Drones in Ukraine: Emerging Lessons
March 5, 2025
Watch the Video
Read the Transcript
This transcript was generated with the assistance of AI. Please report inconsistencies to comms@afa.org.
Maj. Gen. Charles Corcoran, USAF (Ret.):
All right, good morning everyone. Welcome to our panel today, titled Drones in Ukraine and Emerging Lessons. I’m Chuck Corcoran, I’ll be your moderator for today’s discussion. It’s really great to have our AFA President here in the front row. Thank you, General Field, for being here. Mr. Canada, board member, we got Assistant Secretary General for Operations from NATO. So apparently this must be an important topic to some of you out there. So thanks for being here. As we watch the war in Ukraine unfold over the past three years, drones have certainly emerged as a dominant technology that’s being employed by both sides. The technological innovation and the counter-counter moves are incredibly fast-paced with new approaches, tactics, and technological and software updates being fielded in days or weeks rather than decades, which is what we tend to have become used to over the past several decades in the US military. These unmanned systems are being utilized across every single domain of the battle space to varying degrees of success, and we’ll get into that a little bit during our discussion. The significant impact of drones on combat ops is certainly a revolution in the character of war, and given that, we wanna delve into what the lessons are that we should be learning with from Ukraine to make sure that we’re properly equipping America’s war fighters as they go to battle for us. American industry is key to helping us answer these questions and delivering those capabilities to our war fighters, and so we’re fortunate to have representatives here today from a few of the companies on the leading edge of this work, starting with the CEO of Shield AI, Ryan Tseng. We have Steve Milano here, the head of Advanced Effects Business Line within Air Dominance and Strike at Anduril. And we have Mr. Kyle Woo, the Counter-UAS Program Manager at Zone 5 Technologies. Thank you, gentlemen, for being here, and we look forward to the discussion with you. So let’s kick this off. You wanna hear from these guys, not me. We’ll start off with you, Ryan, and I wanna hear your views on what I’ll call a close fight that’s going on over in Russia and Ukraine, the battle space right now. So the use of drones along the front lines is what I’m after here. A story in yesterday’s New York Times reported that 70 to 80% of all casualties, both Russian and Ukrainian, today, are the result of drone strikes. Most of these are along the front lines. Russia conducted more than 8,000 explosive drone strikes against Ukraine from September to December of 2024. That was four times more than they’d done during the same period in 2023. Ukraine’s operating at a similar tempo. So could you please talk to us about what Shield AI is working on in this sector of the battle space, what lessons your team has learned, and how you’re applying that to developing and fielding capabilities for our warfighters going forward?
Ryan Tseng:
Yeah, absolutely, and thank you very much for having me, and good afternoon to everybody in the audience. If you don’t know Shield AI, it’s about a 1,000-person organization. We’re about a decade old. Most recently valued at around $5 billion, and through our existence, we’ve been focused on bringing AI to the warfighter to enable advanced autonomy so that our vehicles can be capable and all the missions that we need them to be capable in, regardless of the conditions that they are operating. Ukraine is an area where we have a pretty substantial commitment. We have an office out there, and last year, we made the decision. We had heard a lot about the challenges of the EW environment, and we decided that our team was gonna go ahead and try to figure out how to make drones effective in those environments, and I think, in a pretty eye-opening moment for the Ukrainians and many NATO forces, Shield AI was very successful in flying one of our aircrafts called the VBAT. It’s a Group 3 UAS. It can fly for about 12 hours, but crucially, it was able to take off without any GPS, go into Russian-controlled Ukrainian territory, fly a six-hour sortie, and without GPS at any point in the mission, call in targetable coordinates that resulted in a HIMARS strike on a Russian SA-11, and we’ve continued to operate since then, whether over land or over sea, or the Black Sea environment, to continue to operate these drones successfully in very intense EW environments, and continue to provide targetable coordinates and actionable intelligence, and so the major learnings for us, I mean, I think a lot of them have already been reflected externally in the media. People talk about the need to be expeditionary, to be constantly on the move because of the danger of drones, and the need to continue to stay out of the enemy’s weapon engagement zone, or sorry, to just, you’re in the weapon engagement zone, but to be continually moving. There’s a lot of talk about the need to make sure that your equipment is going to be successful in EW environments, and I’ll just say from an industry perspective, I think one of the very interesting anecdotes for us was the criticality of being able to update software in near real time based on what’s happening in the battlefield. So leading up to our initial deployment in Ukraine, we were assured that on the ground we would have GPS, and that the jammers would kick in around 200 feet, and so our engineering teams, when we started designing the capability to operate without GPS and very limited communications, we accounted for that in the design. When we went out to Ukraine to conduct the first operation, it turned out we lost GPS at about three feet because the Ukrainians were jamming GPS to protect themselves, and the aircraft went up, turned the wrong direction, and it flew for about an hour going the wrong way. I don’t know if any of you have seen the movie Interstellar, but it was basically a scene out of the movie where we’re chasing this thing through, like over highways and through cornfields, and eventually we brought it down and recovered, and we got a call back in our office in Dallas and said, “Hey, we don’t have GPS up to 200 feet. “We don’t have GPS, period.” And in the span of 24 hours, our team provided an updated software package that didn’t depend on GPS at any point in the mission, pushed that forward, and it was the next operation, another 24 hours after that, that resulted in the targetable coordinates that resulted on the strike on the SA-11. And I think the lesson here, because we participated in a lot of programs, that the idea of pushing software from nothing to combat operations in 24 hours I think is something that really breaks the way that a lot of programs operate in the United States and around the world, and I think we need to think about how we can achieve that level of software agility to be competitive in environments that might be similar to Ukraine around the world.
Maj. Gen. Charles Corcoran, USAF (Ret.):
Well, thanks. That’s some really impressive work, Ryan. Steve, we’ll shift over to you next and the work Anduril’s doing, and I’d like to specifically hear a little bit about the work you’re doing in the maritime domain. I think one of the most significant unmanned systems achievements so far in the war in Ukraine has been Ukraine’s ability to secure localized maritime superiority in the Black Sea off the coast of Odessa, and they’ve done it without any traditional naval vessels. So can you tell us about the work that Andrel’s doing to deliver capabilities in the maritime domain, any lessons that you and the company have gleaned from what we see in Ukraine and some of the capabilities you’ve delivered to allies to date?
Steve Milano:
Yeah, absolutely. So appreciate the opportunity here, and I think a lot of what Ryan said rings true around how we’re identifying and understanding the requirements and the speed of the capability gaps that we’re seeing. And so Andrel’s been on the ground in Ukraine since two weeks after the invasion, and so seeing some capability get seeded into the fight in the form of loitering munitions in our Altaeus line, as well as some of our loitering UAS-type capability and our GOES capability, and as expansive as kind of those engagements have been, I think what Ryan pointed out is 100% accurate, is that it doesn’t matter what hardware that you’re putting downrange, it doesn’t matter what you’re showing up with, it doesn’t matter that you have this huge surface naval vessel fleet, what matters is your adaptability on the battlefield, and we saw that both in the EW spectrum, as well as in our targeting capability. And so as we start to take those lessons learned and fold them back into our systems of systems, it really provides the impetus to drive into the area of just accelerating technical insertion of capabilities into your fielded assets, and getting out of the requirements loop. Because if we were in a traditional requirements loop of we have the surface naval engagements that we actually need to go after, and we needed to acquire some capability to counter those efforts, it would be a 12-month, 18-month, 24-month activity to demonstration and prototype, and while people are dying on the battlefield, you don’t have that luxury of gaining those timelines. And so getting in those 24-hour software integration loops is critical to being able to provide the capability to the warfighters in a timely manner. And so what that actually looks like is leveraging a lot of the investments that are already being made on the software side, and I know it’s become a cliche that software’s eating the world, but it’s true. As much as I’ve been an advanced effects missiles person my entire career, I live and die based off of the software that I’m able to integrate into the system. And I always find it amusing that if I go back in the history of Andro industries, it’s just like, oh, okay, there were Andro industries for a reason, AI, and so at their heart really, really developed kind of a software-based company that’s hardware-enabled. And so I think that that’s a little bit of what you’re seeing here with the three industries, three companies that you have up here on the panel, is that we’re leaning into that adaptability both from a software architecture, but from a scalable manufacturing architecture as well. And so the direct question was about the maritime domain, but I think that that scales across the board. I think that everywhere that we’re actually engaging in the capabilities that we’re fielding, only way that we’re able to field and engage them is getting real-time feedback and having the mechanisms in place. ‘Cause I think the lessons learned there are less about our capabilities because nothing we’re doing I think is pushing the edge of technological advancement. It’s furthering things that were already done. These are engineering problems, not physics problems. And we’re finding new ways to apply those lessons learned. And I think that the real modality that’s going to help us go quicker and what I think we all know to be true is that the latency from demand to execution is too long right now. And we’re seeing that if we can break down those processes and actually deliver capability that we already have resident, it’s all about scaling and your responsiveness to those requirements.
Maj. Gen. Charles Corcoran, USAF (Ret.):
Thanks, Steve. Kyle, let’s shift over to, so we’ve talked close in, we’ve talked some maritime, some software, you know, on the offensive side. Let’s switch over to the defensive side here. Talk about counter UAS. Ukraine and Russia are both, while we hear a lot about the attacks, and I even mentioned 70 to 80% of the deaths are coming from small UASs in a close in fight. Truth be told, a lot of the UASs don’t get through. The defenses are pretty solid. Long range strikes by Russia, for example, only 10% of their drones are getting through Ukraine’s defenses. Can you talk to us about the work that you’re doing at Zone 5 on the Paladin system and others? And then other lessons that you’re gleaning from Ukraine and incorporating into your work at Zone 5.
Kyle Woo:
Yeah, appreciate the opportunity, and excited to be here. As you mentioned, I’m Kyle Woo. I’m the program manager for our counter UAS portfolio at Zone 5. We’re largely split up into two main product lines, the first of which being Air Launch Effects. We were recently down-selected by the Munitions Directorate for the ETV effort in there, and then the other side of the house is our counter UAS portfolio. Being an unmanned systems company, a large part of our counter UAS products are interceptor platforms, Paladin being one of them that he mentioned. When I look at the actual defense side of it, counter UAS is a very, very broad subject, from group one all the way up to group five, and the mitigation that you apply to a group one drone, even the detection that you use to detect a group one drone, it’s very different from a group three. But when I go and I look at the defeat, I kind of go back to first principles of RF jamming can be effective, but they can run a fiber optic line, and they will rapidly adapt. But kinetic is always gonna win. There’s always gonna be a thing in the sky, and as long as you have a big enough rock to hit it with, it’s gonna fall out. So what we’ve kind of done with Paladin, for example, group one, group two threats, these FPV kamikazes that might be $1,000 or less, lowering that cost curve for engagement, we take it out of the sky with a drone interceptor and a 12-gauge shotgun. So now your expense per round is a 12-gauge, and it lowers that cost curve. But that’s not gonna work against group three threats. It’s all a layered approach. And leveraging open systems, MOSA/WOSA compliance, and multi-mission hardware architectures is really how we’re gonna win the battle long-term. Because while a 12-gauge shotgun might be effective right now, it might not be effective in the future. And leveraging the state of our abilities to adapt on a weekly timescale with open architectures is really how we’re gonna win that fight.
Maj. Gen. Charles Corcoran, USAF (Ret.):
Great, thank you. All right, Ryan, we’ll go back to you. We are here at the Air & Space Forces Conference. There’s a lot of discussion among the senior leaders about how best to utilize the resources that they have at their disposal to field the most capable, most lethal force to do the mission that the nation demands of the Air and Space Forces. Given what we’re seeing in Ukraine, given your experience with your company, if you’re king for a day, you’re Chief of Staff of the Air Force, you’re secretary, or you’re advising them, where do you focus the resources that the department has to get the most bang for the buck?
Ryan Tseng:
Yeah, great question. First, I think a lot of the focus on mass makes a tremendous amount of sense. And we talk about what’s happening in Ukraine. You mentioned a small number of things getting through, and so just having enough rounds or vehicles and the regeneration rate, I think, makes a tremendous amount of sense. The thing that, from our perspective, could use more resourcing is on the autonomy side. I think that’s really one of those opportunities to achieve tremendous returns on investment. For a while, I think it was 90% of drones in Ukraine were being defeated by electronic attack. And that’s an example where, if you put the right software on these vehicles, your attrition rate could be cut by a factor of 10. And so you’re sort of faced with the choice, do I wanna build a factory to build 10 times more stuff, or can I apply a software update that reduces the attrition rate by a factor of 10? And so that’s just a specific example at a moment in time where software could have really made an outsized difference and it’s not always gonna be the answer to invest in software ahead of hardware. But I think thematically, if we just go back over the last 30 years, I think there’s been tremendous underinvestment in software capabilities. The great technology companies that are around today, the ones that have disrupted industries, those companies that disrupted the incumbents were the ones that over-invested in software compared to the hardware, because as Steve said, software’s eating the world, cliche at this point. And so I think a question for the United States is, if we do what technology companies have been doing for the last 30 years and over-invest in software, what sort of outcomes can we achieve? And I think that we would find that they would be quite compelling.
Maj. Gen. Charles Corcoran, USAF (Ret.):
Yeah, that’s a great one. I’d like to riff on that a little bit more. You walk around the exhibit hall here, you see hardware. You see small UASs, larger UASs, even CCA models. It’s hard to put software in the booth.
Ryan Tseng:
It doesn’t show as well.
Maj. Gen. Charles Corcoran, USAF (Ret.):
Yeah, how do we change this mindset in the department? Is this uniformed personnel work or is this something that we rely on contractors, companies like Shield? And I invite the rest of the panel to weigh in as well.
Steve Milano:
I think it’s both. I think that you need kind of the combined effort of the industrial base and government to understand the path forward. There’s a lot that we can do on private funding, whether that’s grad work or it’s direct high-rat investment or direct investment into activities. There’s a lot of things that we can do ahead of need, which will drive forward. And I think that’s a little bit different than the traditional development cycles that the Department of Defense has been used to. Usually the developments come out of the labs. And so I think that you’re seeing this traditional, that traditional mindset get broken a little bit, but there’s goodness to that in that we can invest in things that are pushing the state of the art forward in autonomy and in software development. And then you get to pick and choose what’s actually necessary. And so what becomes the incumbent responsibility of the acquisition process is to be very clear about what are the priorities and the things that you care about so that we can make critical decisions about how we’re investing in those things. And so I think it’s more than just demonstrating that we can go do those things. It’s about demonstrating the things that actually make a difference. And by definition, Department of Defense should be telling us what are those things that will actually make a difference.
Maj. Gen. Charles Corcoran, USAF (Ret.):
Thanks. Kyle?
Kyle Woo:
Yeah, and to add to that, I mean, if we fast forward to the future of what we think the threat is gonna be based off of the lessons learned here in Ukraine, we immediately envision AI-enabled swarming tactics being used against us. And an operator, it’s impossible for someone to process that. There’s no way that someone can input a suite of sensors and detection and mitigation methods against hundreds of adversaries and be able to perform a defeat. They need AI and software and modeling and simulation to be able to inform ourselves that that is the route that we’ll have to go. And this is what the technology that we’re gonna need to be able to mitigate that threat in the future. And both Shield and Anduril are leaders in that space.
Ryan Tseng:
Yeah, it’s hard for me to comment on what the department should be doing differently, but I can describe what an ideal outcome I think would be for the industry government team as it pertains to software. I think number one, there’s a question of how fast is your capability generation? I gave a very specific example about our work in Ukraine. Hey, turns out there’s actually no GPS at any point. How quickly can you turn a software capability update and get that created? So now that you have this capability. And so there’s all the way, there’s what I think a lot of people think about as the time until the first flight or the time until the developmental test or the operational test where you get that first outcome and you’re like, hey, I think I’ve solved the problem. Let’s go see if we actually solve the problem. And then the second, so I think we need to think about what is our capability generation rate on the software side? In the same way that we think about our hardware generation and regeneration rate, what is our software capability generation and regeneration rate? The second component of that is what is our fielding rate? So there might be new training elements associated with it. There’s all of the infrastructure necessary to deploy that hardware, or sorry, that software to those hardware targets swiftly and securely. Do we actually have the infrastructure in place that if it works, that we can push it out there at the speed of relevance? And I think that that speed of relevance is getting faster and faster and faster because if somebody is coming at you with complex software-defined systems, try as you might to anticipate what that threat looks like, it will be inevitable that you will be surprised and something that you were depending on is not going to work. And so you need to have a workforce, whether government, industry combined, that can very rapidly generate responsive capability and that can very quickly redeploy that capability and integrate it with a force and be effective. And my impression is that the government industry team can make a lot of progress if we really lean forward in these two areas.
Maj. Gen. Charles Corcoran, USAF (Ret.):
We could spend a couple days tapping into the readiness side of this. Because at the end of the day, we gotta have forces that are ready, that look very, very strong and capable to our adversaries to deter, and if the adversaries are not deterred, those forces have to be able to go defeat. And traditionally, we define a requirement for a piece of hardware and we develop it and we test it and we field it, as you said, and it’s done. Software is never done. And so do you have some thoughts, any of you, on what day-to-day training and continuous software delivery and development looks like in peacetime? How do we build out the training infrastructure in the DevSecOps kind of mindset where the developers that are likely not government, they’re the folks like you on the stage, sitting there side by side with operators and rapidly iterating and putting systems in challenging environments that force us to require a software change. GPS is a good example. It’s difficult to get authorization from the FAA to train with GPS jamming, even limited. So any thoughts on that? How do we make sure we have ready forces?
Steve Milano:
I think just real quick, my perspective here is that eventually you’d like to see as many humans taken out of that process as possible. And so I think that as we develop the future TTPs and future training protocols that are gonna be necessary to keep these algorithms alive and keep everything kind of moving ahead of the competition, is that we’re going to be taking not just humans out of the logistics chain and out of the operational chain, but out of the training process as well to actually be prepared for that. So I think that there’s going to be less humans required to maintain the progress that once we actually get these things instantiated, and we can debate a little bit about whether or not that ecosystem is live and healthy today, but once it is, we will be able to continuously iterate on that from a training modality where you do not need the same number of humans actually training on this on a regular basis because when the balloon goes up, you don’t need 10,000 people that are trained on algorithmic performance on these systems, you need a 10th of that based off of your efficiencies.
Kyle Woo:
Yeah, and kind of going back to what I was saying, I think it’s modeling and simulation, right? In order to keep the guy and the B-Doc or gal up to speed, they need the tools to regularly train against modeling and simulation where it’s not them going against the real threat. And I think we all want the full end autonomy. We want man on the loop, not man in the loop, and every single person in this room wants that, but I have yet to see anybody write the con op and TTP that allows that, right? No one’s comfortable enough yet to give those systems full, complete autonomy, even if that technology exists, but in order for us to get to that point where we’re comfortable, it’s that modeling and simulation approach. Let’s simulate hundreds of targets coming in and see how the system works and prove to ourselves that we can be confident in the AI algorithms that we make in the autonomy.
Maj. Gen. Charles Corcoran, USAF (Ret.):
Ryan.
Ryan Tseng:
Look, I think it’s a really great question and I don’t know if I have a fully thought-through answer here. I think that one of the things that we could just start doing is tracking how long it takes us to execute things. If we decide that, if in an exercise, or leading up to an exercise, we say, hey, we want this capability, right? And it can be based on some war game outcome, but it’s like, hey, guess what, this has changed. And let’s just track whether or not we as a team can put those capabilities into that exercise at the pace of relevance. Can we deploy it at the pace of relevance? Recently, Shield AI, working together with General Atomics, decided to put some of our autonomy capabilities on the MQ-20, and I just want to compliment, I think the Air Force has done some tremendous work on reference architectures, and General Atomics and Shield AI have also done a lot of work. But it was basically a napkin concept, I think about three weeks ago, we said, hey, we should put autonomy on the MQ-20 and fly the thing. And Shield AI spent about $100,000 to go from nothing to live flight, and the combined team did that in three months, where the first month was us just deciding whether or not we actually wanted to do it. But I think it’s a really important thing to be tracking how long it takes you to conceive a capability and deploy that capability, and that’s an area where we, Shield AI, are excited to continue leaning in, is just tracking how long it takes to do capability generation, tracking how long it takes to do the deployments, in partnership with the government, and just figuring out how we can collapse that OODA loop or that cycle time to generate new relevant capability.
Maj. Gen. Charles Corcoran, USAF (Ret.):
Thanks. And since we brought up the topic of man on the loop, man in the loop, let’s transition over a little manned, unmanned teaming. Andril’s on the leading edge, one of two companies selected as the initial CCA providers. Can you talk to us a little bit about the work in that area that Andril’s doing, and any lessons you might be gleaning from Ukraine that’s helping you with that?
Steve Milano:
Absolutely, and I’d be remiss if I didn’t point out our Anduril MUM Tea here, the manned, unmanned team. A little play on words there. It’s delicious, by the way, if you get a chance to stop by and have some. Arnold Palmer with caffeine, limit yourself to a handful a day and you’ll be okay. So, yeah, so I think across the board, and actually, I’ll probably start off with the kind of advanced effects and long-range munition side of the house, because that’s probably where we have the most recent activity. And so, wasn’t my news to break, but since you already mentioned it, the Enterprise Test Vehicle Program, DownSelect with Zone 5, and Andril as the two performers moving forward. The design cycle, too, is really focused in on network collaborative autonomy, integrating those behaviors, understanding kind of what the capability stack will get you in that platform, whereas phase one was more of a hardware-centric, can you build something fast, going back to those design loops and understanding what they actually look like. From six month, from contract to work to six months later, can you fly a clean sheet airframe? And I think that was, it was an impressive prototyping experimentation activity, but now it’s becoming a, okay, how do you turn this into an operationalized system? And so, where this gets back into bringing autonomy into the hardware stack, it is, how do you fly things that don’t require you to have 300 or 400 flight tests to prove out every modality, because every different permutation of a captive carriage or a live flight test needs to be modeled for you to get safe separation data and those types of things before you can ever get certified to go and actually strike targets. And so, what we’re going through right now and what we have, what Andril has a lot of experience on is taking autonomy behavior integration across platform and tying that back into new and novel platforms. And so, the ability to take learning off of surrogate aircraft and other aircraft that are flying in operational spaces allows us to take that data from operational theaters like Ukraine, fold it back in, understand how autonomy performs in both stressing EW and in other environmental spaces, and understand how does that actually tie back in and enable you to do something better, faster, quicker on existing hardware. And so, as we get through, I look at these examples of where acquisition is doing it really well and they’re going quick. I think a lot of what Ryan said, you guys are doing a lot of that demonstration work, a lot of what the ETV program is kind of instantiating as a baseline model for how you can go quick and take lessons learned and integrate. These are good use cases for how can you deliver other capability. But leveraging that data back into actual hardware and then building it out into the systems of systems and understanding what does autonomy give you in the manned-on-man teaming attribute, it’s a complex scenario. No manned fighter wants to fly with something that’s unsafe, and so there’s a whole process for us to go through to make sure that that proves it out. But what starts with single system autonomy morphs to homogeneous autonomy, then heterogeneous autonomy. And as we kind of go and scale through those different permutations, I think that we learn a lot from the things that are flying today and make sure that we’re folding those back into behaviors as we integrate them.
Maj. Gen. Charles Corcoran, USAF (Ret.):
We’re getting low on time here, so I’d like to hit one that’s near and dear to the Air Force, I guess. It’s one of the things that we pride ourselves on being able to do as an Air Force, and certainly Space Force, as part of this, is holding global targets at risk. We bring range and payload to any place on the Earth, right? And while drones have arguably made the close-in fight a no-man’s land, the struggles both Russia and Ukraine have had with drones is on the long-range strike front. They’ve had some successes, but they’ve had more struggles than successes. There’s challenges with battery life, challenges with assured communications, navigation, et cetera. For all of you, where do you see this going? How do we get to the promise of affordable mass in the long-range strike area for our Air Force and for our nation? Or are we going to be relying on bombers and long-range cruise missiles, things like that?
Steve Milano:
If I promise to be quick, can I go first? So I think that the caveat that we will start seeing to affordable mass is affordable effective mass. Affordable mass is important because there’s a capacity problem that we’re up against. And so things like we’re building with the ETV program, like the proof point in that is, do you have surge capacity? Can you actually build out and ramp capacity very quickly? But the very next question is going to be how effective can you be at that low-cost point? And it’s going to be a challenging problem. There’s going to be different CONOPS and modalities that we go after, but I think that that’s going to be critically important as we look at the long-range strike capability and offer a credible option, not a replacement, a credible option in the toolbox to bifurcate the need for deep bomber strikes into enemy territory.
Kyle Woo:
Yeah, and as you were saying, with 90% of drones that are being launched right now are being mitigated, but there’s still 10% that make it through. And I think one of the lessons learned from the Russia-Ukraine war is quantity, right? And then this is kind of building off of what you were saying of the ETV effort is, how can we as a country build in quantity at the low cost that Ukraine has seemed to be able to do? And we were talking about this just before this being up here is like, there is no one company that can solve this problem, right? The munitions directorate chose both of us to solve this, to be able to build this at scale. And then mass quantities also drives down price, right? We all have a phone in our pocket. If we own, you know, a company only built 10 of them, they would be pretty expensive. But when you build things at scale, it drives costs down. And then leveraging the autonomy to be able to communicate with each other so that if, you know, one of our things is fine with one of their things is fine, we can all work collaboratively.
Maj. Gen. Charles Corcoran, USAF (Ret.):
Great.
Ryan Tseng:
Look, I think it’s a super complicated problem. And so I’ve had the honor of, you know, working alongside, you know, people in the DoD on the mission of the last 10 years, I’ve learned a tremendous amount. I’ve also learned a lot about what I don’t know. So I feel ill-prepared to come to the table and say what that future is gonna look like, but I have conviction around a couple things. Number one, I think autonomy is very important to the future. And I think that that applies both to short- and long-range platforms. And then I think that, you know, there’s just a very, for me, there’s just this very challenging problem of like cruise missiles and ballistic missiles can be pretty long range and they can hold infrastructure at risk out to great distances. And like, as long as those missiles have that range and that lethality, it just creates, you know, a serious conundrum based on my understanding when we think about maintaining the scale and the tempo of operations that, you know, we feel like would be needed to continue to be effective in the new world. So I don’t have answers other than thanking all of you for your service, for being the smart people that you are, thinking through this very challenging problem. I’ll say that Shield AI, and I think everybody up here in industry is ready to contribute in whatever way that we can. We know our technologies, we know how to build stuff, and hopefully we can learn from all of you and as a team solve those hard problems.
Maj. Gen. Charles Corcoran, USAF (Ret.):
Thank you all. Yeah, I’m glad you all hit on the price point. I got a stat here I wanna read about how Ukraine’s building drones much cheaper than the US. So the US Army intends to buy 11,000 short range reconnaissance quadcopters over the next five years, $260 million program. The projected cost per unit is 20,000. The camera on these little guys is $4,000 each. Ukraine is manufacturing similar drones for 1/20th of the cost, including a $150 thermal camera. So getting competitive, getting that price point down is key. Anything on that, Ryan?
Ryan Tseng:
I think there are a couple things. One is volume, right? So volume is one of the most powerful things in manufacturing and production that you can have going. One of the lead engineers on our team was one of the key people in the Starlink program, building the Constellation, also the ground terminals. He led the engineering of that program for a while, up ’til it was worth about $100 billion Constellation. And if you look at that dish, it looks a lot like a picture of an ASA radar. And obviously there’s differences, but this thing is forming a beam, it’s tracking satellites, and it’s doing incredible stuff from an RF perspective, from a software perspective. And you all know, or maybe you don’t know, but ASA radars are super expensive, and you can pick up a Starlink terminal. Obviously there’s a difference between them, but you can pick up a Starlink terminal for, I don’t know what the price is right now, it’s like 300 bucks. But volume is an insanely powerful lever. And so I think that anything that we can do to increase volumes is important. Number two, I think that the way that you approach manufacturing, obviously there is an art there, and there’s a workforce cost. But I have a lot of conviction that the United States can be competitive on the price points when the volumes are there. The third piece, and this is where I think industry and government need to work together, is just the overhead cost of program requirements and the way that programs have traditionally operated. It drives quite a bit of expense, right? And I think that finding ways to be more agile could significantly reduce price points. Anybody else comment on that?
Steve Milano:
I think just lever to pull on that requirements thread a little bit. I think that there’s a lot that we can do to reduce kind of the overhead, the actual indirect cost of producing the thing, because what you’re looking at is what you’re seeing going into some of the low-cost Ukrainian drones is direct cost of those things with very limited overhead. And that’s both industry overhead and government oversight, program management type execution. And so the more we can reduce that, better off I think we’ll be. We need to get streamlined in those processes for the things that are important, but maybe not critically complex, right? We just need to build a lot of them. The other thing is on the requirements front is the allocation and flexibility of requirements, which I know is odd for a products manufacturer to kind of articulate, because typically stable, consistent requirements are where we live best. But if we have flexibility in the requirements baseline, it allows us as an industry to kind of bring forward more options that may potentially not have been valued if you were beholden to a series of mil standards that would drive you to a very specific solution that’s not readily available or at higher cost.
Kyle Woo:
Yeah, we all build unmanned systems, and I think we can all agree the most expensive components that go into those are the sensors. It’s the radars, it’s the infrared cameras, it’s the IMUs, these are the components that largely make them so expensive. Part of that is we all have to be NDA compliant, right? We need to source these from trusted partners to put inside, and some of that comes with cost. Some of that gets dropped down with quantity. But I think a large part of it is diversifying US companies that can manufacture this so that we don’t have one company that we go to for infrared cameras. We can go to a suite of companies. And I think that competitiveness will eventually drive down cost and also with the quantity that gets us to lower cost systems overall.
Maj. Gen. Charles Corcoran, USAF (Ret.):
Great. Unfortunately, we’re about out of time, so I’m gonna give each of you 30 seconds to make a final comment if you’d like, and then I’ll make some closing comments when we’re done here, so, Ryan.
Ryan Tseng:
Again, just wanna thank all of you for what you do. We’re honored to have the opportunity to contribute to the mission.
Maj. Gen. Charles Corcoran, USAF (Ret.):
Thank you.
Kyle Woo:
Yeah, thank you so much for the opportunity. Appreciate it.
Steve Milano:
Corky, appreciate you doing this. And I think I mentioned it to both of you before we came up here, this is an ecosystem. Yeah, we tend to compete on a lot of the same things, but we definitely learn a lot together, and so I learn a lot from you guys. I appreciate y’all.
Maj. Gen. Charles Corcoran, USAF (Ret.):
Thank you all very much. You made this panel. How about a round of applause for our panel? Good friend of mine, Paco Benitez, who happens to be a Shield employee, has a newsletter called The Merge and podcasts. It’s crazy successful. If you’re not a subscriber, you should check it out. But after the unfortunate loss of some US service members at Tower 22 in Jordan, he was waxing poetic on some observations that I think are worth repeating here. So these are four observations he made after those unfortunate deaths. First, high tech doesn’t win and low tech doesn’t lose. It’s all about the right tech properly applied. Second, as those in the profession of arms know, every technology and tactic has an offset, an asymmetry waiting to be found and exploited. Failure to continually evolve and adapt cedes time and advantage to be offset by others, a position of failure. Third, this is an infinite game. It has no end and doesn’t neatly fit into fiscal years, palm cycles, programs of record, or a line of congressional language in a bill. And finally, innovate or die is a common mantra in the startup world, but in national security, it’s literally the way of life. Those are great words by Paco. I’m gonna end on something that may be controversial and then anybody can stand around and debate me afterwards, but I think the reason we’re talking about all this close-in drone warfare and all this death that’s happening in Ukraine is fundamentally because neither side could get air superiority. And I’ll leave it at that. Have a great day.