingly on embodied or physical AI. Far fewer than those that are focused on agentic AI. Agentic AI has gotten all of the real value proposition from investor perspective so far on the venture capital side. My view of that is we've got companies today that are being valued at hundreds of billions or trillions of dollars, and what those companies do is they help us humans with mental tasks. We're all familiar with ChatGPT and things like that. What we're focused on now, and some of our competitors that are venture funded companies, they are focused on how AI can help with physical, real-world tasks, not just mental tasks.
I think we're just at the very early innings of that whole industry evolving. Far fewer players in that space. For every 100 companies focused on agentic AI, there may be one or two that are focused on embodied or physical world AI, it's a much less competitive dynamic. When we get to the precision component side of our business, there are a lot of players in that space, a lot of them are more kind of old line machine shops. Not nearly as automated or advanced perhaps as we are because we came to the game a little more recently and had the ability to put our capital into advanced manufacturing opportunities. There are more and more folks in the venture capital space getting funded there also. I think the short answer is venture funded companies are a lot of our competition.
We don't see ourselves competing today with the big defense primes. They are our customers, not direct competitors. When it comes to kind of the mid-level primes like us, in some cases, they're customers and partners, and in other cases, they're competitors, but that's kind of the nature of the aerospace and defense industry, as you know. One day you're collaborating with somebody on one program, and the next day you're competing with them on another one. It's all good from our perspective. We want to be able to be a good partner and supplier to many, and in some narrow cases we'll be competitors.
Maybe just a couple of questions on the business model, but maybe just to begin, how do you think about the recurring revenue potential? Where are some of the biggest opportunities or partnerships for platform adoption? As you kind of scale, what would you kind of classify as recurring?
The software business is really where we are focused for a recurring high margin business. On the industrial manufacturing side, as I mentioned, that's a per robot per year license. As we grow the base of robots that move from being basic and dumb to being smart and intelligent, if a customer wants to continue to have that degree of autonomy and intelligence and agility in their robot, they'll keep paying us on an annual basis. We like that part of the business. That's at the very early stages of that. We haven't even gotten up to bat on that side. We're just approaching the plate now, again, continue the baseball analogy. On the drone side, I think of that as almost like the razor blade business.
It is a one-time upfront license fee. On the other hand, if you look at the volumes that the Department of War and our allies are going to be requesting and purchasing, most of these drones are a one-and-done mission use case. As more drones get sold and there's a greater desire and need for autonomous operations on those drones, we'll keep selling licenses. I don't think anybody believes that the demand for drone volume is going to go backwards. It's only going to go up for us.
I guess kind of related to that, just thinking about custom or customer engineering, and the scalability of the product, how customized is this? I know you've had different iterations, but is this really a scalable business on the software side because there's not much change with customer to customer once you kind of figure out what that competitive moat is.
Yeah, the basic architecture is all the same. As I mentioned, when we have a new drone OEM that comes to us and says, "We have a customer, a government customer that wants to see your autonomy solution on our drone," it's historically taken us about two weeks to get the software ported and implemented on a new UAV platform. I think this is imminently scalable. When I say ported in two weeks, that's a couple of guys working on it. This is not a big labor lift.
I guess related to that, what are the biggest obstacles to customer deployment or, just thinking about the roadmap over the next couple of years, what's kind of that unlock to full deployment?
I think it's as I said at the beginning of this year, it's getting customers to understand the capability exists today, that it's not a science fiction or a lab project at this point, that it is proven and tested, and that it works. When you think about defense procurement cycles, we've all heard about the fact that requirements have to be written. Once a requirement is written, it gets published. Once it's published, there's an RFI or an RFQ, ultimately a contract gets awarded. We're at the point now of trying to get requirements written that reflect the fact that our autonomy solutions exist. It can be a lengthy process, although, the Department of War is now moving with a speed that we've never seen before, and that's very encouraging for companies like ours. What's the long pole in the tent?
Creating awareness, proving out the system, getting requirements written, getting contracts awarded.
I feel like we've maybe focused a little bit more on the software side, but you did mention vertical integration early on with some of the recent acquisitions. Maybe just a little bit, if you think about more on the hardware side that there's probably some crossover on, where do you see the biggest near term opportunities just given the newer portfolio?
First off, our flight computer, which we call the BRAIN, it is a guidance, navigation, and control flight computer that has the ability to have our autonomy software fully embedded on it and shipped as a combined product. We call that IntelliSwarm. I think that there are tremendous opportunities with this proliferation of new defense tech companies that are bringing new UAVs and missiles and loitering munitions to market. We're seeing some great traction with that product. On the hardware side, I'd say that's the most immediate. We've already had some sales for it. The sales are growing. In terms of complete systems, we are in the process of bringing to market a cruise missile replacement that is one-tenth of the cost of traditional cruise missiles. We've had our first flight test, which was successful. That product is called the SwarmStrike.
That is intentional because we think that that, with our software on it, can become a full swarming solution that is unlike any other capability that is in the U.S. arsenal today. We are in the process of completing that product. We expect it to be commercially available to the Department of War in the next 12-18 months. We also have a quadcopter product, which is very unique. It is a mini bomber product. It is not intended to be an FPV drone. It's not intended to be just an ISR drone. It is intended to be able to drop a kinetic device or munition with a highly precise level of accuracy on targeting, then go back to base, get refueled, get recharged, add another munition, and have it go back and deliver on target again.
That is a repeatable event which will create a lower cost per effect or cost per kill than even the lowest cost FPVs. We think there is a very unique opportunity, and when we put those drones in a swarm, which by the way, this drone was demonstrated in this most recent exercise that I mentioned, it can achieve a lot of the military's objectives. Not everything has to be a suicide or kamikaze drone. We might be able to deliver a much better cost per effect with a munition being dropped from a drone, going back to base, and getting reloaded. Those are examples of two of the platforms that we are currently working on. We're also working on a near-hypersonic missile. We have a contract with the U.S. Navy for that product.
People may not be aware that the F-35 does not have the ability today or does not have a product today that is a missile that is near hypersonic or hypersonic that can be released from the bay of the F-35. If we want hypersonic capability, the missile today has to be mounted on the bottom or the belly of the F-35, which kind of defeats the whole stealth aspect of the F-35. We've been awarded this contract by the U.S. Navy to develop a near-hypersonic missile that can fit in the bay of an F-35. That's just an example of the kind of breadth that our team has in terms of capability.
I'm going to ask you about AI, which we've talked about a lot and is literally in the name of your company. Maybe just thinking about the production side, just given it is such a big part of what you offer on the product side, how does maybe AI change your business? Where is it integrated into your company outside of just the product that you offer?
We use AI in two different respects. The two different big categories of AI that everybody's familiar with, agentic AI, or what I refer to as digital world AI, where you're asking algorithms that largely reside in the cloud to help you with your daily workflow. Our engineers use AI from other vendors, just the way any large company or even small company does today. We're all relying increasingly on these digital cloud-based tools that we refer to colloquially now as AI. As we build out and enhance our manufacturing capabilities and we bring some of our UAV and missile systems online and we start producing them at scale, we fully expect to use our own homegrown IQ product that will enhance the manufacturing capabilities using robots that historically have only been automated and now will become fully autonomous with our IQ homegrown AI products.
Maybe just as we come up on time, one last question. What milestones should investors be watching? If you think out over the next five years, what does success look like for Palladyne AI?
I think over the course of the remainder of 2026, look for us to have integrations of our software with more partners on both sides of the house, both the industrial side and on the drone side. Look for us to have more wins at exercises where we're getting the Department of War to understand that the capability exists today. Look for more contract wins with money actually being awarded to us from the Department of War. On the component side, same thing. We'll be announcing customer wins, revenue producing contracts. I think it's just kind of a drum beat, a battle rhythm, if you will, of customer wins, proof points of the technology working, demonstrations where customers are seeing that this capability actually exists today. What does it look like five years from now?
I would expect five years from now for us to have a significant revenue stream from high-margin software businesses balanced by full system sales to the Department of War that integrate our software, and also a healthy and growing components business where we can maintain relatively high margins because of the uniqueness of the precision components and avionics that we're selling.
Okay. Ben, we'll leave it at that, and really appreciate the time, and look forward to following the company's success. Thank you.
Thank you, Greg. Appreciate your time.