Good. All right. Welcome, everybody. Thanks for coming this afternoon. I know this is the last session of the day, so we'll try and make it exciting for you. So my name is Ben Wolff. I'm the CEO of Palladyne AI, and we're here to talk about artificial intelligence for both the commercial and industrial marketplace and also defense. So we are what we call America's cross-domain force multiplier. What that means is our artificial intelligence and other systems that we sell are intended to make robots smarter so that they can be more efficient for their human users. We do that both in the defense space and in the commercial space. So our products that we have or are bringing to market include embedded or embodied artificial intelligence for both robots and drones. It includes proprietary UAVs and missile systems that we have under development.
It includes electronics and precision components for UAVs. And we have engineering services on both the commercial side and on the defense side for robots and for drones. And then finally, we have our own manufacturing facilities where we make precision-manufactured components for various defense systems. I'll get into all of those different aspects of our business in a few moments. So in terms of who we are today, what we look like today, obviously we're listed on NASDAQ. We're headquartered in Salt Lake City, but we have offices in other places around the country, including Kansas City, Tucson, Arizona. We have a small contingent in Huntsville, Alabama, where our propulsion systems are based, and also an engineering team in Boston, Massachusetts. We're about 140 employees as of the end of last year, and the company has been around in one form or another for more than 30 years.
We started life in the 1980s doing a bunch of R&D work with the Pentagon on robotics, human form factor robots. We worked on that for several decades, generated more than $350 million in revenue from DARPA and other DOD or Department of War now programs. In 2007, Raytheon acquired us, and we became the robotics division of Raytheon. We served in that capacity from 2007 until early 2015 when my wife and I partnered with the management team to do a management buyout, and we bought the business and the assets from Raytheon. After that, we brought in a number of key investors that had real interest in what we were doing on the robotics and artificial intelligence side. We brought in Caterpillar and GE and Schlumberger and Delta Air Lines and Microsoft as our core investors back in the kind of 2016, 2017 timeframe.
During that time period, we continued to advance our hardware systems and our artificial intelligence platforms. Fast forward to 2021, we went public through a D-SPAC merger. We raised a fair bit of money then, and in 2021, in the fall of 2021, I retired from being the CEO of the company. Thought I was going to go sit on the beach. Turned out it didn't work out real well. Two years later, the board of directors gave me a call and asked me to come back to the company because things had not gone according to plan. We're very candid and open about that. We did not succeed in our hardware development efforts, but what we did succeed in is developing some of the world's most advanced embodied artificial intelligence systems, and that's what we're going to talk about today.
I came back to the business in the fall of 2023, early 2024, and have been with the business ever since then. We restructured the business to focus exclusively on those artificial intelligence platforms. Today we have two products that we're bringing to market. I want to be really clear, these are not proven yet in the market. We've done a lot of trials with customers. We have some great customers at the U.S. Department of Defense that are supportive of what we're doing, but we've just launched these products and they're just getting into the market now. The first product is what we call SwarmOS. It is a drone swarming artificial intelligence software. It allows drones of all different types. We are hardware agnostic.
It allows drones to be able to swarm and collaborate in the air autonomously, which means they can perceive what's going on in the environment or on the ground and in real time adjust what they're doing in their mission to be able to respond to what's happening on the ground without requiring human intervention. That is a novel concept. It's something that people have talked about in science fiction for years, but we actually have it today, and we're demonstrating it every day with our military partners and customers. On the second part, we have Palladyne IQ, which again is intended to teach industrial robots, manufacturing robots, how to be smarter.
What I mean by that is with our software on an industrial robot, like you see pictured on the right, you can teach a robot how to perform a new task in a matter of minutes instead of spending months hand-writing code to enable a robot to be able to perform a new task. So it's about training a robot more quickly. And more importantly, it's about teaching that robot how to perform many different tasks at the same time. When you see a robot like this on a factory floor, oftentimes it's single-threaded. All it can do is the same task over and over 24/7. With our intelligence on board, that robot has the ability to now perform a wide variety of tasks and to respond to a lot of variation or variability in either the task or the environment. So that's our Palladyne IQ product.
So let me talk about what we do on the defense side. Both of those products I just described are relevant to defense. On the IQ side, on the Palladyne IQ side, we have our software working on robots at an air force base that are doing things like corrosion stripping and paint stripping off of aircraft parts, something that historically has been a very labor-intensive job for human beings. And now we're enabling robots to do that same difficult and dirty and challenging job. So that's what we do on the IQ side. On the weapons system side, our AI is being built onto drones of all different types. We have a partnership that we've announced with Red Cat and with Draganfly, and we're working on partnerships with other parties.
In addition, through some acquisitions that we did late last year, we have a company that is now part of our portfolio that is in the manufacturing business. It produces precision manufactured parts for things like the F-35 and for the Abrams tank. It also will produce our own drones, which we're developing pursuant to our GuideTech subsidiary that we recently acquired, where we are producing low-cost missile systems and uniquely differentiated quadcopter drones. So the companies I just mentioned, GuideTech, Warren & Key, MKR, those are the portfolio companies that we just acquired late last year. With those companies, together with our artificial intelligence, we have a complete vertically integrated defense company. We can take from design and concept all the way through scale production of UAV systems and missile systems. That's something that's unique for a company of our size.
So talking about swarming, what's so novel about this? You hear a lot of companies talking about AI on drones. Most of the time when people are talking about AI on drones, they're talking about enabling a single drone to be able to use some amount of intelligence or data to be able to complete its mission. It's kind of like a fancy way of saying autopilot. That's what most people are talking about when they're talking about AI on drones. In our case, we're talking about enabling multiple drones in a swarm to be able to communicate with each other without having to go back to the cloud. There's intelligence that's actually resident on each one of the drones.
Each one of the drones that's collecting sensor data, whether it's from a camera or an IR sensor or an acoustic sensor, whatever kind of sensor you might want to put on the drone, they are each collaborating with the information that they get among the swarm so that each of them have a better picture of what's going on in the environment below them. Initially, we're focused on targeting on the ground, but you've all heard of counter-UAS, counter systems that can take drones out of the sky. There's no reason that our capability couldn't be used for targeting things that are in the air in addition to things that are on the ground. That's an area of expansion and opportunity for us. Importantly, as seen in this slide, we've got one soldier managing a fleet of drones, a swarm of drones.
That's an important capability and something that allows the U.S. military to scale their drone deployment. You've all seen in the news that the Department of War is very focused on having swarms of drones. The problem is today, virtually every drone that's out there requires a soldier to manage that drone. It's a one-to-one ratio of one soldier to one drone. With our system, on any variety of different drones that you want to think about, we can have a single soldier managing a whole swarm of drones. So that allows scaling of deployment of drones. In terms of the drones or UAVs or missile systems that we're bringing to market ourselves, we have an incredibly innovative team that does a lot of development and design work for other prime contractors. They're also doing it for us in-house.
This is an example that you see here of something that is one of the weapon systems that we're developing in-house. It's called SwarmStrike. And we went from original design concept, meaning nothing had been done on the project at all, whiteboard drawing, to having our first test flight in less than six months. And that test flight, you can see in this video. So this is a system that has incredible capability relative to anything else in the market at a price point that is one-tenth of the typical price point that all of our competitors and peers have. So our whole focus is on delivering highly cost-effective, attritable munitions. They may be UAVs, they may be missile systems, but this is one indication of how quickly our innovative team can move to address a need in the market.
That's great because right now you hear Trump and the Department of War talking a lot about rapid innovation. The purpose of this video is to show you we know what that means. We could do it in less than six months, start to first flight. Maybe. We got a technical problem up here, guys. Is it advancing? It's not advancing. Stay tuned for station identification.
Is this slide showing your website?
It is. All right, that's the one I want. Can you blow it up to be presentation mode?
Yes.
There we go. SwarmStrike was one of the two systems that we're bringing to market. The other one is called Banshee. Banshee is a mini bomber concept.
One of the things that you've seen the Department of War and the White House talk about recently is the desire to have a very low cost per effect. That's a fancy way of saying a low cost per kill or low cost per mission. In our context, we have a large quadcopter platform that basically has a mini projectile or a mini munition on the underside of it, and it drops it like a bomb. Think of the way a B-52 works, same concept, but for a drone. We are not aware of anybody else that has fielded a product like this. Talk about a low cost per effect instead of having a $2,000 or $3,000 or $4,000 low cost FPV drone that rams itself like kamikaze style into a target. In this case, the bomber drops the munition, as I hope you'll see here. Nope.
There we go. One second. So it drops the munition, and for less than $1,000 per target, we can actually achieve the objective of there we go. That gives you a concept of, and again, this is first flight of Banshee, did that in a matter of months from first design to execution of the first concept. All right, moving on to our manufacturing capabilities. I mentioned that we have the F-35, we make parts for the F-16, for the Tomahawk, for the Bradley, a number of defense programs that we actually manufacture high-quality precision parts for. It's not by happenstance that we sell these products to Lockheed, TDI and Kratos, Boeing, Stark, and others. These are big prime defense contractors that both our facility and our products are qualified to sell to.
This team, together with these facilities, gives us the ability to manufacture our own UAVs in scale while at the same time delivering high-margin revenue-generating products to the rest of the defense industry. We have more than 120,000 sq ft of manufacturing capacity. One of the challenges that startups that are just getting into the drone industry often have is, great, you've got an interesting product, Department of War has expressed interest, but can you scale, and I'm proud to say that we've got a team that has decades of experience in scaling manufacturing, and now we have the facilities and the machinery that's necessary to be able to do that, so bringing it all together, we've got some of the best world-class design engineers in the world when it comes to new UAV systems.
We've got our embodied AI that will highly differentiate these systems from what anybody else has in the world. And we have the manufacturing capability and facilities to be able to scale it. Now let's talk about the commercial side of the business. I mentioned making manufacturing robots smart. That's our Palladyne IQ product. This system has now been working on 160 different models of robots. So we are qualified in working on KUKA, ABB, FANUC, and Universal Robots. That represents about 70% of the market in the U.S. for industrial robots and cobots. Our product lives on the edge, meaning it's installed on the robot itself. It does not live in the cloud. This is not a system like a SaaS system. It's not like what Palantir does. It's not like what Google or ChatGPT does.
We call it embodied AI because it actually lives on the machine itself, and it allows the machine to respond to what's going on in the environment in real time. This product has been under development since 2018. We have had a fair bit of government funding to develop it because the military has been interested in how they can reduce the labor cost and increase the efficiency and throughput consistent with what President Trump was just talking about last week, modernization of our defense industrial base. So what this whole system does, this AI system, I know there's a lot of words on this page. Again, this slide deck is on our website, but this is intended to convey that what our system does is it functions much the way we humans do. It allows the robot to perceive what's going on in the environment.
Now that's a technology called computer vision, and you see a lot of companies out there that are selling computer vision, but that's just one small part of our system. Computer vision allows us to see and perceive the environment. But then what our algorithms do is it takes the input from computer vision and allows us to think and reason to solve a problem. A problem might be, I've got a new part in my hand as a robot, and I don't know what to do with it. I've never experienced this part before. That's something that would bring traditional automations to its knees. As soon as you introduce variation or variability to existing automation, it doesn't know what to do with it. It says stop, it shuts down, and it requires more code to be written so that the robot can actually perform its task.
In our case, because of our perceive, think, reason, learn, and act process, which is a closed loop, complete feedback loop, there is no need to intervene and have a human tell the robot what to do. It can figure it out for itself. All right, so let's move on to what our financials look like. 2026, we announced guidance earlier this week that said we expected to have $24-$27 million revenues for 2026. The businesses that we acquired this last fall, they're effectively cash flow neutral other than a little bit of R&D that we're going to invest in those two flying systems that I mentioned. We had $47 million of cash on hand as of the end of last year. So we've got a nice long financial runway in front of us.
We are spending roughly on the order of $2 million a month, $2.1 million a month on our R&D, and we expect to start recouping some of that investment as we get these new AI products into the market and start generating revenue this year. So we know that we are at the early stages of this concept of embodied AI. It is not the kind of AI that customers or investors are familiar with. Although in reality, if you think about what Tesla does, we're all familiar with embodied AI. Tesla has put AI into its cars. Tesla is now putting AI into its humanoid robots with one big difference from what we do. They also have a lot of AI living in the cloud with constant connectivity between their cars and their robots.
We don't have that cloud connectivity because we don't need it with the way we've designed AI. But you're familiar with embodied AI because of what those two products are doing at Tesla. We've got a different approach to it, but because it's new and it requires a lot of education, which means a longer sales cycle, we think of our business over the next three years with a crawl, walk, and run mentality. So this year, $24 million-$27 million in revenues based on kind of modest market penetration of our AI products. 2027, 2028 is when we really start taking wings with our AI products. In terms of the key catalysts, I'm not going to read through all of them here, but this is what, if I was an investor and I am an investor, I'd be looking at for us to deliver on 2026, 2027, and 2028.
So please reference this if you get a chance on our website. So back to kind of the core of where I started. We are two distinct businesses, commercial and defense, but all leveraging our basic artificial intelligence platform. So why now? I'll tell you why I'm really excited about the business today. A lot of it has to do with what's going on in defense. If you look at what the White House has talked about over the course of this year, or I should say last year, and the actions that the White House took just last week, its executive order saying that the CEOs of the prime defense contractors can't make more than $5 million, and they can't issue dividends or distributions until they have submitted plans to Pete Hegseth that show how they're going to modernize their manufacturing. I believe that's just the start.
Trump is extremely focused on modernization of our defense industrial base, and our IQ product is perfectly poised to be able to take advantage of those tailwinds. Additionally, on the defense side, this whole notion of having low-cost, attritable weapons that recognize the fact that we are likely to be engaged in conflicts that are more about economic attrition than we are about large weapon systems that can bomb the heck out of everybody. I mean, this is a war of economic attrition. We're seeing that in Ukraine. You saw that a bit in Israel, and I think that's what we're likely to experience going forward. That's why Trump and Hegseth are so focused on it. Our artificial intelligence and our upcoming drone platforms and our components business are perfectly poised to take advantage of that. So that's all I've got for you. Any questions? All right. Yes, sir.
Does the software have a special, some kind of a training data set or something, or is it mostly people have come up with the?
Yeah, we don't need a database of training data the way you see with LLMs. We have what we call domain-specific language that is narrowly tailored to collate or curate the information that's coming in off of sensors and to respond to it in real time. So we don't do training or learning the traditional way that you think about in the way a ChatGPT does, for example. It's all done in real time without reference to a database.
What would you say is the long-term competitive advantage of this or duration of competitive advantage of this?
So first of all, I think about AI breaking down into two categories.
You've got AI that is embodied like we're doing, where it's actually resident on the machine. Then you've got cloud-based AI, which is what you see Palantir and Google and all these other companies doing. I think there are strengths and weaknesses of both. The strength of the cloud-based AI is you have much larger data sets that could deal with a much wider range of situations. I mean, when any of us types in a query or asking for information out of ChatGPT, they're literally boiling the ocean to come up with an answer because it's up to speed on every topic on the planet, and as a result, you need to have massive compute power, which costs massive energy costs to be able to support that.
In our case, our domain-specific language model, it sits only on the robot, and it's narrowly tailored to only deal with the kind of applications that the drone or the robot on the manufacturing floor might deal with. So we couldn't take a drone and have it operate in some set of circumstances it's not familiar with. We couldn't take a manufacturing robot and expect it to go drive your car. It is domain-specific. But the benefit that we bring, the advantage that we bring, is it doesn't have to be connected to the cloud, which means there's no latency. There's no lag time between getting the answer in the cloud and getting it to the robot, which is incredibly important if you're trying to respond to manipulating something in real time for a manufacturing robot, for example. The other thing is we avoid the cost of cloud connectivity.
For a while, we had one of our robots connected to one of the large cloud services, and we were spending $35,000 a month just on cloud connectivity. So this is a much more economically viable solution by having the AI resident on the machine itself. Now, we can still connect to the cloud if a customer wants to download data or extract data from the robot, but that's a one-way transmission of data. It's not required to give instruction to the robot. And the guidance, can I follow up on this?
Yeah. Guidance for next year, how much is hardware software and how much is embodied AI?
Yeah, we haven't broken that down. I'll tell you that the majority of it will be defense for 2026.
Hardware?
That we haven't gone to. Any other questions?