Serve Robotics Turned Delivery Robots Into a Platform Business

Right now, somewhere, a company is making a move in plain sight. The street reads the headline, "Autonomy AI sees the signal. We deliver it before the market catches up. Autonomy AI, your models, our intelligence." Request a private strategic briefing at autnmy.ai Ali, since you last put on the road to autonomy, the company's done one thing and one thing only. You've grown and you haven't just grown robots. You've grown revenue. How do you continue to grow several botics? Well, class of all, thanks for having me again. It's been a little while. You know, it's, what do they say, like, 10 years of work? But, you know, one night of all the pieces coming together, basically. So it looks like overnight success, but really, it'd be been added for nine and a half years at this point. And we, I think we had the right time and right place with the right piece of technology. So it's a lot of tailwind as well. That's zoning on LA and I have to admit to the audience, I've been hooked on the World Cup. I've been watching three matches a day, sometimes four matches a day. And going through your Q1 filing, you just go see a 500 bots in LA. I have to ask, are any of those at SoFi Stadium moving stuff around the park and lots for the tailgateers here? You know, that would have been pretty cool. We had some conversations. It didn't end up happening, but we are very close to it. So if you go into that area anywhere from downtown LA, West LA, there's like a lot of areas in proximity that you'll see our robots all over. I was at two games in LA actually myself. So I was there with my daughters. We actually went to one of our depots and I got to show them the, you know, where the robots live. And that was pretty fun. Oh, wow. So the audience thinks of your traditional robot tax depot or your truck depot. They're big. The average depot, this will blow your mind for way more is 90,000 square feet. In some cases, it's going over 110, 120,000 square feet. I'm assuming since your robots, let's just do you probably eight, eight, the size of a car. What size of a depot are we looking at for sir? Yeah, they're like a garage. It's a small garage. Like it's not something as impressive as Ubers. We can have usually about 100 robots packed into a couple of thousand square feet. So nothing, nothing that fancy or crazy, but that's kind of the power. They can be very close to the population center with low cost of real estate. And what happens at those depots? Is it charging? Is it maintenance? Do you replace a wheel? We replace a lighter? What happens in those depots? Yeah, I mean, all of the above. So it starts with cleaning actually. So when robots come home at night, they get cleaned. They get just video health checks on robots, making sure everything is in good condition. And preventive maintenance that sometimes we have to do, we know that by X number of hours or miles, the robots would require a tire to be fixed or whatnot. And then there is of course, active maintenance. There's something has gone wrong or I don't know, someone drew something on a robot. We got to get rid of it. That sort of thing. So it's very logistical. It's very, the kind of thing you expect if you're operating a fleet of cars. It's very similar. I think the big difference between robot taxis and cars is your robots can have advertising them. They can get wrapped. And that could become a very significant source of revenue for you. Does the wraps happen at the depots? Let's just say, I don't know. Give me an example, you're an LA minions. They took over carnies. There's great chili dogs on sunset and they built this whole installation. At your depots, essentially, could you have wrapped them as little minions to run around for movie premieres? Does that happen at that depot level? Oh, yeah, we've had a lot of movie premieres actually. And that's exactly where it happens. So our team works with folks to basically create the designs. We actually just announced this week that we're also doing collapse with local artists. So for example, for the pride in West Hollywood, we had robots that are designed by a local artist as good, who is just fantastic. He's been doing this for us for a while. And he makes this really cute kind of, I call them gay bots. Marcia is the name of one of them, who is this very famous figure in the gay rights movements. So we try to make the robots much more part of the community. And of course, brands can also take over and do wraps. If you look back at the history of art installations for vehicles and for all sorts of purposes, you are vehicle, I always go back to the BMW painted cars. We had Warhol. You had a Bosch. Some of those were actually the Peterson Museum in LA. Was that the inspiration for what BMW did with that for what you're doing now with Serf? You know, I like to kind of make it more intellectual, but honestly, brands came to us. I kind of like to joke that I didn't think I'm going to get a PhD in robotics and then sell ads. But here we are. So we had so much interest. And this goes back all the way to when we were part of Postmates. And it was a much more trickier thing because Postmates was a brand itself and they didn't want to take risk. But we had all these brands come into us and saying if they can put their brand on the robot, some of them were daily very partners, some of them were not. And then at one point, we decided to experiment. Interestingly, as soon as we did the first, I believe it was a grocery store, it was a pink dot, which is a very famous brand in LA. Soon as we did that, 7/11 reached out to us. So our partnership with 7/11 came when we basically advertised the competitors. So it's just something that draws a lot of attention. So brands really like them. If you look at ad businesses, listen, I call them an ATM alphabet with Google. It's an ad machine, which is an ATM. And if you look at ad businesses and I don't think that Andy Jasek is enough credit for this, Amazon has a massive, massive multi-billion dollar, shouldn't it be a hundred billion dollar ad businesses inside of Amazon? That's growing. As you look at this from the CEO's perspective of where you're taking surf or boxes of business, how big do you see ads getting overtaken? Have you seen that being one of your main revenue drivers potentially? I do. For the for the delivery robots, I think there is a huge opportunity. If you think about buses as a, you know, most direct example, they make, you know, decent amount of revenue from advertising on buses. Most buses you see have an ad on them. But it's not enough to really fund the cost of a bus and the driver. Now with robots, the fact is it could. It actually, you know, the cost of the hardware is less. The cost, there's no driver involved, you know, so as a result, I'm actually kind of excited about an idea of robots that are largely able to create value in other ways, including advertising. And the delivery fee itself can be really, really marginal. And you basically kind of create free logistical layer for cities, almost free. I think that there's something really powerful about that. So we're going to keep exploring ideas like this. But I definitely see advertising as a big success story. It is a given example. I live in South Florida and I noticed that your robots are all around Brickle and my daughter goes, oh, there's another one. There's another one. There's another one. So you've got the child's imagination and now looking at, okay, is this one brand new, does this one brand new? Because it all ties into your revenue grows in Q12226. The company reported $3 million in revenue. That was a significant over the previous quarter. What's driving that revenue growth? Yeah, look, we have more robots, but we also have more use cases. There's robots in more places. And we've also unlocked new revenue opportunities. So let me cover off them. Obviously, we went from something like 50 robots to 2000 robots last year. Now, they're not all operating all day long. That's growing as we speak. So by Q1, I think we were around 800 or something, if I remember. So that's still a significant growth year over year. And then on top of that, we have new places. Robots are operating new use cases. You mentioned hospitals. We actually did that acquisition in middle of that quarter with Moxie, the Dedy General Body, it's robot joining us basically in about 25 hospitals right now. So that's another opportunity. It's the same technology basically being used in a very similar way, which is moving things, but rather than just moving food in cities, we are now moving medications and supplies for nurses in hospitals who are always very short staffed and really can't benefit from that. So that's another unlock. And then on top of that, we have new revenue opportunities. Like we are monetizing our platform. We've built all these tools and capabilities that can help other companies get their vision to market much faster, rather than building everything from scratch like we had to do. They can just focus on what, as they say, what makes the beer taste better. They can put their focus on really their secret sauce. So we are also monetizing that or our data or insights. So we kind of look at it as we've built something that can have a lot of long tele applications basically and we should make it accessible to folks. And we've just started to scratch the surface there. We have scratched the surface in right way because Q126 in your SEC family reported 45% of your revenue comes from recurring software licenses. Is that allowing third-party companies to use the serve platform? Exactly. Yeah, that's exactly it. We have actually just a couple of weeks ago, we launched a premium version of what we call autonomy assist. It's basically whenever a robot, your autonomy requires help. It needs to connect to a human and ask for input. Well, there's a lot of technology involved in that from really low latency connectivity that you can count on. Because remember, we are in cities. We don't have WiFi coverage. everywhere we go, we don't control the network, we have to just up operate where we are. And if something happens and a robot can't connect, that's a problem. That's a human who has both physically help a robot. And it really doesn't work. It would cause issues for our customers who have to wait long. It would cause issues for the costs of doing business. So we basically had to build a lot of infrastructure just to allow robots to get access to humans where they need to. Now, we opened that up. The very first sign up for that service was a major company that had an application for this. And they got on board immediately to start testing it. So there's a lot of pieces like that. That's just one of them that we've been kind of further along that we are opening up to more folks to use. All simplifying, I would say, it calls server mode operations. And so obviously, your technology works very well with delivery bots. So we start to think about that from potential expansion into remote assistance in warehouses, remote assistance of potential robot tactics. How should we think about it from a form factor standpoint as you build this business? Oh, all of the above. There are tracking companies looking at it right now. Autonomous tracking. There are companies in entirely other spaces I can't name right now, but hopefully we'll have more announcements. We've actually had companies use them in Ukraine in the war for removing minds. So it's a-- being able to connect to that machine in the real world, it's easier said than done. Because once you're out there, there are also some natural connectivity issues. So you want a really, really good robust connectivity layer so that you can operate thousands of machines reliably. And that's the kind of thing we build. That's, again, one example out of many. If you look at Ukraine, the Ukrainian government open-sourced or battlefield data set for companies that are-- they have to go through an approval process, but you can learn, you can transfer from that. Do you ever take any of that-- I'll call extraneous data into it as you're putting it into develop your algorithm for the boss operating in cities? How we not to Ukraine in the other set, that's not as directly applicable to our core application right now. But yes, we do look at the data that's out there. But it's not as much of a source of data for us compared to what we actually have ourselves. We have, again, 2,000 devices right now. They can generate multiple times the number of images that GPD force vision model was trained on. So that's every single day. So there's significant volumes of data coming in. In fact, some of the other pieces that I mentioned in the surf robotic platform that we are enabling and giving access to others is around, how do you gather the data? How do you find interesting stuff in the data and surface that and then run it into simulation scenarios and then feed it back into your models? All of the tooling around that is stuff that we had to build that now we are offering to other folks. That's interesting, because now you're starting to see a lot of buzz and it goes back to Jensen. When Jensen goes on stage and gives a presentation, the world pays attention and I don't comes to black jacket and the world pays attention. And Jensen's been talking a lot about world models. And we've seen companies are developing world models raised hundreds of millions, in some case, billions of dollars. For all practical simple terms, your robots are connecting, collecting world data. Could we potentially see that emerge as another business? So you could have the licensing business, the software business, you have the bots business and then you have the ads business. But could we see that emerging as a new line of business? >> It already is, actually. Yeah, we've had, again, these are really early days. I want to be very clear, like we are providing these, two partners at a very early stage and we'll see if this becomes a repeatable business. But we are out there exploring all that. And that's already something we've actually done. >> One thing I think about from your business, staying in the business line, the thinking is that if there's a large corporate park, multiple buildings, big things, or I'm staying in LA here, sorry, I got to do it. One of our other studios where obviously, the offices are all spread around a lot. Could we ever see, sir, potentially license the technology to a Warner Brothers or to a corporate park where they can just operate it in their own domain? And say, okay, Commissary, he's going to, building a, Commissary going to building B. Could we ever see that emerge? >> Yeah, I mean, that's actually not too far from what happens in hospitals right now. Because, you know, that's not a delivery as a service model where we are kind of running every single delivery. We actually have robot station there all day. And folks can request them to go from A to B and bring them stuff. So, it's actually very similar to that. So, we already are having experience with that. I think when we are, you know, when we look at our technology as a platform, you could pick and choose. You can grab, you know, just the connectivity layer, you can grab just the data infrastructure, you can grab the hardware, or you can get all of the above and basically run it, you know, as a full end-to-end system. So, that really is depending on the use case. It could be something that folks do and it will make sense. Like, we recently launched a laundry verity called it a partner. And that's the same robot. It didn't have to change anything. It's just the same robot moving on the same sidewalks, but it delivers laundry rather than food. - How did that come about? Was that your idea or one of your team members that is, let's come into the laundry business or another thing that they can't be said, "Hey, we're short on delivery drivers. You've got these bots. Can you deliver laundry?" - You know, that's a good question. I don't know who talked to whom first, but I can tell you we've been looking at those adjacent delivery opportunities. Any non-food application is really interesting because generally it's a higher value delivery and it's off peak. You know, you don't need that lunch. You don't need that dinner. It can happen in other parts of the day. So, it's really complimentary and valuable to what we already, the infrastructure we've already built basically can be used to do these things. Now, the distinction is that people don't do three laundry a day, the way they eat three times a day. So, you can't replace food with non-food. You can compliment them together. So, we focused on food, but now we are at a point that it looked 2000 robots out there. What else are people gonna wanna do with them? And we've been kinda exploring that. - Let's have a hypothetical scenario. So, let's say you have QSR, quick service restaurant partner that delivers it. And then you have no scrubs that deliver to laundry. Is that that same bot? Instead, okay, well, I have a downtime of saying, now, or I'm gonna go deliver laundry, or I have a downtime of two hours, I'm gonna go deliver food. Is that how that works? Then the serve platform actors that work is to determine to tell a bot where to go? - In a lot of cases, yes. On the laundry case, I can't answer that right now. There's some other considerations around cleanliness and that sort of thing you have to consider. Like, if you have a garbage collection robot it's not gonna go to delivery your food after, right? So, like that sort of thing, you have to be mindful of, but generally speaking, yes, it's the same infrastructure, same robots. You don't need to design new hard, very you don't need to deploy new robots. It's either in your city and probably just, you know, have a block away from where you are. So, it would go to whatever you ask it to. - And I've noticed in the cities that I've spent time in, your robots become part of the fabric of that city. And part of your Q1 release, you announced to the market that you were going to start an international expansion this year. And then in '27, do a major international expansion. What's driving that international expansion? - We have a lot of interest from other partners in markets across the world from Europe, Asia, Australia, Canada. And we always, you know, look at them in an open mind, especially if we see a certain concentration of demand, shortage of labor, if those pieces kind of come together, good policy frameworks in place, it does make sense for us to want to diversify our geography as well as our partnership. So, at the same time, as I said kind of earlier, our broader vision of building robots that are doing more than just delivering food, they're navigating complex human environments and basically doing whether moving goods or performing services. As you're building that, you want diverse data sets. You want to be in diverse environments. So, we get to go to new places, experience them, collect that kind of data that we can feed into our models and get the robots to be even more robust, even in the existing markets, having that diversification of that actually helps you. So, we're kind of building on that flywheel at the same time as diversifying our partnerships and geography. - Let's take an example here. Two markets that you've publicly disclosed as part of your earnings, Tokyo, in Sydney. When you think about deploying a robot tax, you send there for training, let's you're doing a zero-salt program. How should we think about it when serve announces a new market? Is it okay service expanding to Sydney? Do you send five bots, 10 bots? How long do they train until they start generating revenue? How should we start to think about that? - Yeah, you know, there are a few different pieces of the puzzle. First, you want to identify partners who are we working with, who gives us access to demand. Usually, that could be the same partner that also helps us understand the nuances of this new country that we're going to, whether on the policy making, how the processes work there or actually operation and culture of delivery, et cetera. So usually it starts with these anchor partnerships that help us understand operation and secure demand. There's also the policy side, so we want to make sure that, you know, via ahead of that game, there could be compliance requirements, et cetera. And then you get into the, you know, the, again, the needy of, you know, getting depots, hiring people. And then we usually send a small number of robots to start, you know, putting feet on the ground, collecting data to train the models, actually test how the models perform on day one, and then start dealing with any deficiencies. We do some mapping closer to the launch, but we don't have to have maps to get started or to operate somewhere. But subset of high traffic corridors for us usually would be valuable to have maps because you can move faster. So we do all of the above. And then once we are actually ready, and if you feel like you have ticked all the boxes, you want to start with a small fleet. You don't want to just dump a whole bunch of robots. Like the Scooter model, I don't think works anymore. And it was a good idea to begin with. But especially now that folks have vies and up, I don't think that's the approach to take. You want to put a few robots, give people a chance to get to know the robots and kind of understand them, remove some of their fears, engage in that conversation. And then as we do that, all of this takes time. People are worried about robots and AI taking over. I'm like, you haven't been in the real world yet because this stuff takes times. It's usually measured in months and years to just get off the ground. And then you want to gradually grow that and mature it in the market. So those are all the different steps you have to take. But even once you're done with the basics, you still don't want to rush into dumping a whole bunch of robots in a city. Like going from 40 robots to 500 in LA, it took us some time. It didn't happen overnight. And it was only after we were there for a few years already. - Yeah, in LA it took you three years. When you went from 40 to 500, it took you three years. So we look at the city market and we look to the north, you have Tokyo. And then it looked a little bit to the west. You have Singapore. And then to me, I see a large growth opportunity for your business in the APAC region. And let's just say this is hypothetical. Let's just say you get four, five thousand bots deployed in the APAC region. Should we look for potential manufacturing in that region to one cut down on tariffs, to cut down on shipping imports? I asked that because Hyundai, for example, has a very large facility where they build cars in Singapore. Should we look for localized manufacturing if that's something that you really truly ramp up and expand? - Yeah, I mean, look right now that we do, we have sub assemblies that are done in Asia and then we have final assemblies that happen in the US as you know, by Magna International in Michigan. So we could do end-to-end robot manufacturing in Asia when necessary, when it makes sense. It actually does make sense. If you're going to, for example, let's say we wanna go to Australia, not even in necessary just Asia, but if you wanna go to Australia and you don't have the US tariff, it probably makes sense to just get the robot build outside of the US, rather than shipping it here first. So unfortunately, these tariffs do have consequences like that. And we are quite flexible in how we are approaching that. We can, the nice thing about these robots don't have very complex in terms of hardware. They're not like a car with so many parts and such a complex infrastructure. The simplicity gives us more flexibility. - Yeah, and if you look at your Gen 2 to Gen 3 model, you significantly cut costs on that. You've cut costs, but here's the thing going back to Asia. I have to say it, Gen 2 could run a light rain. Gen 3 can run in heavy rain. So you cut costs, you enable that. How did you do that? - You know, a lot of hard work, a lot of learnings in the field, which I think is like a big part of our story has been, we've always been out there kind of figuring out what has to be fixed and what needs work, rather than sitting in a lap and developing robots with, you know, releasing highlight reels. In our cases, usually low light reels. Anything that goes wrong, everybody sees. But you're actually out there and you're figuring out what goes wrong. So our hardware, the Gen 3 versus Gen 2, you said it. So it goes about twice as fast. It goes about twice as much battery range. It has five times the compute. It has better sensors than before. Better drivetrain. It actually has four-wheel steering and four-wheel suspension, both of which we didn't have before, has bigger cargo. So it's just an incredibly better, more reliable, more robust product that comes at one-third of the cost of Gen 2. So you can literally build three Gen 3s with a cost of one Gen 2. And this was years of work. Again, a couple of years for just that particular model, but also years before that of iterating and learning what works and what doesn't. A lot of what we didn't do was as just as important as what we did do for Gen 3 and I'm very happy with the outcome. And by the way, we are continuing that. We never stop iterating on the hardware. There's always the next minor versions and iterations as well as the next major versions. So this is going to be, it's like iPhone 3 versus iPhone God knows what right now I've lost track. So he's going to keep going. It's great. Yeah, if you go way back, you can say your first Gen was the iPhone with the Edge network. Didn't work very well. Now you've got your 4G underweight, your 5G. Because let's get some stats here to the audience. The new block can go 11 miles an hour in handle heavy rain. It can go 48 mile hour range and 65% cost reduction. So those are the numbers. And I know I've asked you this offline many times. You have this wonderful partnership relationship with Magda. And my opinion, Magda is one of the world's greatest contract manufacturers in history. What role did they play in helping you achieve these manufacturing goals? Yeah, we've been partnered with them for some time. As I said, they do our assemblies in Michigan for us. I think they're a good talk partner. We generally do all the design and IPR selves. It's really critical to us and we are closest to it. But having a partner that actually makes cars, there's a lot of lessons that these guys have learned over the years that we can really internalize and not have to learn from scratch. My goal from the beginning was try to-- there's plenty new for us to figure out. As much as possible, we should not learn anything that other people already have learned. And there are some domains we can really learn from. One of them is the auto industry, because we are also making things on wheels that move around every weather condition or every road pavement condition in the city. And there's operational things like scooter companies that have learned about putting a lot of devices in the city unattended and collecting them, charging them, cleaning them, maintaining them, all that stuff. We don't need to learn all that from scratch. So we build our team with those skills, kind of imported. And we found partners like Magna who can teach us a few things so that we don't have to make mistakes. Looking at things, you mentioned scooters. Obviously, you learn a lot from that. But I have to say-- and this is just my humble opinion-- I've seen some of the most interesting edge cases on X and in TikTok and Instagram that I've ever seen with your bots. How do you handle, I'll call, the interesting curiosity in situations that unfortunately arise all too often when people want to get famous on social media? We've had robots licked. I mean, like literally people going on licking robots. I mean, the kind of things that happen-- it's also kind of in retrospect kind of fun to think about the kind of thing that you would have never expected that you get to experience with the robots. Look, you have to build a certain kind of DNA and culture in the organization to deal with that. Because the reality is that the unexpected things, the unknown unknowns, are going to happen when you create a new product category, a new service, and you put it out in the wild. So we kind of always had that internalized. That's why within one week, really five days of starting the project back in Postmates in 2017, we actually had a robot on the sidewalk. It was very small, like a set of pieces we bought on Amazon and put it together. But we wanted to go out to see what we learned in the real world, rather than again sitting in a lap. So it's all part of that DNA of wanting to be outside to actually experience the world. Because that's the only way you're going to find out what needs to be thought about and addressed. And then the team culture needs to be one that embraces that. That is not afraid of it. That is honest. Then something happens to say, yes, sorry, guys, like we got this wrong. Here is what we're thinking. Here is what we're doing. We're transparent. We're responsible. And then hopefully you end the trust of society as well, so that you can go through this learning experience together. Because there's only one way to get to the destination, which is be out there and kind of learn. Honestly, it makes the biggest difference. If something happens, you say, yeah, it's our fault. People respect that. And they move on. And the kid in me has an idea for some of the interesting use cases I've seen of bad behaving people. You figure out a way. You cut a deal with Warner Brothers. And you make a peppy look you. So when somebody tries to misbehaving, get close to it, you'll let off the skunks smell. Oh, boy, I'm definitely going to touch something again. You know what? That's new. I barely get new ideas, but that's definitely a new one. So props to you. Oh, thanks. And so as you move to Gen 4, Gen 5, Gen 6 hardware, do you see the form factor changing months? Are you pretty comfortable where it is now? And our city's comfortable with that current form factor? I think I love a lot of the work that's being done in robotics right now with like, legged and veiled legged and all the new kind of capabilities that brings to the table. I think there's a reason why we invented veiled so many thousands of years ago, and it's worked pretty well for us. So I would bet that this form has a place. There are going to be other form factors. We've already tested a lot of our technology on four-wheel robots, for example. So we know that we can transfer these when the time comes to a new kind of technology. enable new applications and new ODDs, et cetera. But I do think that this as a form is always going to have its place. And that's why we are investing in making the best version of it possible. But it's no one to say that we won't have a LEGO robot as well or some other form in the future that would work side by side. Maybe they even help each other. Maybe there are things that one can do and the other can't. And therefore, they can actually complement each other. So I think we are going to preserve this while we build more types and more and more forms. You can say that your multiple forms of robots will all be friends in January. You close your diligent robotics, your health care robotics company. From a technical perspective, are you taking some of their stack and putting it to the serve stack, you taking the serve stack, are you merging the stacks together to improve it, or how is that working? Yeah, absolutely. I mean, look, we've built a lot of valuable technology and they have also built things that are relevant to us. So I mentioned that connectivity layer, the assisted autonomy stuff. That's something that they're working on integrating. And it's just such an easy win. But it's not just that. I mentioned the machine learning and AI pipelines. That's something every team has to build. At some point, if you want to go from a 50 or 100 robots to thousands, you're going to have to build that infrastructure. And we just happen to be ahead of most people in that. So they're really like, I don't know, VAMOs and couple of companies, perhaps, that are at the scale. Everybody else has to build this. And one of the reasons why that acquisition was exciting was there was so much synergy there. At the same time, they have capabilities of operating inside indoors, in corridors, in elevators, whether they have an arm that can operate in elevators, can scan key cards and get access to the protected areas. All that is something we didn't do and we didn't need to do. But now we have access to that IP and capability. So it's very much this idea that the two pieces come together and add to each other. Like we kind of make a more robust overall platform that can work inside and outside and elevators and press buttons and doors and all that. There's a trail there, which gets very interesting. And then those robots can operate in strenuous, unpredictable situations. That had to be some value in that data. I think we are operating in unpredictable situations every day. So just having access to data from a whole new environment, which is halvets and indoors and hospitals. And your robots get better because sometimes our robots on a sidewalk are going to a really narrow area with a lot of people around. And that's something that a hospital robot gets to see every single day. So I do believe that again, there is a lot of positive cycles when you bring these pieces together, both in terms of getting access to each other's technology but also creating unified models that can make robots navigate complex human environments. Should we think about healthcare as a growth market for server robotics? Absolutely, yeah, we are very excited. So look, Moxie, in a way, was where we were a couple years ago in terms of the hardware maturity. So there was the previous generation of our hardware. It was more expensive. It had certain areas of improvement, like improving sensing, improving compute, improving battery. All those areas, we are now investing in Moxie and making it better. And the nice thing is we've done this before. We know exactly how to do this. We've done it really, really well. So Moxie and the diligent team have a really capable team but now they also get to access our expertise, our supply chain capabilities, our learnings, our technology. So we are now making those investments and we see a lot of pipelines here for more hospitals. There are 25 hospitals right now but they love their Moxie and there are thousands of hospitals just in the US and it doesn't have to just stay limited to hospitals. There are a lot of other environments that are similar indoor, elevator, doors, you know, all that stuff that Moxie solves. So once you have that, it's really nailed. You can really expand beyond it. - You can expand and you have this great line that closed out your Q1, 2026 investor, Deck. You said several botics is the operating layer of physical AI. - That stood out to me. Your numbers were good, but that was the one line that I took away from your Deck. It says that you have grand ambitions or where you want to take this company. - Thanks, yes, I'm glad that stood out. Look, I think a lot of attention right now, rightfully has been around data centers and models and all that kind of the building blocks. But when you get to the physical AI, to the robotics, there is a whole other aspect of the real world. From models that understand the physical world, not just text and images, but actually, you know, what's happening around them. To, you know, all the, again, the assisting of robots remotely when necessary to data infrastructure, to actually, like all the operation tooling for having thousands of unattended machines in the world, to safety layers, to human operating procedures. Like there's just so much more if you want to have robots at scale commercialized. And I think that's what we really have created. That's so unique and valuable. And you mentioned it at the beginning, how we've grown. I'm like, it's right time and right place with the right technology. It's like the stuff that a lot of people now need to scale their applications and we can actually help them. So it's a, it's a win-win for everybody. They get to go focus on the real value proposition and we get to kind of provide that foundation. - And when you do that, everybody wins because there's growth. And as Jensen says, physical AI is a growth market. I agree with Jensen and there's a great bank of merit report that the AI has left the chat and has become physical, which yes and deed, it has market breaks coin that terms. Look at that to Martin. Ali, as we look to wrap up and we're having you on again sooner, we're not going this year and a half nonsense. Until we have you back next time, I know your public interest, you've got to be careful. What is the future of server robotics? - We've just started. More robots, more places, more use cases, helping more people. And we want to be, we want to enable that. We want to accelerate that. Whether it's our own robots or we help other robots come to life, that's our vision. - I like that. Server robotics is the platform that is bringing robots to life. They are becoming the infrastructure plumbing for physical AI. The future is bright, the future is autonomous, the future is server robotics. Ali, thanks again so much for coming on the road to autonomy. - Thanks for having me. - Billions and soon trillions in value will be created in the autonomy economy. By the time a trend becomes consensus, the alpha is already gone. The gap between uncovering signals and reading headlines is widening fast. When it's a headline, it's no longer a signal. Enter autonomy AI. We decode signals before they move markets, giving you conviction in the autonomy economy. Autonomy AI, your models, our intelligence, visit autnmy.ai.