Why This Company Could Become the "Microsoft of Drones"

Today we are talking to Alex Fink, the CEO of Swarmer. Swarmer is a drone software company that IPO'd back in March and saw their stock jump 500% on its first day of trading. There's a lot of hype around the company, so in today's conversation we learned more about what Swarmer does, their business model, how they went from being founded in 2023 to hitting the battlefield in Ukraine just one year later, and how they plan to expand the business moving forward. It was a very interesting conversation, so let's get into it. Well, on a very basic level, we help drones fly autonomously and cooperate in large groups without pilots. So the basic idea is Ukraine wanted to launch millions of drones but didn't have millions of pilots, and we wanted to allow it to do so with whatever pilots it had, so we created the system that allows one operator to control hundreds of drones. So I'm actually really interested in the origin story there because you founded the company in 2023, so not too long ago, and then you were in the battlefield by 2024, right? So I mean, what was that process like of just coming up with this idea, creating the software to do it, and then you're in the battlefield? Yeah, so I was building cameras and imaging devices before this for about 18 years. My co-founder was the head of computer vision and AI at Ring in Kyiv, right? And then they were acquired by Amazon, and when the full-scale invasion started, I guess he started thinking that maybe doorbells are not the most important thing to be working on right now, so he started volunteering for the different drone squadrons, he started helping startups in the ecosystem, and then one day he calls me up, and we've been friends for years before that, and he says, I know, here's what the problem is, and then he described what I just told you. I registered the company pretty much the next day in the US as a Delaware C Corp, right? And then we continued with this parallel track of business and legal and all of that organization in the US, engineering and field deployment in Ukraine. So the company was founded in May 2023. We got our first investor check in December 2023 from a fund created by Eric Schmidt, the former CEO of Google. Then January 2024, we had the first customer contract. April 2024, the first combat deployment. And then from there, I think you kind of know the highlights by now, more than 100,000 missions, the biggest series A in the history of Ukrainian defense tech. Eric Prince joins as our non-executive chairman and the first IPO out of this Ukrainian defense ecosystem. Yeah, I mean, the scale of how fast you guys kind of spun up, hit the battlefield, now, you know, got investors, now IPO. It's pretty incredible. So I'm kind of like, I want to learn more about like the technology here. So does the technology just take any drone off the shelf and it has the ability and it gives like pilots the ability to kind of control these drones? Or does it have to be like a specific drone that you have to, that like that your customers have? I'm just kind of like, can you take a DGI drone off the shelf and just start customizing it? So we will not work with Chinese components. Let's start with that. But even taking a step back, it's any drone that has sufficient processing power and the radio that we can use, right? So sometimes we start talking to manufacturers and the first thing we tell them is, looking at your bill of materials, we recommend that you upgrade the radio or that you upgrade the processor. Otherwise, there's not enough good things that can be done with this hardware, right? But assuming that the hardware is sufficiently strong, then yes, it's an integration process for us. It's not plug and play yet, right? But it's weeks to make it operational and to make it interoperable with everything else that we've integrated until then. Which basically means that the moment you're Swarmer enabled, now you can be included in the same mission as a whole bunch of other hardware from other manufacturers that is also Swarmer enabled. So you're still focused on the software side, right? So you're not in the hardware game. I think that's something that I want to clear up because a lot of people hear the word drone company and they're assuming you're making drones, but you're on the software side. So you can work with, I guess, pretty much any hardware outside of the Chinese ones. You don't want to work with them, but you can work with any drone hardware and then enable the pilots to control multiple and give them all these additional features and capabilities that can help them fight wars. Yeah, so we're a drone company like Microsoft is a PC company. Gotcha, okay. Microsoft's basic premise from 1980 was there's 200 different IBM compatible PCs will be the operating system that can work on any single one of them, assuming they have sufficiently strong hardware that follows the same architecture, right? So we kind of have the same premise. There's more than 500 drone companies in Ukraine alone, right? So we want to be able to run on every single one of those. We want to run on anything that is done in the West as well, in other countries. And when we say drone, it doesn't have to be the flying kind, right? You have marine drones. You have terrestrial drones. A launcher, even if it's stationary, if it can be controlled remotely, it's a drone from our perspective. A guidance kit on a bomb is a drone. A missile is a drone. So basically anything that has a processor and a radio and is unmanned can be controlled and included in the same set of missions where the mission could include multiple types of drones working together. Gotcha. So it's essentially, it's like any robot. It's not just drones. It's like anything that is autonomous, you can implement your software and give the ability to control multiple at the same time. Yeah. We've avoided the word robot because people have a certain image in mind when you mention it. But essentially, yes. Again, anything that is unmanned and has a radio and some processing capacity can be treated as a drone. It can be controlled remotely. Or in our case, it's not that we control it remotely all the time. We tell it what the rules of engagements are. And then our software runs on board the device so that it can make autonomous decisions. There's been a lot of worry about this, right? This kind of blew up a little bit more with the Anthropic Department of War drama that went down about a month ago where they don't want their technology to be used in weapons that autonomously make decisions. How have you guys thought about that? Is that something that you guys think about when it comes to your software? We think about it a lot. And we have to define it correctly. Otherwise, other people will define it incorrectly, right? So from our perspective, there's a difference between a decision on what is a valid target and a decision on what is the best way to achieve a target that was already set, right? So from our perspective, setting the objectives, choosing the targets, prioritizing the target, that's the task of a human. Our system is not involved in that in any way. But now that the human gave us a list of targets and prioritized them or set weights on them, it's the job of our system to figure out how to maximize the value of the mission with the assets we have. So if we have five targets, nine drones, very contested environment, we will figure out how to hit the most important targets, which drone takes which target, in what order they do it, what approaches they take. All of that is something a human does not concern himself with, but the human needed to choose these targets and approve them. Okay, so it's like a human made the decision on this is the ranking of how important the target is. And then from there, the drone will make the decision on how to best execute that mission, essentially. Yeah. So again, using the same operating system analogy that I've used before, if you have a dialog box on the RPC that asks you something and asks you yes or no, you click the button, right? You don't care which process then gets launched on which core of the CPU. What order are those commands scheduled in? All of that is too much information. You want to know that the button does what you want it to do, right? So we think that launching, whether it's aerial or marine or terrestrial assets should work in the same way. You decided what needs to be done and how that is prioritized. Then the software determines the most efficient way to use the assets to achieve that objective. You know, drones have become like a, have kind of become, you know, a very important tool now when it comes to modern warfare. I mean, we saw that with the Ukraine-Russia war. And then even with Iran and the US, you know, especially early on with the Iran conflict, Iran was launching their drones across the Gulf region. I'm trying to figure out how I should phrase this. I guess, has your company learned anything as the situation with, you know, Iran is developing and how the US is kind of, the US and their allies across the Gulf region are dealing with that? So we're learning all the time. That's the benefit of being founded in a war zone, right? Where the tactics essentially change every two or three months. So we're constantly learning and adapting. We're constantly gathering data from all the missions we fly and from all the missions everybody else flies as well. And we're updating what the system does and how it works all the time so that if somebody manufactured the drone and our software is on board, even if that drone is in a warehouse for the next two years, it will keep improving and changing its behavior, right? Every time it gets a firmware update, suddenly it will learn something new based on what happened in a battlefield that we were active in, right? So observing other conflicts where we are not yet involved, we're learning at the higher level, not from the actual data of what each system does, but more in terms of general descriptions of the tactics that we see on the battlefield. But in Ukraine itself, we're also seeing all of this. I mean, the threats that were hitting countries in the Middle East were already hitting Ukraine for two years before that, right? It's not like there was something new there. It was just something that the U.S. military experienced perhaps for the first time. Yeah, and I'm kind of curious to get your take. The U.S. military's drone technology, how does it stack up compared to... We've seen what Ukraine's military technology is when it comes to drones. China as well, they're like the leader when it comes to manufacturing drones. How do you think the U.S. military's technology stacks up against China, for example? Well, I think the U.S. certainly has the best systems, but what we're seeing in Ukraine is that the best systems are not what actually works in the modern battlefield, right? What works is a larger quantity of cheaper, perhaps less capable systems that are able to work in large numbers, right? So this is what we're seeing in Ukraine. And I think the U.S. is now starting to adapt to that. That's what drone dominance is about. That's what swarmforge is about and all the other programs that we're seeing. But it's going to take time. It's a very large ship to turn around, U.S. acquisitions, right? And so right now, the U.S. isn't quite ready for this type of conflict. It has some work to do to adjust. And the thing that we're seeing in Ukraine is that nobody knows the exact end state yet. So it's not like the U.S. can now make a 30-year program of how to adjust to this new type of warfare. It has to come up with a more nimble process that can pivot a little bit every few months as the tactics that we're observing on the battlefield change. I think the biggest thing that people have learned was that these drones cost relatively small amounts of money compared to the defense systems used to kind of block these drones. You know, a missile might cost a million bucks to hit a drone that costs like $40,000. And that gives, you know, a country like Iran the advantage when it comes to like attacking is that there's an asymmetric warfare. Like these drones are so cheap compared to like the defense systems. Yeah, let me give you specific examples there, right? So the Shah had cost somewhere between $20,000 and $40,000 depending on the size of the warhead. Patriot missile cost somewhere between $3 million and $4 million. And typically one is not enough. So we've seen instances where countries in the Gulf were shooting eight Patriots to take down one Shahad. Now, it seems like it doesn't make economic sense. You just spent $24 million to shoot down something that's worth $25,000. But the refinery behind it is worth billions, right? So it makes total economic sense for them to do this. But this is also why we saw several of these locations run out of interceptors just because they use them all. But cheap drones still keep coming, right? So obviously the world needs to adapt to this. That again, large quantities of very cheap, less capable, but capable enough devices are basically wreaking havoc on the entire world. And so we need to adapt to this both in the offensive capabilities, which is what you're seeing in the Ukraine, right? But also in the defensive counter U.S. capabilities and counter U.S.V. So in the water, you have the exact same thing happening, right? What is hitting ships at sea, no matter how big and powerful these ships are, is the equivalent of a jet ski with a remote control. Interesting. That's what made the entire Russian Black Sea fleet essentially hide in port. And now they can't even stay in port. They went upriver to hide inland. They are not operational at all. And that was supposed to be the second strongest Navy in the world. But it just doesn't exist because they were hit by six or seven foot boats without a captain on them, but just remote controlled or controlled by an autonomous system. Do you? So you mentioned how like, you know, drones are being used for offensive purposes, right? We've seen that in Ukraine. We've seen Iran use that. Do you think that's, I guess, do you see swarmers technology or just in general drones being used defensively? And how do you do that? Because like for offensively, it makes sense. You just launch a bunch of drones. Can you kind of swarm the enemy with a bunch of drones? What's the best defense against that? I'm just kind of thinking through it. I'm not a military expert here, but I'm just kind of curious to get your take. The only defense against a swarm is a swarm, right? Because if the enemy launches a swarm at you, then it's really inefficient to try to take them down one by one. First of all, because the incoming swarm probably knows how to reassign targets among itself. So even if you take out all the targets that were aiming at the particular site, some of the others will just reassign themselves and go take out that site anyway. Right? So now you have to take out all of them. Otherwise, something gets hit anyway. You can't just hit some. Then the other question is, OK, if you're trying to intercept it, your interceptor is probably not going to work 100% of the time. That's why you saw eight Patriots get launched to take out one Shahid, right? So the math is, let's simplify the numbers. Let's say that your interceptor works 50% of the time. You want the chance of impact on the target you're protecting to be less than 1%. How many interceptors do you need to launch? The answer is seven, right? Because two to the seventh is 28, right? You need to be above 100. So if you launch seven at every incoming threat, you'll probably run out of interceptors. But if your interceptors operate as a swarm and they can reassign targets among themselves too, now you have 30 incoming Shahids. You just launch 60 to take them out, maybe with a few extra to spare. And they reassign it among themselves so that on average, there's two per target and you take out all of them. So a swarm is, I don't want to overstate this, but I think it's almost mandatory to solve this kind of problem. And I'm not sure if you're allowed to share this, but we can cut this out if you can. But is swarmer, is your technology right now used for defensive purposes as well? Or is that something that's planned down the line? I would say that's a really obvious plan. So if it's not used yet, it probably will be. Gotcha. I want to now kind of pivot a little bit and talk more about the business itself. Like I said, you guys recently had your IPO, a lot of enthusiasm. So I want to kind of dig into the business model. Again, you guys are a relatively new company launched in 2023. Revenues in 2025 were $309,000. It was a 6% drop from 2024, but the company had an $8.5 million loss, which was more than four times larger than 2024. I want to learn more about the business model. Like what's driving the sales growth? How do you see a path to profitability down the line? So first of all, as I mentioned, the tactics in Ukraine keep changing quite a bit every few months. So when we launched, the primary product that Ukraine was using was larger bombers, especially the kind that was active at night. And we're using infrared cameras to basically seek out heat signatures and take them out. So we started from those products, but over time, they became less and less useful. And what you started seeing a lot more of on the battlefield is the use of smaller kamikaze drones, typically 8 to 10-inch quadcopters that just go and ram the target and explode upon impact. So we had to pivot ourselves as well because it's one thing to just integrate with a new product. It's another to go and embed ourselves with the units, see how it's used, figure out how to define the type of missions that commanders in the field want to run as well, etc. So this is why you're seeing this past transition from smaller revenues on our original type of deployment that declined over time. And what you're seeing right now with the numbers we have under contract and under MOUs, almost all of it is the new kind of product that we're seeing in Ukraine, smaller type kamikaze drones. And just for an idea of what the size of that market is, last year alone, Ukraine manufactured 4.7 million of these small drones. Okay. Right. And that's just Ukraine. We still don't know what 2026 is going to look like, but obviously we have a long way to grow in this market. Our market penetration is still not that great. Most of the stuff that is flown is still flown manually. We think that can be improved over time. Gotcha. So, I mean, you're still just in the early stages of just kind of even, you know, addressing the full market potential that you guys have just in Ukraine. I'm sure have you looked at, you know, trying to get US contracts, things like that, or is that still too far out right now? No, we have sales teams right now in every continent, right? It's just the sales cycles are somewhat longer in the West than they are in Ukraine. In Ukraine, if it flies, it goes, right? Elsewhere, you have to go through the process. And right now the processes are not very nimble in most of these countries. So this is why you have things in, not backlog, but under contract, right? But they take a while to realize. The other thing to consider is that we sell to drone manufacturers or to prime who then sell to government, right? So sometimes the moment in which we can realize revenue is not entirely controlled by us. It occurs when the hardware manufacturers ship to their end user, to the government. So even though we already sold software, it's already under contract. We know we're going to get the money. We don't know exactly when. We will when the contract gets fulfilled by our customer. So we are the subcontractor in these relationships. There's an upside to this, which is that we don't have to deal with most of the headache of government contracting, right? But the downside is I can often tell you how much is under contract. I cannot tell you exactly when it will be realized. And so I guess that makes sense now because I think I read somewhere that you described your business model as B to B to G, correct? So your, are your customers, the majority of your customers, these drone manufacturers who then you're selling software to, and then they go around, then they go and sell their actual drones to right now in Ukraine and potentially other countries on the line. Correct. And again, I go back to the Microsoft analogy because it answers 90% of the questions. It's like Windows selling to Dell, right? And then when you buy the Dell laptop, then I'm assuming they pay the license to account for that revenue, right? Yeah. So that's exactly what we do. We have a per unit fee that is paid upon delivery of the hardware to the end user. And then we have an annual maintenance fee from year two onwards, which in Ukraine, nothing makes it to year two. So that never gets triggered. But in the West, things are going to sit in a warehouse for a few years. And so every year we will get paid an additional fee to keep upgrading that drone and maintaining it and making sure that it's always running the latest and greatest. Are all your customers as of right now in Ukraine? All of the revenue you see from previous years is from Ukraine. Okay. The things you see under contract for the next two years is a mix. Is there concerns? I think one of the concerns that an investor could potentially have is that maybe you're too reliant on just a handful of customers. I guess you're trying to solve that problem by kind of branching out. So you have sales teams in the US and you're kind of working that those channels now to expand. I can tell you there's a lot more diversification and variety in the things under contract and under MOU. So yes, we were reliant on a very small number of customers early on. But when you're a small company, that's what you're supposed to do. Right. You're just supposed to go all in on the one good opportunity to be able to gather data, to be able to grow. That data is now a part of our moat, essentially. But now we have the resources to go much farther and much wider. And bear in mind, there's a conflict in Ukraine. There are conflicts in the Middle East. Our chairman is active in a few more conflicts, right? That you can kind of Google and figure out which countries those are. So I think we will see a lot more interesting, both sources of data, sources of revenue, types of hardware we run on, types of missions we run on, because every single one of these is going to have different mission templates, right? There's going to have different tactics deployed in the battlefield. And we're going to cover all of it if we can. If conflicts start winding down, does that impact the future growth of the company? I don't think so. I think that I wish I had the resources right now to use this technology for disaster response, for firefighting, for border patrol, for agriculture, for a whole host of other civilian use cases. But right now, the military demand is overwhelming everything. And so I think that for a while, that will continue to be the case. Even if all the conflicts just magically ended at some point this year, you would still have all the countries adjacent to the conflict. Armed themselves to the teeth over the next few years, because now they understand most of what they have in the warehouses is obsolete and they need to completely change their inventory and their tactics and what they have. Right. And then when that process runs its course, then I hope we'll have some resources to address the civilian use cases, of which there are many. Gotcha. So that's something that's new. I mean, I didn't realize that you were also thinking about using this technology for civilian use cases and not just the battlefield. We actually even had two contracts already in Ukraine for civilian purposes that just went on hold because nobody has bandwidth, neither us nor the customer to work on that. Like one of them is for demining. It's really difficult to focus on demining when you have active mining going on. Yeah. And so that had to wait. But yeah, we're absolutely, we started this from the beginning as a dual use company, focusing on all the possible use cases of autonomy in unmanned devices. And even if you think of the basic use case of dropping a package in a particular location, nobody said the package has to explode, right? The package could include blood, it could include water, could include medicine, could include many different things. And not everything is a package either. So surveillance, you can surveil military in the middle of a conflict or you can surveil a border or you can surveil a site that you are defending from potential attacks. Site defense, interceptors. I don't know if this is a military use case or a civilian one. It's kind of both. Gotcha. Because we're going to have a World Cup in a couple of months in this country. We don't have a war technically on US soil, but protecting that stadium, is that a civilian or a military use case? It seems like one in which you need to swarm right now. Gotcha. Last question for me. So, you know, you're starting off as a drone software business. Do you think you'd ever consider expanding into the hardware side and try to vertically integrate fully? I know there's a lot of interest there as well. Or do you want to stick to the software side and just kind of dominate that layer? Because I mean, Microsoft, you make the Microsoft example. They have their Surface line. They have Xbox. They have their own hardware lines now. Would you ever consider kind of dipping your toes into that? I can't predict the future or make future looking statements, right? But I could say that we have really good visibility into what works and what doesn't. Because we are hardware agnostic, we tend to work with many other companies. So we have really good understanding of who the best companies are, which tech teams are the best, which tech is the most promising. There is certainly the possibility that we'll start assembling a larger portfolio of items under one roof. Are they going to include this technology or that technology? I cannot comment in advance. We'll just look at every opportunity as it comes along. And I should have asked this early on. So you said that what the software lets you do is it lets one pilot control a swarm of drones, right? Is there any AI involved at all? I can't believe we went this long without talking about AI. Is there like an element where there's just, instead of one pilot controlling 50 drones, are you going to have an AI element where you just have a pilot press a button and it just like, the AI does it. Well, that's exactly how it works. So basically the upper layer interface that the operator uses allows you to choose a mission template, choose the assets, choose the targets, choose the roles of engagements, click go. Everything else is done by the system. So there is no actual pilot involved. Oh, gotcha. When we do demonstrations in the West, we have to have a pilot on site to comply with local regulations, but that pilot doesn't actually have a controller. Oh, so, okay, okay. I thought there was actually like one pilot controlling a swarm of 50 drones, but you're saying that the pilot just inputs the command like into like a command line given all the weights and whatnot we talked about earlier and then it presses go and then the drone swarm does its thing. Yeah, so the pilot is on board of the drone itself. Now, I should be cautious with the word AI because AI means different things to different people, right? People immediately imagine running on the drones. I don't believe drones need to think in natural language. So all these entropic or chat GPT discussions with regards to running on the drone, I think they're somewhat moot, right? To me, drones think in much more compressed, much more specific terms. There's no reason for them to think in English, right? And so the data we transfer between the drones is much more compressed. Every bit counts because the bandwidth of the radio is actually what determines the maximal size of a swarm I can have. If the drones need to transfer more information to each other, then I can have less drones in a swarm communicating with each other. And if I compress the data more than the size of a swarm can be larger. So from my perspective, AI in this case, the way we think about it as anything that can be rule-based is rule-based, right? Anything that can be solved with if-then-else or with linear algebra is solved with linear algebra or if-then-else. Only the things where machine learning is just the only working solution, we'll use machine learning. But even then, we will bound it in kind of a small sandbox that is only asked one question. What is the optimal path from A to B in 3D space? That's an AI question. Machine learning is good at that. But there's no free thinking or logic here, right? It's just answering a particular question. The same thing with object recognition, the same thing with object tracking. There are specific discrete questions that only machine learning can solve right now. And we will run that inference on the node. But it will not be thinking like a human without any constraints. There has to be a rule-based constraint. And on top of that, there has to be a human constraint. That's fascinating stuff. It seems like so much is changing right now, especially over the last couple of years. And I think I heard one of your TED talks say that we're going to have a future where it's just going to be robots fighting robots. And we're not even going to have humans involved. And it's just going to be metal on metal. And that's what warfare is going to be. So really fascinating stuff. You're on the front lines, literally and figuratively. So I appreciate you taking the time today. And we'll be following along and seeing you guys over the next few weeks and months. And congrats on all the success. Thank you so much. This was great. Appreciate it. Thank you. Well, all right, guys. Hope you enjoyed that conversation with Alex Fink. You know, it's pretty crazy to me that Swarmer was founded just three years ago. And now they're a publicly traded company. They're definitely capitalizing on all the hype around drone technology these days. I do think the market might be getting a little too excited, though. I mean, this company has a market cap of over $500 million at the time of this recording, while only making $300,000 in revenue last year. And they lost over $8 million as well. So it'll be interesting to see where things go, especially once the hype settles down. Let me know what you guys thought about today's conversation by dropping a comment on Spotify and YouTube. And while you're at it, consider giving us a five star rating wherever you listen to your podcast. All that engagement really does help us out and it helps other people find the show. Also, if this is your first time listening to this podcast, just a heads up, we post a daily market recap throughout the week, breaking down everything happening in the market. Each episode is only 10 minutes long, so it's a quick way to stay in the loop of everything that's happening in the markets right now, especially these days with earnings season underway. So definitely get subscribed to the podcast if you haven't already. Thank you guys again for listening, watching, subscribing, commenting. Shout out to Mike and Connor. For all the work behind the scenes. And we'll see you guys back here tomorrow.