From Zero to 150 Robots in Just 20 Months | Mike LeBlanc, Foundation Future Industries

Humanoid robotics development timeline

0:0020 months ago, we didn't have a robot. We had smiles on our faces and the idea of humanoid, right? And today we have robots already deployed out in the world working full shifts. I think all of this is a lot closer than than anyone will think. I wrote something last year about humanoids and how the promise is way out ahead of the reality. And there are all these problems with power supply and the strength of actuators and the generalization of the models that control them. We're also trying to get the robots out. That's why I say onto the battlefield in Ukraine to really hope to actually start putting them out there and seeing how do the soldiers want to use them? How do they actually work in a fight? How do they go up against drones? How do they go up against explosions?

Foundation Robotics Lab ambitions

0:45I first interviewed Mike LeBlanc, co-founder of Foundation Robotics Lab, back in January when the company was making a public relations push, claiming they were building humanoid robot soldiers. Mike was claiming they would produce 10,000 robots this year, when even Tesla fell far short of the 5,000 it had hoped to make in 2025. Mike is good company, smart, engaging, and by his own description, an optimist to the core.

1:19I, on the other hand, am a skeptic to the core. At the end of our podcast recording, I asked if I could tag along to Ukraine when he delivered robots there for testing. The trip happened in February. We got to Kiev, but the robots were stopped at the Polish border and never arrived. Nonetheless, Mike's public relations team managed to get a big spread in time, the venerable media brand now owned by Mark Benioff, Salesforce's CEO.

1:52The article talks a lot about robotic warfare, but fudges the question of whether any Foundation robots ever made it to Ukraine. It said simply that, in February, two phantoms were sent to Ukraine, initially for frontline reconnaissance support. But that was enough for news aggregators to pick it up, and pretty soon I was seeing headlines like, Robots in Real War, Ukraine Battlefield Tests Phantom Mach 1, Ukraine Deploys Humanoid Robots in Battlefield Reconnaissance,

2:30and robots are heading to war, being tested in Ukraine.

Touring the San Francisco operations

2:35When I visited Foundation's operations in San Francisco in April, my skepticism grew. Mike's claim that the company went from an idea to building a complete humanoid in under two years left out that he and his founders started by buying an existing humanoid robot company named Boardwalk. What's more, some of the military contracts he mentions were carried over from that earlier company. Foundation's shop felt like a research lab, not a factory.

3:08The robots on the floor were relatively lightweight machines, nothing that could withstand the recoil of a military weapon. An engineer giving me the tour gave one a shove, and it stumbled onto its knees, held up by a harness. The operation was producing about one robot a day, which made 10,000 this year seem impossible. The whole time, I kept trying to square what I saw with the company's pitch.

3:40Machines that would cross rough terrain in a war zone and carry weapons.

Updates on production and new models

3:46I spoke to Mike recently, and he has several updates. A new, more robust model of the robot has been completed, six feet tall, 185 pounds, with arms that Mike claims can each lift 80 kilos, more than 170 pounds each. And the company is opening a new production facility in Texas. They've scaled back this year's production target to 6,000 robots instead of 10,000.

4:17Mike said that Foundation is building its own AI model to control the robot. But see my earlier episode with Sergey Levine, a renowned AI researcher, with his own startup, Physical Intelligence, where he talks about how difficult that is to do. Mike said Foundation is building its own actuators, the tiny motors that control the robot's joints and are the most critical part in any humanoid. That's ambitious.

4:48None of this means Foundation won't get there. I've been confidently skeptical about companies before and turned out to be wrong. As I told Mike, Foundation's PR is running far faster than its robots.

Mike LeBlanc's background and career

5:03Here's the conversation. Why don't we start with having you introduce yourself and talk about your earlier startup, and then we'll get into Foundation and what you guys are doing. Great. Yeah, so I'm Mike LeBlanc. I was born and raised in Ohio and then left home to go to Annapolis, Maryland, where I studied the classics and saw the books behind me, learned to translate ancient Greek out there, and got deep into the classics.

5:33And as I did, that started to make me really look at war in a deep way, you know, looking at the facilities and the food tower and around the cities, and joined the Marine Corps, much to my family's dismay. I was in the Marine Corps for 12 years total. I did three tours to the Middle East during that time, and I was at the Pentagon. And it was at the Pentagon that I started looking at startups, and I was looking at electric tanks and starting to vet what, you know, pocket drones, how these things could actually come in, how they could be used.

6:05And so as I did that, it started to make me really look at a lot of these companies. I was getting married at the same time, and so got out of the Marine Corps, went to Harvard Business School, and then as I was out of there, co-founded Cobalt Robotics, which was back to, you know, we did our seed through Bloomberg Beta, our A through Sequoia, our B through Cotu. So, you know, ran that for a number of years, grew it into nine countries, four continents, having robots all over,

6:35and then sold it in 2024 and started Foundation Future Industries, which is now the sole provider of humanoids to the Department of Defense.

The reasoning behind human form factors

6:45Yeah, and so we were talking last time we spoke about this humanoid craze that has kind of taken over in robotics. And I was saying that it seems that there are many form factors that are more practical than the human form. And I was asking why, and maybe you can answer for listeners,

7:15why everyone is focused on humanoids now. And I know that the bad answer is the world is built for humans, and the easiest way for robots to navigate is in the human form. But there are all kinds of problems with the human form factors. You know, balance and speed and battery life and all these different things that, frankly,

7:46it'd be easier with either a wheeled robot or even a robot dog. We talked about the robot dog. So, yeah, give me your view on the human form factor. You know, I think that we're really a species of workarounds. In everything that we do, we have workarounds, we have shortcuts, we have ways that don't really fit the way that we're supposed to do things, but that's how they get done. That's, you know, part of becoming an adult is realizing that the org chart doesn't represent

8:17the full view of relationships, right? And so, as you look at all of these workarounds, you do have all kinds of things that just need to be, you know, through a factory, that need to be carried somewhere that you didn't think that they did, things, parts that need to be discarded, things that, you know, they need to adapt. And as you really start looking at that, you need dexterity and locomotion to do it. And so there are a lot of things that you can think about what form factors that could fit. But the only form factor that we know fits all of these jobs is the human form factor because humans are doing it today.

8:50And so, you know, to everyone who wonders about the human form factor, and I always did, you know, I would turn them to the question of why do we still have so many humans in all of these factories, right? I walk into them and they're packed with robots, right? There are all kinds of ABB and FANUC and industrial robots, but there are also human beings everywhere. And so that's always my first question when I walk in, as I say, what are all these people doing here? And that's what we'll spend the next couple hours as we tour any facility talking about is why are they still humans?

9:23Yeah. And is it the form, the human form, or is it the human judgment? I mean, is it the physicality or the mental capability that humans provide? Well, the human hand is incredible. So that's a big portion of it. You know, I think about that every time I'm balancing a mug of coffee as I carry in groceries and have a kid in my arm, right? And I'm still able to open the door. The human form factor is pretty amazing for things like that.

9:55But it also is, it's, you know, being bipedal and being able to move around all these spaces, being able to sidestep, move forward and backwards without taking up a lot of room that, you know, that wheels might or that a base might. And then also the human judgment, right? When you put all of those things together, you start to really get an appreciation of the human design, right? You start to learn a lot about the eyes when you're going into perception. You start looking a lot at the mouth when you want to have the robot have speakers. You know, all of these things are actually designed very well.

Dexterity, strength, and hardware design

10:25Yeah, and I mean, you mentioned the hand. I wasn't going to go into the weeds right away, but I did a piece for somebody on the Wuji hand, this hand out of China that uses, it has the actuator, the motor, you know, instead of the, most of the hands human, the robot hands today, the form human form factor hands

10:56use a tendon system. So the motors are buried in the, what is the wrist, and then there are tendons that run up and work the fingers. But there are a lot of reasons why that's not ideal. And there's a company in China that makes tiny actuators and they have a subsidiary named Muji Tuck and they have built a hand with the actuators

11:28actually at the joints. And it's an incredible, if you haven't seen it, look it up. It's an amazing hand. How, what, how do you, so the form factor, meaning bipedal, bipedal, I mean, there are two legs for locomotion, but how, how are you, you guys working on dexterity? I mean, what, what hands do you use? Well, there, there are two questions about, about the hands, I think.

11:59One is how dexterous are they? So how small of an object, how thin of an object can they pick up? Can they pick up a pin? Can they, you know, can they do electrical wires, things like that? Um, the other is in strength. And so how much, how much can they lift for how long can they lift it? Uh, you know, can, can they pull these things up? How much torque, torque can they produce? And so when you're, when you're really looking at dexterity and strength, I think that there are ways to have actuators in the hand that have lots of dexterity. And, and I think if that's the hand that I'm thinking of,

12:29it's, it's an incredible one. Watch it all over YouTube. And, uh, and immediately went to my hand team and I said, why aren't we using this? And they, they wrote back strength. Yeah. As a matter of fact, that's right. It's only got like 40% of the strength of a human hand. Yeah. Yeah. And so that's, that's where I think right now, as we're seeing that there's so many humanoid companies, it's a horse race for all of this. And you have to really think about where, where you want to land in that market. We are, we are fully dedicated to, you know, businesses and defense.

13:02That's, that's what we're making our robots for. Uh, most other people are making it for, uh, the homes and business. But when you're making those, those kinds of design trade-offs and you're looking to fold a lot of clothes and do laundry, you're, you're looking to, you know, empty a dishwasher, all of those things, they don't, they don't require that much strength. And so when you're looking at defense use cases, when you're looking at, uh, industrial use cases, you want as much strength as possible. And so I think for us, that's why we've, that's why we've still kept things in the forearm, but these, these are all the, the kind of choices we're out there with,

13:34with customers. We bring back all of their pieces. So if you walk into our office, you'll see parts from all of our customers. Uh, you'll see a lot of auto racks. You'll see a lot of, uh, you know, big, big metal clamps, uh, that we're working for ports. You'll see all of these things because that's what we test on. That's, that's the standard that we need to get to. Yeah. Yeah. And, and you, uh, uh, you guys are building both the hardware and the software. Is that right? Yes. Yeah. I think, I think most, most of the companies are there.

14:05There are a couple of breakouts. Uh, there's, there's physical intelligence. There's skilled that are both, both really just working on, on the brain. Uh, I think all the ones that were just working hardware seem to be getting scooped up. Uh, but, but we'll see. I, I, I think that really this is, this is one of those problems where the humanoid is a lot more figured out than, than I thought before I got into this. Uh, and, you know, as I've gone around and seen everybody else's robots and, you know, you go to Germany and, uh, DLR, they're kind of

14:37NASA over there and looked at their humanoid. You really start to see all of the parts of the same. Right. And it's, you have a, you have a choice of what you're going to buy for, for each part, but it's, it's kind of, it's, you know, it's really dwindled down. The only one that really is the standout is, uh, agility still has very different form factor, but for everybody else, the, the hardware isn't, uh, you know, you, you kind of know how to place it together and everybody needs to solve hands. Everybody, you know, everybody's facing the same challenges right now. Yeah. And actually that's interesting because that's one of the things that surprised me that

15:10you were able to jump into this extremely technical market. Um, I mean, you had robotics experience with cobalt, but still, uh, uh, it's, that's, that's a pretty brave leap, but is it? Well, I, wait, before I, before I take any credit on that, my co-founder, Sanket, you know, he, he has all the, yeah, between him and, and Patrick, uh, Bendersbach, our, our chief AI scientist, they've got so many papers all published all over and cited all over the

15:43place. And these, these guys are the brains behind this. Yeah. Yeah. No, I understand that. And, uh, but, but is it like the auto industry that, uh, you know, the different factories stamp different, uh, shapes of sheet metal, but, or, or whatever they're using now, fiberglass or plastic, but, uh, but the, the actual, uh, parts, uh, most of the parts are coming from suppliers who supply to all of the, the major,

16:20uh, car makers and, and, you know, you, you have some custom customization, but, you know, a car seat is a car seat and, and, uh, steering wheel is a steering wheel and, you know, the bearings and all that stuff. Uh, is it like that? I mean, is there, uh, a world of suppliers to humanoid robotic makers or, are, are, are these car companies like yours building bespoke, uh,

16:53parts in the house? We're, we're building a lot of, a lot of bespoke parts in house. So we, we have CNC machines, a lot of this, you, you know, especially for me doing the second time around with lots of manufacturing in the past, you, you start to see, you think that your suppliers are these big vendors that make all of this great stuff and you can never, never touch their business. And then you go visit them and you find three guys in a storage unit somewhere. And you're like, you know, hooked up to a generator and you're going, is this is what I've been paying? No

17:25wonder I, no wonder I can't scale. Right. And so, so you kind of end up taking those things in the house. And we've, we've done a lot of that this time round of, of really saying what, what is a part that we, that we can make? How can we get as close to just taking things from nature and turning it into this, into this robot? Uh, but you know, with that, you, you are still going to have a steering wheel. You're still going to have tires. You're still going to have all of the basic parts, right? But some people are going to be building Ferraris and some people are going to be building how to civics. Yeah. Although when you get to actuators, that is, uh, an industry in

17:57itself. And, and I, I, you know, there's no point in, in trying to build actuators from scratch, uh, when their companies have been doing it for decades. So, uh, and we, we are building us from scratch. So our actuators, yes. And our, our actuators are really our, our differentiator, I think. Um, you know, because you, you do realize anybody that's, that's, you know, breaking down a humanoid robot, almost all the cost is in the actuators. That's, and, and that's where almost all of the speed comes from. That's where, that's where a lot

18:29of the problems are going to come from. So it's, it's, you know, it's one of these critical parts, right? The hand needs to be, you need to get the hand right so that you can do all the skills, but you need to get the actuators, right? So that it doesn't break down all the time, that it doesn't, that it can actually make economic sense, all of these things. So, so we are building those, those in-house. Um, and we've been, you know, we've, we've been really excited about what we've been able to do on those. They're, they're extremely smooth and watching, watching a robot, you really, really can see the speed and, you know, uh, yeah, it's, it's,

19:00it's really amazing what our robot's able to do on that. Yeah. That's interesting. Okay. Well, let's, let's,

Training military tasks and vision language models

19:07pull back a little bit. I'm sorry. I got so into the weeds so quickly. Uh, so you're building humanoids for industrial use and, uh, the military. Uh, and when we spoke before you were saying, uh, that you're building, uh, at least for the military, you're focusing on, on specific tasks, uh, I think you mentioned, you know, getting out of a Humvee or, or, uh, you know,

19:39being in a combat situation and having the robot go and place a charge on a door to, to blow open a door, which is for obvious reasons, a very dangerous thing for a human being to do. Uh, yeah. Can you talk about, uh, the, the, the, how narrow the tasks are that you're training these robots for? And, and does, do you, do you have, are you developing a control model? I mean,

20:13an AI model that, that controls a robot that is, that generalizes enough that you can, uh, use different hands on the robot for different tasks, that sort of thing. So we'll, we'll keep, we'll keep all the same stuff on the robot for, for defense, for business, the robot, the robot will stay, stay as the same product. Uh, but yes, we are, we're building world action models that will be generalizable to all of these tasks. Right now, what it's looking like is about a hundred hours

20:45of video data that we, that we put into each of these tasks. Uh, and, and that, that seems to be giving us out what we need. Uh, and that's, that's how we've gotten our, our first robot, you know, working 24 hour shifts, five days a week on, on the assembly line was, was through getting about a hundred, a hundred hours and, you know, tweaking there, there are all kinds of little things that go into that, but, uh, but that, that makes it fully autonomous. And so I think throughout defense, we were the, we're the only company that's selling to humanoids, uh, selling humanoids to the department of defense. And, and so you kind of, you kind of go where

21:19the case leads you, right? You have, you have some people that are looking at logistics, uh, like motor transportation, fixing trucks, uh, aviation, doing all of this maintenance, refueling, things like that to, to aircraft. Uh, but you also have people that are interested in, uh, you know, kinetic side and, you know, really protecting soldiers. And that is just, just like you mentioned, putting, you know, putting a water charge on a door, being able to blow that up, uh, being able to carry out, uh, dead bodies off of the battlefield is very, very dangerous tasks. A lot of people do being the point man in a patrol. So, so there are a lot

21:52of those that go into this, this real protector and shielding mode. Uh, and I think right now it's, it's really just take, take as many videos of all of those things as we can. And we, we start building the models. So we're, we're getting more specific about which ones we're going to automate, but we're, we're very much hoping to get these onto the battlefield this year. And, uh, you mentioned, uh, uh, the model, the, the, the kind of model these, uh, you're training with these are, uh, vision, uh, language action models. Is that right? Is that VLA's vision

22:29language action? Yeah. Yeah. And they, and, and specifically, uh, latent space variable analysis is what we're doing, which is ultimately you're, you're understanding the relationships between objects. You're understanding the physics of what's happening through these videos. So rather than having to mimic any, any direct movement or, you know, like a, like a tele-opping would remember all of the different ways that I, you know, you're grasping a cup, this, this actually understands the relationships between all of these things and then can make decisions about how it should, uh, how it should go about interacting in the world. And we've, we've had a lot of success of

23:01really automating that with a lot fewer hours than I, I think it takes on a lot of other things. It's like a world model building of internal representation of, of the environment or of the world at large, not a specific environment, but it generalizes. So it understands laws of physics and that sort of thing. Is that, is that right? And then, and then you use the vision language action, a model to train it using a video. I'm guessing, I don't know.

23:35Yeah. Yeah. So it's, you know, it's really interesting. So everybody, everybody saw chat GPT and, you know, all, all of these companies got perplexity and opening high and all these that, that are all using large language models. And those, those are all working. We just threw a ton of compute at it and we had all of this language already out there. Right. So it's able to pull from all of this is huge corporates of language that we have on the internet, but we don't have all of that kind of real world data. So, so that big, there's big dearth and, you know, and actually

24:05just picking up, you know, cups and things like that. So we do have, we do have videos, but we don't know if they're, if they're close enough, if they're at the right angles, if, if there's a way to use, you know, I don't know if there's a way to use, uh, you know, Seinfeld, uh, reruns and be able to train a robot, but we haven't figured that out yet. So instead people have to turn to either doing this in simulation. Uh, so they, they just have to build a robot, like a video game and kind of train it there and then see how that comes on there. That's, that's a very good way to do it. Uh, except that it's very tough to overcome the gap between the simulation and reality when the robot actually

24:39does it. Uh, the second would be tele-opping. So that's where you always see these people with VR goggles on and the gloves, you know, and they're trying to try to control these things. You can train it to just do that over and over again. And the robot will soon learn, okay, that's, that's how I go there. Uh, and the other way would be video. And so for us, it is video. It's first person video and third person video. So we'll have someone wear a GoPro. We'll have them wear a, you know, either chest cam, or we've got two, two that go on a helmet. Uh, and then we've got, uh, other,

25:10other ones that are sitting on tripods outside of them to kind of show the larger action that they're working with them. And, you know, I think, I think as I've, as I've gotten, you know, more involved with the, with the AI and looking at what our AI team does, it looks to me like these are all, uh, it's, it's not really, you choose one, everybody is using all of them, but it depends on the chef, how much they use of each ingredient. And so I, I think that's where we're pretty heavy on, on, on videos and what we're putting into these. But what that ultimately creates is a robot that

25:41can understand and interact, you know, the, the world with the task that, that we're, we fed into it. Yeah. But I mean, generalization is the Holy grail for all of these, uh, robot control models. Uh, and, but you were saying that you focus on specific tasks, uh, that, does that make the training easier? And, and, uh, the, the, the, the robots operation that much more exact or, or not, I don't know.

26:17This is so ultimately, as we train all of these different robots to do these tasks, they'll, they'll all be out there collecting all of this different data that we will turn into a generalizable model. So I think this, this was true, you know, true six months ago. I think it's still true today. We're the only ones that are sort of taking the approach that Tesla took on cars. So, and, and that includes Tesla, uh, Optimus Tesla, I think is brute forcing a lot of, uh, tele hopping, but you know, the way that they did it with cars was they just had people drive and the car would just collect all of that data of exactly what was happening in the camera and

26:50exactly what did the driver do. And by doing that enough, it was able to create a model that would, that would work to be able to navigate the world. And so I think for us right now, we're picking very, very similar tasks to each other, but there are, there are lots of differences that come in, in the actual, uh, in the, in the real world and pulling those things off. And so as we collect up all of that data that will feed into a generalizable model. Yeah. Because even if you have a robot trained to do a fairly, uh, specific task, like placing a charge on a door, the minute it's on the ground,

27:28moving across the ground, all bets are off because there's all kinds of unexpected obstacles. How, and is, how do you deal with that? Or is that the, are you reaching that level of generality that, that the robot can handle? Yeah. So it's, it's, it's been really interesting as we've broken down these tasks because in almost everything we've done there, there's the base case and then there are contingencies. Contingencies have a very long tail, uh, but don't pop up that often. And so there's

28:01some ability to use humans and training side by side to, to have humans, you know, handle whatever that situation is. Whenever a robot's out there, for example, it will be out there with Marines, right. They can, they can go and handle some of these situations. Uh, but that's again, where we're finding contingencies are, are long, but they're not, you know, the, the probability of them is not as infinite as you would think, but everybody knows the videos of, you know, a Tesla that is the driving, somebody has been, you know, sleeping in their, in their car and there's a tree in the road and it crashed into that because that, that was just a tiny, tiny chance of that happening.

28:35Those things are going to happen. Uh, that's, that's largely going to be the case in these, in these early models that we deal with them. Uh, but it's amazing how many contingencies you can really plan for and how much video data you can collect to, to make up for them.

Testing robots on customer sites

28:48Yeah. Uh, and, and where, how are you testing these? Do you have, uh, uh, because again, you, you need to train them in an environment somewhat similar to the environment that they're going to be operating in. Well, as, as much as possible, we're, we're getting onto our customer sites and doing these. Uh, so that's, that's where we have all of our, you know, we already have robots that are working in auto manufacturing. And, and I, as I say, that's where we're the first humanoid company to do that for 24 hour shifts. Um, a big reason for that is we've, we plug the robots in right now.

29:22Those aren't, those aren't just on battery. Uh, but being able to take over 24 hour shifts, it teaches, teaches your robot a lot and it teaches your team a lot because our, all of our engineers, we have PhDs surrounding this thing at all times and they're hearing the loud noises. They're feeling how hot it is in there. They're, they're seeing how the press goes down and, and how they have to respond to all of these, all of these different actions. Uh, and so we've, we've been doing a lot of that where we're also trying to get the robots out. That's why I say onto, onto the battlefield in Ukraine, uh, to really hope to, to actually start putting them out there and seeing

29:58how do, how do the soldiers want to use them? How do they actually work in a fight? How do they, how do they go up against drones? How do they go up against explosions? Uh, you know, what are all these things we need to build? Uh, well, yeah, that's interesting. Uh, because these robots aren't cheap, particularly, uh, at the scale you're at now, I mean, certainly scale brings down unit costs, but, uh, going to Ukraine, uh, how many robots can you afford to, to lose?

30:29Yeah. Well, I'm, I'm, I'm headed there next month, uh, to really, to really look at this up close, but from, from what I understand right now, there's, you know, it's not, it's not quite trenches that everyone is in, but they're, they're sort of strongholds that they're in. And the huge danger is anytime you're out of one of those strongholds, if you have to go and get supplies from a truck, if you have to go and, you know, set, set a landmine, if you have to, anything that you have to do outside of that stronghold, the, the drones are just coming for you and, and you are, you are really naked out there as soon as you

31:04do that. Uh, so the, the thaw and what we've been talking to some of the military units over there about is, is instead of sending a human to do those tasks, send a robot. So, you know, will those robots get hit? They may, but they're, they're certainly, you know, cheaper and you'd rather be a robot than a human. Yeah, sure. Uh, and we talked about robotic dogs, which to me is just a much more stable platform, uh, than, than a robot on, uh, on legs, uh, on

31:36two legs. Uh, again, why, why not put it on a, you know, put an arm or two arms on a robotic dog? Uh, even you could even have, you know, a dog that's six feet off the ground, uh, but it's just a much stabler platform. Is there a reason? Yeah. Yeah. I think, I think we're going to see a lot of different form factors coming in land drones, right? We have, the skies

32:07are filled with them right now, but there are a lot of people competing in the sea and, and on land. Now, uh, the, the dogs, what the dogs have added to, to war, what they're, what they're really going to add is, is really incredible. Uh, but I think, I think, again, you're still going to run into things like if you need to search house, for example, if you need to go upstairs, if you need to drive a vehicle, if you need to, uh, you know, use a cruiser weapon, there, there are just lots of examples to pop up that don't make a lot of sense for the dogs that, you know, again, it was kind of funny when we were developing

32:38this, cause we started saying, do we need a head was one of the early questions. Cause we said we can do it right in the chest. And as we actually started doing it, cause it's also ultimately two light industrial arms, right? And we thought we can just put a camera right in the center. It's much more helpful to have a camera that's up. And so all of these things kind of start to make you appreciate the human form factor of going, no, that really does make sense. And you, you do like a neck to be able to crane and be able to look at things. You, all of these things really go toward a purpose. And, and yeah, Brian, any of this, if you want to, if you want to be, you know, pulling out, uh, pulling out

33:12bodies, if you want to have something that can detain people, if you want to have, you know, all of these different interactions, you could probably find a different form factor for each. Uh, but it'd be very hard to find a form factor that can do all unless it's a humanoid. Yeah. Uh, so you're optimistic about, uh, a bipedal robot walking through uneven terrain and carrying a heavy weapon or, or, uh, or, yeah, I, I think the, the uneven surfaces

33:43have been amazing and, and not just, not just for us, you know, I see what the other, what the other humanoid companies are doing and it's, you know, it wasn't too long ago that, that they were all kind of, you know, petering out there and that, you know, for, for us, that was, that was right during the, uh, during the election. And we all said, you know, our robot walks just like the president, you know, we were like, it has a presidential gate because it walks like president Biden. Uh, you know, I think a lot of people had, had that kind of shuffle going, but you know, as we've moved forward now, you see, you see robots that are running marathons. You see, you know, you see very, very agile robots

34:16and we've already started to hit all these uneven surfaces, be able to move through things. You know, we put objects in front of it in those, which objects to walk around and which ones that can actually plow through. Uh, so no, and this, this is all in 20 months of work. So yeah, I'm, I'm very optimistic about, about what we're able to do on this. Yeah, that's fascinating. And, uh, so you talked about the deployment in Georgia, uh, where the robots are, are telling to, to the ceiling with electrical cables that feed

34:50power to them, uh, to, so that they can operate 24 seven without people having to swap batteries and stuff. When you're on the battlefield, how do you handle power supply? We're, we're doing that with battery. And this again, really goes to the use case of how long do you need that robot to be running? So having a robot run 24 seven is what makes it make sense economically in a factory. If you're going to replace, you know, three, three shifts of workers that are each getting paid $60,000, you can make it make economic sense. Uh, on the battlefield,

35:23they don't need it to be working 24 hours. Right. So they, they do have lots of loitering drones and, and they have more concerns about what, what their airtime is. But for us, it really is these, you know, they would want to be able to turn on their robot, go pick up the supplies. The truck is here. We can, we can pull those out all out. We need to get a better perspective on, on a certain building, or we want to go look at the enemy's flank, send out some reconnaissance robots. So it's, it's going to be much more directed actions, I think. Yeah. And so your battery, what is the battery like?

35:57Right now, four hours, uh, you know, and we're hoping, we're hoping to extend that. And I think that batteries are, are one of those interesting things that again, you, you kind of have all of these different startups, right? A hand, a robotic hand is a company all in itself. A battery is a company all in itself. You kind of have all of these different ones, but batteries are a place that I think that the, uh, the industry is making a lot of gains. Uh, and, and that, that makes it easier for us to start picking up and getting much longer times. Yeah. And so the idea would be a team would go out, uh, with, um, a couple of these robots or one

36:32robot and a whole bunch of battery packs and they just swap the battery packs. And yeah. Yeah. We've, we've had a lot of, you know, the, the imagination goes, runs wild when you have a human shaped robot. And, and so we've had, we've had a lot of different, uh, different decision makers in the military thinking different things. Uh, but one, one that we've heard a lot of is attaching robots to trucks, figuring out some way to have robots kind of tucked up and on the sides of trucks that as they arrive, they can just pop these down and have the robots, you know, first walk

37:05a perimeter around to make sure that there are no landmines, no IEDs, anything like that. Right. There, there are all these kinds of different use cases that people are coming up with, but a lot of them converge on that. And, you know, so you'd kind of have the truck as a central place that would have batteries. Uh, and, and we will get to the point that, you know, our robots can change each other's batteries. So, you know, you will find this fully autonomous, uh, command center. Yeah. That's how, how, I mean, this all sounds very futuristic, sounds possible,

37:35but, uh, it still sounds, uh, like it's not, we're not there today. How, how confident are you that you'll be able to deploy successfully on a battlefield in a certain timeframe? I mean, what's timeframe you're looking at? Uh, you know, I, I think I'm, I'm just as confident as I am that my kids are going to be able to drive cars, that they're going to get married and buy houses. And, you know, a lot of this is you, you look at it in its infancy and it is hard. It's, it's hard.

38:09My kids are six, seven and eight years old. It's hard to look at them and start thinking, oh, they're going to do taxes, right? Cause, cause they're not there today. Uh, but when you really look at how quickly they're growing and when you really think, you know, when I think back to two years ago, how my kids were talking, how they were acting, how they thought right. And where they are today, uh, that's where, that's where you start to go. Yeah, we, we have this very linear view of what growth and improvement looks like, but for robotics, especially, I think if you look 20

38:40months ago, we didn't have a robot. We had smiles on our faces and the idea of humanoids, right? And today we have robots already, already deployed out in the world, working, working full shifts. We have, you know, we have robots that are, are ready to walk uneven surfaces and go out into the battlefield. So, you know, that's all, that's all just in 20 months. So I think 20 months from now is, is going to be shocking to a lot of people. I think all of this is a lot closer than, than anyone

Automotive factory deployment details

39:07will think. Let's talk about the Georgia deployment. Um, what exactly are the robots doing? Uh, so it's almost all injection molded parts. So, you know, all, all of these other FANUC robots and things all make all of these parts. They clip them together. They come out on a conveyor belt. And when we arrived there, it would be, you know, a human that's standing there, a person on an eight hour shift. The, the parts come out in different spaces on the conveyor belt, right? So there was,

39:38there's a little bit of adaptability that needs gone there. They, they would put a label on it. They would do a quality control check with a, with a camera. And then after they marked it good, they would put it into the packaging. If it was bad, they would, they would move it to a separate rack where it would, um, get, get taken away. And, and so, you know, as we, as we looked at that, we, we thought we could probably handle something like this. And I think it was what April to November, never do math in public, but I don't know, six months in or so, uh, that we did a proof concept on this.

40:09And, and we brought our first robot out. That robot was on wheels when we did it. We didn't have the legs yet. Uh, and, and the robot was able to do all the parts. So we, we did that for hundreds of parts for all their executive team. And they all said, this, this looks like it's worth moving forward. Uh, and so we, now we have a contract for 150 robots going and doing those tasks. Uh, and that's, that's what we're, we're deploying and we're, we're working to, to get all of those robots out. Um, but they're, they're very, very similar tasks, different parts, but very similar.

40:41Yeah. What kind of a factory? I mean, what's the product that they're assembling? It's a, it's an automotive OEM. So it's, they, they make parts for, you know, they, they actually, they make a part for my, uh, uh, Land Rover, you know? So it was, it's funny as you go out there, it's a little plastic cladding part on the outside of the door. And I go, oh, yeah, you can pop it off and see that's made in, that's made in Georgia. And the power, uh, presumably was that difficult to design the, the cabling so that these robots don't get tangled up? Not at all, oddly

41:16enough. Uh, you know, I, I thought that was going to be a real thing. It was the thing that we talked about a lot before, before we went out there. Uh, but really, so all, all 150 of these, operate in a 15 by 15 cell or smaller, 15 foot by 15 foot. Um, they, there's nobody that moves that much. And the first robots that we've deployed, they really only take a couple steps. You know, at most, they're probably taking five steps. And, and so it's interesting. We have, we'll have the robot move, uh, just move backward instead of, uh, instead of like turning,

41:48turning around and doing these things. And so the cable really just hangs there and can, you know, it's, it's like, um, like, uh, you know, my dog, when I put them on a, on a running leash out there, it, it just doesn't get tangled. Yeah. Uh, yeah, that's interesting. Uh, the dog analogy, the, uh, uh, where do you, so you guys are 20 months in, I mean, it's incredible. Um, and maybe it's, uh, it's, it's the fearlessness that comes with someone coming in fresh. I don't know, because a lot

42:24of the researchers that I've talked to are so cautious about capabilities, you know, uh, and frankly, what I've seen until fairly recently, uh, with, uh, optimists or they don't blow me away, uh, and, and they don't look very robust. Uh, but you know, you, you guys have, you know, made all this progress

42:57in 20 months. Uh, where, where do you see this going for you and for the industry? You know, to me, this, this really is one of those cases where, uh, the attitude and approach makes a huge difference in how the technology, uh, develops. And, you know, I think everybody saw there was a viral video, I think it was out of Germany or Russia. I can't remember where it was out of, but the robot that walked on stage and fell and yeah. And the guys run out to cover it like a car accident and they're very frantic about it. Uh, it is, it is not hard to find, uh, videos of our robot

43:32falling. Uh, but what you'll see when it falls, you know, is Sanket and I will be standing next to it. It will fall. And we continue conducting the interview as we lift it back up and get the robot working. Uh, because I, I, I think that we coming into this, you know, we're, we're not shy about saying this is very new technology, all of the stuff that we're solving, you know, we're, we're very, very honest about where it's at. And, but as you do that throughout the whole team and you don't have these, these kinds of marketing lies that are built up and things that you're hiding behind

44:06and trying to make these highly curated videos, uh, you have a team that is actually able to deal with the problem that's happening. And when you have very smart people that are directed toward those problems, you can get those problems solved. So I, I do think this is an interesting time where the, the experts in all of the specific tasks, uh, have kind of too much project fatigue for them to really see how quickly this is going to move. And so I'm, I'm still shocked. I'll talk to other founders and they'll be saying, you know, yeah, I think we're 10, 10 to 15 years out.

44:38And I go, yeah, you better get ready. Cause we're making 10,000 this year and I'm not going to keep those in a storage unit, right? These things are getting out in the world. Yeah. And that's, uh, I think I told you, I, I wrote something last year, early last year, about a year ago about humanoids and how the, the promise is, is way out ahead of the reality. And there are all these problems with power supply and, and, uh, the strength of actuators and the

45:12general generalization of the, of the models that control them. Particularly, there was talk and there continues to be of household robots, uh, that will, you know, be your, your housekeeper. Uh, and I would have been very, not skeptical that it won't happen someday, but that we're even within 10 or 15 years of that, that it's going to take a long time to get that level.

45:44This is, this is so funny. Cause I have, you know, from, from your alma mater, well, I call it the New York times I have hanging on my wall, the New York times, you can print any old copy of the newspaper, any page. Right. And so I went through the archives and found, in, I think it's 1906. I think it's October, 1906. There is an article that says why there will never be heavier than air flight. And it says so many people have been trying planes. And, and the, the analogy that it uses, this is the math that it does. It says, uh, you know,

46:15if you look at how long it took birds to evolve, right, you would already have, you would already have almost a million years. And to do that with metal, which is even heavier is going to be even harder. And so the, the writer says that it's going to be between 1 million and 10 million years for human beings to, to achieve flight. And three months later, you had the Wright brothers, uh, that were, that were first in flight. And, you know, so I, so I have that hanging, I have that framed on my wall because I, I think that we are living in exactly that time where, where the humanoids

46:47are in their infancy. I think that a lot of people working on them, uh, don't believe in them. I think a lot of them, you know, still think, yeah, there's funding coming in, but these things aren't going to turn out. Uh, but as I look at this, we, we are the Wright brothers. These, these are going to be working far sooner than anybody thinks. And, and people are going to be shocked at, at the change that it makes in the world.

Competitive landscape and market strategy

47:06Hyundai is famously, as they were featured on 60 minutes using, or at least talking about deploying Boston, uh, Atlas, Boston Dynamics, humanoid. Uh, when Boston Dynamics is tied up with DeepMind, Google DeepMind, who would be terrifying to me to have to compete with them. I mean, first of all, with Boston Dynamics, because they've been at this for such a long time, uh, and the criticism

47:38of their robots used to be, yeah, but it's all teleoperated and there's no AI controlling the robot, but now they've got DeepMind and, and, and they've got a, a refined form factor. I mean, it's pretty impressive doing, uh, in the video I saw doing something similar to what you're talking about in Georgia of picking up parts, scanning it for defects, and then moving it to where it needs to be. Uh, how do you see the competitive landscape? Is this a market that is going to be so large

48:14that there will be many players or, uh, do you, do you feel pressure from people like Boston Dynamics and DeepBrain? Oh, we, we feel a lot of pressure. Yeah, absolutely. I mean, this is, and that's, that's why we work, you know, six, six days. Some of our teams work six days a week. Some, some teams work seven days a week, uh, because this, this is a huge race against all of these, but you know, when I, when I look at Boston Dynamics, I loved watching that 60 minutes. I watched it with my kids getting ready the other day, the other morning, uh, they are doing almost

48:47exactly the same task. So if you look at what the parts they're using, I mean, they look almost identical to what they're doing. Some nuances in the video are they're doing it in a replica site to the side of the factory. They're not doing it on the manufacturing line. They're not running anything close to, to 24 hours. And, uh, as you actually watch the tasks that they're doing, they're moving a lot of things from racks to tables, which is a capabilities demonstration, but it's not the actual task. I'm sure that's not, nobody needs put parts taken from a rack and put

49:19on a table. And so when I look at that and I think about how long Boston Dynamics has been around and in space. And when I look at us and what we're doing today and how long we've been in space, uh, I don't feel that word. So you can, you can throw in lots of, you know, Google's deep mind. I mean, going against Elon, what's, what's more scary than going against Elon. Right. But, but as I look at those, I, I think a lot of people are making these design choices that are going to be much more lightweight. They're going to be hoping to get into the home and the ones that are focused on business like us. I don't see anybody that's moving anywhere as close to as fast as we are.

49:54Uh, so we'll, we'll see how many, how many winners there end up being in all of these spaces. Uh, but as of right now, I think as long as we keep pushing on this as hard as we have been, we've, we've got momentum and we're going to see these things really fly. Yeah. And, and do you, do you think that then there's room for many players? Like there'll be someone specialized in military grade robots. Uh, there'll be some that specialize in, uh, you know, household robots, or do you think there'll be kind of an iPhone of robots that, uh, that can be

50:29applied or tweaked? Yeah. I, I think that's going to look a lot more like cars. Right. And that's, that's why we've been really focused. We always say we're, we're building the pickup truck. Everybody else is building the Prius. Uh, I, I think that you are going to have to have these lightweight, low bomb cost, uh, you know, consumer robots. If you really want to get it out there to, to lots of homes, it just doesn't make economic sense to make it very expensive. So you really need to hit, you know, even the 20,000, 30,000 that, that Elon's talked about. Uh, that's,

51:00that's still pretty expensive for what these, what these robots are going to do. Uh, for us working in, working in businesses, when you're working on ports, when you're replacing specialized jobs that, you know, I just, just looked at, uh, Ford is, is having trouble filling all of these positions for $180,000 a year. When, when you're facing those kinds of things, you can make a much more robust robot. And so I, I don't see that landscape changing anytime soon. And I think there probably will be different models that are doing those things. Yeah. So there's room for, for many companies. Um, once it's solved, uh, it's, it's sort of an endless market, right?

51:40Yeah. I think, I think as, as varied as the cars you see on a road, right? So you see, you'll see a cycle, right? There's going to be somebody with a tiny unit tree, you know, those things are like three feet tall, but they'll have that for crawling in attics. I don't know what they're going to have them for, but you'll see that all the way up to, you know, very heavy and much larger robots that are able to lift a lot more. Uh, and who knows, who knows how big and robust those can get. Yeah. And, uh, you know, you don't have to talk about this, but you're new and you're

52:11venture, uh, funded, um, I guess you have a little bit of revenue coming in, but how long, when, what's your window? I mean, at a certain point, a lot of these companies are going to run out of venture funding. Uh, and, you know, get gobbled up, I guess, by people like Boston Dynamics or Optimus or, uh, yeah, no, the, the thought of being bought by Boston Dynamics is, uh, is absurd to me,

52:42but, uh, you know, they, we, we might buy them. You never know. They've, they've got some good stuff. I really threw it into something. Uh, we've, we've got very robust funding. And the, the interesting thing is this is, this is not a hard thing for investors to see and to see the power of, you know, I think if you look, if you look across human history, we've, we've really been subsidized by slavery for almost all of it. Uh, you know, when we finally finished slavery, it was with the help of the industrial revolution. And so robots were actually able to help us stop that. Uh, but then

53:15we just started new slavery under the name of globalization, and we just shipped it off to China. And so, you know, I, again, I was a classic philosophy major. Aristotle says the only way that you'll ever truly get rid of slavery is, uh, with automatons, with things that are able to do their own tasks. And I think that is the age that we're living in. And so when you're dealing with such an existential question, everyone just understands that in their bones. And so as you can show them, you know, this is a very visual product as they see what we're able to do autonomously. When people

53:46come, we bring investors to the factory to actually look at these robots. And it's, it's not that glamorous, right? They're not that fast. They're moving. They're moving within the same, same time that a person does. It's 60 seconds per part that they have on these. Some, some of the work cells are 52 seconds or, but they're all right around there as they, as they see it doing that. And they talk to our customers and they go, you actually sent three people home because of this, right? You're actually recognizing ROI. I think that is more surprising to them than seeing a machine that's able to walk

54:17like a human. The fact that we're actually employ robots to save money. So I think there's always lots more funding when you can, when you're saving customers money, you, you have very, a lot of power to go scale. And when you have a lot of power to scale, you have a lot of power to raise capital. Yeah. And I hate this question, but I, I feel compelled to ask because it's such a part of the, the discussion these days, uh, because I'm a tech optimist and I think that the labor market will

54:47figure itself out, but you, uh, you know, just that, you know, being able to replace workers with humanoid robots, does that concern you or how do you see that play? No, not, not in the least, you know, I, I really, I look at these tasks and you'll, you'll watch these people. I would, I'll, I'll sit up, they have little ladders that you can climb up and I'll, I'll sit there and I'll watch them for hours. I'll be looking across the factory and I'll be watching these people pick up a piece of plastic, hold it under camera, put on a label, put it in packaging.

55:19And I think all of these humans have stolen robot jobs. These were all stolen from robots and we're taking them back. I think that is, nobody wants to go to work. We, we hang with all these people. My, you know, my wife will come on site. She brings all these, uh, you know, fancy Starbucks drinks for everybody. And we walk around and we talk, talk with everybody in there. Um, I, I don't know anybody in the plants that we're operating in that is upset that the robot has taken job. It's, it's just the same as a FANUC robot coming in and doing something. Uh, you know, our, our most frequently asked question I always say is, uh, can I take a selfie with it? Right. People, people love to come

55:55up. They love to say we're working with humanoids. And, and I think that, you know, overall the best, uh, the best example is really ATMs, right? When the ATM came out, people said it's going to put bank tellers out of business, but sure enough, it actually just meant that we expanded the number of banks to the point that we have more bank tellers today than we ever did before the ATM. Um, so, you know, with any technology right now we're replacing a lot of jobs, but then I think very quickly people see that they can expand because they have this technology. So they go, we can start a new plant and then people start realizing they can do fully new tasks because they have

56:29this new capability. And so, you know, where, where all of that goes, I'm, I'm very optimistic about how many jobs that's going to create, how many human beings still need to be involved in those processes and how cool the work is that they're, that they're going to be doing because it's, it's very fun working with robots. Yeah. Yeah. Uh, and you were saying, uh, at your offices, you have robots roaming around, is it, are these prototypes or, or is that part of the testing? No. Yeah. So these are all, you know, if, if you walk in, you'll just see a lot of engineers

57:02focused on their computers and looking over at the robot. Right. And God knows what they're looking at and what they're looking for, but that, that is the job is really, we, we run them on these replica sites. We build exactly replica sites of our, of our customers, um, work sales. We do those both in our German, uh, German, uh, office and in our San Francisco office. And, but these, these robots are always walking around doing some kind of testing. Uh, sometimes, you know, we had to have, we had a customer that wanted to see how long the robot could stand. Uh, they

57:32wanted to make sure that it could just stand there for 24 hours without moving and having to catch itself. So there was one day that we were just videotaping that, uh, there are other days that for hours, the robots, uh, we just test how they do just walking back and forth, just a walk and a turn, just a runway walk. And we'll, we'll run that for hours just to see, see how it all goes. So it's, I always say it's like Willy Wonka for adults. It, it has this just really cool feel when you walk in there of, and, and you feel optimistic when you walk in there about what the future is going to look like.

58:02Yeah. Yeah. And do you think that you're, uh, on par with the other players? Do you think you're ahead of them is from what you've said you're, you're deploying more quickly? Uh, or do you think, uh, there are others that are ahead of you? I mean, I, you know, I don't want you to rank your competitors, but, but who do you watch and, and not necessarily worry about, but who do you see is, is on the right track?

58:34I think, I think figure and optimists look really good. Uh, you know, I think the reason that we've been able to deploy more and on, and on full contracts, not on, not on pilots and these, these little tests, uh, that really comes down more to solving the whole problem. You know, and I see this, this is in, in startups. I see this all the time that they, they kind of go in and they handle just some base case, but they can't make an economic case. There was, um, there was a robot a couple of years ago that was dipping fries into a deep fryer and taking them out. And they were saying, we're going to

59:07replace the fry cook, but they couldn't open the bag of fries, right? That, that to me is that, that's just encapsulates everything that, that people do wrong in their, in their go-to-market strategy. And so it's interesting when, you know, when you go back and look at Edison, you know, the most important part, I think of what Edison did, there were already light bulbs around, right? He didn't, you know, and the electricity has always been around. He didn't invent electricity, right? But he realized you needed to have a power plant and he realized you're going to need to do that for all of the homes around. And so he, he solved the whole product. He was able to put it out there and then made, made the innovations to make it cheap enough that

59:40it was worth people getting. It's, it's those, those kinds of human and economic conditions that you really need to satisfy. And that's why I think that we've figured out a way to get these robots into action a lot sooner than anybody else has by really just looking at the whole product, putting ourselves in the customer shoes and saying, what, what am I paying today? How robust does this robot need to be for me to really trust that this robot could do my part? And, and that's, that's what we've really, that's been the real innovation that we've figured out to get these robots out faster.

1:00:12And what, what is the pricing? Do you have a target price point? I mean, regardless of, obviously it's, you know, there, you, you can't, you know, optimize cost when you're starting up, but you were saying 10,000. What did you say? How many robots this year? These are, yeah, we'll, we'll make 10,000 robots this year. Yeah. So right now we're, we're only working with six customers. We're, we're keeping it very tight. And that's, that's because we have a hundred robot minimum at, at any

1:00:46customer. So the, the contract size we're working with are, you know, the, the largest contract that we're working with is 2000 robots, but they all kind of need to be in that range. And yeah. And for, for these robots, for the initial users, it's all a hundred thousand per year, per robot. It, it will go up from there as they become more robust, as they're able to take over harder jobs, the prices will go up. But for all of these initial customers, we have the same deal, a hundred thousand dollars per robot per year. And that's, are you essentially renting the robot or are they, they buying them?

1:01:19Yeah. Full lease that includes all, you know, us coming and fixing them. If they go down, it has all over the air software updates. Uh, it, it even has, if the robot goes down, uh, we have built into our service level agreement that a human being on our side has to come and do the task while the robot is down. So there, there are all kinds of contingencies about how this, how it's built up. And a lot of those, we took the same structure from, from cobalt, my last company, uh, where, where we did have, you know, hundreds, hundreds of these robots out there and, and had to really learn how, how do you fix a robot when it goes down in Singapore?

1:01:51Yeah. Yeah. Although I, I don't think you're going to be sending, uh, employees out to the battlefield to stand in for a broken robot. Uh, is that the same? Is the cost higher for the military grade?

Selling to defense bureaucracies

1:02:10Uh, for the military, we've been working different pricing in, in different use cases. Uh, and they, they have their budgets work differently. It's very hard for them to go on a subscription. So we're doing some, uh, outright buys, uh, and then they have, have some maintenance that they're able to buy on top of it. But yeah, there, there've been, we're, we're still, still evolving what those contracts are going to look like. Yeah. And, and you, you have contacts in the military having come from the Marines, but how did you penetrate, uh, that's incredibly complex bureaucracy? I've spoken to a lot of

1:02:43startups that, uh, have, uh, products that are relevant for the military, but they don't want to touch it because it's such a time suck trying to get through. I think, did you say that you're, you're, you have a relationship with Anderil? Well, so, so Anderil, we've had, uh, you know, Palmer Luckey has spoken through our robot, uh, several times. So we actually dress it up in a, in a Hawaiian shirt and the shorts and a, and a mullet, mullet wig. So it's, it's been very cool to see. Yeah. It's to see him talking through the robot

1:03:14and we'll continue to do that. Uh, whenever, whenever he can't make an event, a robot will be there. Uh, but no, really, really breaking into all of these things. I think that the military gets a bad rap for, for its ability to purchase things. Uh, I, you know, I've sold, I've sold to Amazon, General Motors, you know, Uber, DoorDash, Salesforce, you know, a lot of, a lot of big names, Citibank, you know, at, at Cobalt, they're all hard to break into. Yeah. It's, it's hard to sell things to a business. And, and the reason that, you know,

1:03:45there, there is a lot of red tape in, in a lot of these places, but the, the real problem is there aren't just full decision makers that can just simply say that's going there and nobody's going to say anything about it. It's every decision that they make is political. And so you have to build coalitions. You have to, you know, you have to get a large tent if you want to put these things out there. And, and I think the military is, is largely the same. So we, we've talked a lot with, with Congress about this. We've talked a lot with generals about this. We've talked a lot with, with just soldiers on the line. Uh, and

1:04:16you, you really have to work all those different angles, just as going into a big enterprise deal. You have to talk with the workers. You have to talk to their supervisors. You have to talk to the CEO and, and the private equity company that owns them. You gotta, you've really got to understand all those different aspects. And so, so I think it's, you know, largely the same for us going into the military. And which service are you working with? Uh, so we, we have contracts with the army, Navy, and air force. Uh, and then we're just beginning with the Marine Corps and we hope to hope to begin with, um, Department of Homeland Security

1:04:47this year. Yeah. Uh, and, and the, the point, uh, to get into those, do you approach each or are you going through the joint Jake joint artificial intelligence center? We go talking to everybody and this is, you know, for, for me, I, I of course have, you have everybody in my peer group is a Lieutenant Colonel. So I know a lot of battalion commanders. Uh, so there's just a large group of those in my phone that I can, can call it all these times. Uh, but there, there are also a lot of, a lot of generals that, uh, that I know

1:05:18through Harvard, a lot of Marine generals go through Harvard. Uh, so, so I know them there, there are just a lot of contacts that pop up and, you know, we, we were able to go to the white house, uh, last year. Um, we've, you know, we've, we've been in contact at the highest levels of government talking about this. Uh, and so I think a lot of it is putting it on number one's radar, getting a lot of, a lot of people comfortable with us and, and with what we're doing. Uh, and we'll, we'll start to see some, some very big contracts come out of that. Yeah. Well, you're, you're a very good communicator. Uh, I, I have, I have

1:05:55to say though, I remain, uh, cautious. Uh, well, you're, you're going to have to have me back on Craig. We'll, we'll check back in and 12 months. That's right. And I'm going to go, I'm going to go, Craig, we're sitting here on January 8th, 2026. And it didn't seem like it was going to happen. And we have 10,000 robots out today. Yeah. You know, yeah, absolutely. We'll see, we'll see where we're at.