अनुलिपि 🔗
speaker-0: real world is messy so stuff will definitely 100% go wrong.
speaker-1: the rubber down fully.
speaker-0: Yes, the the robot as well. I think the biggest mistake that a lot of startups make is they go out too broadly too quickly. It works in the lab and everyone thinks ooh yes amazing let's go the easy part about scaling hardware is that you just need to do more of it. It's more or less the same, you just need to replicate. It's an execution game, not a product definition game, which is a little bit more complex, right? If your vision is too bold, if you wanna push too hard, you you lose the people. Then you ask them, Hey, how's it going? And they're like, we're we're reorganizing every six months, right? Somehow in Europe, product management thinking it's very immature, for example, if you compare it to the US.
speaker-1: Looking into the future, what potential bottlenecks
speaker-0: And I'm thinking like
speaker-1: So welcome to a new episode of Robotics Scales, the podcast where I talk to founders and investors that shape the future of robotics. And we're talking about what it takes to scale a robotics company globally. Today I'm super happy I'm joined by Mario Maurer. He's the global business development manager at Robotics at Maxon. Before joining Maxon, he was the CTO of Anybotics. So I'm really happy to have you here today, bringing actually those both perspective on one side from a robotics manufacturer builder, a scale up, but also now from the supplier side from Maxon. Warm welcome.
speaker-0: Thank you very much for having me.
speaker-1: So before we start, a quick shout out to our location partner, the DERC Campus. They are the robotics space here in Zurich. So if you're building robotics company and still looking for an office space, check out the link in the captions. And otherwise, happy to provide more information on the DERC Campus. Now Mario, I'm really happy to have you here today. Today we're going to talk a bit about what it takes to scale a robotics company, what breaks when you're scaling, but we're also going to look into what should robotic startups build versus buy, and also looking a little bit into the future, what could be some potential bottlenecks, assuming that robotics will scale the same way as AI right now. Now before we start, maybe a little bit about yourself. Can you share a bit about your role at Maxon?
speaker-0: Yes, sure. Yeah, thanks. So at Maxon, as you mentioned, I I have two functions essentially. I oversee on the one hand the business development in robotics, meaning I try to make sure that all our sales organizations pitch our products to the right applications, to the right customers. I help with early deal shaping and essentially driving go-to-market of the existing products that we have for various robotic applications. And maybe as a side comment, Robotics at Maxon is pretty broad. It goes from intra-logistics, HEVs, surgical robot to robots in space. So it's really everything in-between, which also makes it hard to figure out what to pitch into which application. On the other hand, since I have a technology background, a developer background, we're also driving a lot of product management. Meaning, together with the team, we're trying to steer our RD organization into the right direction such that. Let's say in five years' time we have the right products available for robotics customers in the market. And this is also very interesting and very challenging because robotics moves very quickly. The trends change, the approaches change, and this has implications to hardware. So we're also trying to anticipate these developments at Maxon to make sure we have the right products in, let's say, five years together with the team. So that's the two functions I do pitching what we have and making sure that we have the right stuff in the future.
speaker-1: Yeah. And I think you already gave us a good segue to talk about the hardware and start with that. Cause what I'm absorbing is that hardware is becoming a commodity maybe faster than we think. What's your take on that? we, do you think hardware is already a commodity or how much is it still a mode for robotic startups?
speaker-0: Yes, it's a good question. And it's at the heart of what makes robotics hard. So if you compare what how the ecosystem looked like ten years ago, it has made large strides towards commoditization, right? Nowadays it's much easier to get a broad variety of actuators off the shelves, of batteries, of 3D cameras, of compute, of machine services for parts. And so on. So the ecosystem in the last 10 years has evolved massively, right? When I started soon, 10 years ago at Antibiotics, we had to do a lot of it ourselves in the end, right? But nowadays, the whole supply chain, the ecosystem has evolved massively, driven by smartphones, by e-mobility, by autonomous driving. That all these components are so much more approachable. Still, it's If you build a robot, you can only use, let's say, off the shelf commoditized hardware to a certain extent. Yes, it helps to spin something up quickly, but the devil is still in the details, right? At some point you need to optimize your hardware product, you need to dive deep, maybe you need to start customizing certain things. And then you go away from the commoditization. Yes, it's still accessible hardware, but if you need to really scale a complex robotics, Product, you still have to really understand how all these nitty-gritty details and subsystems behave, right? It goes into details like what cabling architecture is the right thing, how do you have to change the electronics of your battery management system? just to give some examples. So striking this balance is still very hard despite the good access to commoditized hardware for robotics, I would say.
speaker-1: Yeah, and speed also plays a tremendous role. One of the reasons why we see also the Chinese robotics companies accelerating because you have this high
Load full transcript
density of the supply chain does also speed. What do you see with here in the Swiss system? Is that something that's rather than holding the companies back? Or do you think that's also the opportunity to outsource certain things and buy them off the shelves that also give like some extra speed?
speaker-0: Yes, I think it it depends on on the stage where you are as a company. So yes, the Chinese ecosystem, and it's really an ecosystem, is super quick, right? They have everything you need closely together in terms of suppliers, in terms of competencies. and this allows them to be super fast, right? But this also I would say in large part applies to a global audience. So you have also access To this ecosystem. You can buy a lot of components from China. They ship globally, right? Because it's a very export-oriented ecosystem as well. And you can leverage that as well. The key question is that you always have to be, and it's not just a question about China, not China. It's always a question when you grow a company about where do you place your supply chains, right? Where are your supply chain risks? How you do multiple sourcings. China is an important puzzle piece in there, but it's not the only puzzle piece. so yeah, in short, if you want to be quick, yes, you can leverage fast ecosystems. Why shouldn't you? And as you mature, on the other hand, you have to be sure that you select the right suppliers, right? Whether they are in China or not, does it depend on the type of technology that you need.
speaker-1: Yeah, and we talk about that like the supplier perspective a bit later, of course, also dependencies that come with it. But now if we look at cost and scale, assuming cost and scale are winning, where do you see them to the advantage for robotics companies?
speaker-0: That's an interesting question. And the question is what does it mean cost and scale? I think that's it's maybe two different discussions, right? So scaling for robotics company means that you as a company can produce hardware quickly and you can grow the amount of hardware that you produce and that you ship, right? Without growing the invests behind it. Linearly. So you grow the output faster than you grow the people that you need to drive this output, essentially, right? This this for me is scalability. And then you you also have low complexity in your design. So you can you can churn out hardware quickly. That's on the cost side or sorry, on the on the scalability side of things. Cost is always a discussion, obviously for a company and for a supply chain, but there again it depends. into which application you go as a robotics company, right? A surgical robotics company has an entirely different cost structure than a B2C robotics company. And and there is maybe two aspects to it, right? In the if you go really high volume, for example, automotive and robotics in large parts or in some parts goes into that direction. Then it's really about why does this shaped piece of metal cost more than the raw material that I dig out of the ground? And you really margins come compressed down to singlet percentages, right? And the question is maybe with this question from the Chinese ecosystem, right? Yes, today China is is cost competitive, absolutely. But I think in future automation in manufacturing will also make the US or Western European ecosystems and supplier bases much, much more cost competitive, driven by automation in manufacturing. we see that a lot as well. I think this is gonna become an important driver in the next decade. Not tomorrow, not in the next year, but in a decade or so. Hope I answered the question more.
speaker-1: I know it was a tricky one, but it's an interesting point you're mentioning. So you're saying Europe's investment into automation, partially being then also robotics, is then decreasing the hardware costs that it takes Europe to build their own robotics.
speaker-0: Yes. I think what we're seeing today in terms of fast iteration cycles in software, for example, large language model agents being able to quickly iterate on math problems, on engineering problems. And we're only still at the beginning of this, right? But it's incredibly how fast this goes. This speed will somewhat translate also to manufacturing, to producing hardware globally in Europe, in United States, in China. But I think what I want to say is the companies that can iterate also quickly on hardware, that can produce
speaker-1: .
speaker-0: complex hardware with a high degree of automation, those companies will have an advantage because it will follow the speed that the the development ecosystem currently paces out. So it's it's about speed in hardware in manufacturing going forward, that will be a key driver, I think.
speaker-1: Now if we move a bit from manufacturing to actual deployment and also now looking a bit into your experience with Anybotics but also now the robotics companies you work in, what's usually the first thing that is breaking when you move from like a prototype or let's say the first 10 robots you deploy into like large-scale deployment? say like thousands of robots or more.
speaker-0: Yes, everything. I think
speaker-1: Not the robot, hopefully.
speaker-0: Yes, the the robot as well. So I think it depends. So yes, stuff will definitely go sideways. What you see in the lab is only a small projection of reality, especially as you scale up, especially as you start deploying more and more robots, right? You will discover many more edge cases in the real world, and the real world is messy. So stuff will definitely a hundred percent go wrong. It'll it also goes wrong for us as a manufacturer of hardware, even though we did it a hundred times already, right? There is still stuff that's That trips you up because it hardware is complex and the interplay between hardware subsystems, especially in robotics, is just devilishly hard. But I think where you can mitigate these risks or these shit happens, the things that will
happen is by smartly setting up your deployments, right? By being very cautious about How you go to market, where you deploy first, and especially don't deploy too quickly, right? It's really about smart product management, that you're conscious about what really your system can do and what it cannot, and select the applications, the geographies, the customers, really, depending on that, that you are maybe over conservative at the beginning, start learning slowly. I think the biggest mistake that a lot of startups make is they go out. Too broadly too quickly. It works in the lab and everyone thinks, ooh, yes, amazing, let's go. And it's also a bit natural, right? Investors demand a certain amount of growth, so there is a tendency to to throw it out there and show that revenue coming that the robots deployed grow. This is a super hard balance to strike from a yeah, from a risk perspective.
speaker-1: Yeah, I think you mentioned two really interesting points. think one is the timing and the second one is a use case, right? To identify what's maybe the first niche because each use case then has a different environment. You need to train on different cases. And that's what I often see is that robotics, mean, and in this technology that you can use for so many different use cases, often different environments, From logistics warehouses, in a manufacturing side. So really focusing or identifying where do you have the best product market at least in beginning to focus on. What I want to pick out of is also the timing because I'd be interested as a former CTO when would you decide what would be your criteria that the robot is now ready for the deployment.
speaker-0: Yeah, it's a hard question. And I think it also boils down to selecting the right customer, the right partner, especially in early stages, right? It's at the beginning of a company, it's all about learning, learning from the market, as you say, from the application. And you will you will most likely do several pivots, right? You will realize, ooh, okay, that's actually not working. Let's try something slightly different. And all these small nuances, it boils down to To learning. And in order to learn, yes, you need customers, you need the right customers. So it's really about finding the customer that is willing to accept these risks, that has a vision as well, right? That that does not necessarily see the product today, but they see the path for a stable towards a stable product. And that they are open in sharing learnings with you. It's a they are investing you as a company, right? By assigning their time to your learnings. And finding these partners is crucial from the very beginning, right? You can yes, of course, you need to have a somewhat stable system that you can show it to someone that's that that it needs to be a little mature enough that someone says, okay, let's let's try this one out in in our application. And it's all about that. And then really focusing on the right people. Maybe also be courageous enough to say no to certain customers if they turn out to not be the right fit. There is a certain learning curve also behind that for sure.
speaker-1: Yeah. Yeah. First of all, building this confidence that they believe in you and then trust you that become a sparing partner for the future. And what I often see is that there is a difference between an ideal customer, the ICP, and then usually the early customers. So usually maybe your first customer might not be your end up your ideal customer, but it was the one that helps you to bring the product to certain maturity, then start scaling it or go into a different direction. Right.
speaker-0: Yes. And I think maybe extending to what I said before, a hard topic of this is also especially in robotics and it depends a little bit on how complex your robot is, right? But you quickly need a team of ten to twenty people to get even the very basic first product prototype out the door and have a simple enough product, right? So the investments that you need even for very early stage deployments have a sufficiently robust thing is can be quite substantial. And that is also a tricky balance to to strike, right? How much do I invest really? When is the product ripe enough? All those questions. It's super hard questions and again it depends a lot on where in robotics you are.
speaker-1: Yeah. Now when we look at these different areas, right, so with the hardware, software, infrastructure, people, what was, what do you think was the like the toughest part to scale?
speaker-0: I would say the experience behind it all, the the product experience behind it all. It sounds, it sounds maybe a bit silly, right? But I think the easy part about scaling hardware is that you just need to do more of it. It's more or less the same, you just need to replicate. It's an execution game. It's not a product definition game, which is a little bit more complex. So scaling hardware, yes, it sucks because you're gonna have all these issues with the hardware, wrong connectors, wrong batteries, whatever it's gonna be, right? But at least it's it's more or less the same thing, plus minus, right? Scaling people is is also okay in terms of a challenge, I would say. maybe scaling culture behind the people is is a tricky aspect because you have to you have to balance the ambition, the vision of the company and the product with The view and the capabilities, the competencies of your organization, right? If your vision is too bold, if you want to push too hard, you lose the people. And so this culture behind execution and making sure that people understand the mission. And this changes a lot, even as you scale as a company, as you scale from 20 to 60 people, you already start seeing aspects of this. Software, I think, is more and more getting Commoditized and at least compared to 10 years ago, it's much easier to nowadays rewrite or revamp software architecture. So I think that has gotten a lot easier as well. And
maybe one more aspect for scaling is partnerships. And it goes a bit to what you said in the beginning: make versus buy, right? At what point do you switch your partnership, your supply chain model? Because that also might have underlying architecture implications on your design. So, yes, in in short, maybe, or yeah, maybe one more aspect is when you scale at maybe you have to change the architecture of the product, be it the software architecture, the hardware architecture, the team architecture, the organizational design. And these architectural changes are always inflection points where you really have to think about how do you change now? And this has implications into software, into supply chain, to everything. So these inflection points are. Probably the challenging part when you scale.
speaker-1: Can you share maybe one of those inflection points or maybe, yeah, one hard decision you had to take?
speaker-0: So at some point in your company you need additional organizational layers, right? And we hear that from at Maxon, we hear that from companies as well, young startups, right, that have grown massively. Then you ask them, hey, how's it going? They're like, we're we're reorganizing every six months, right? So it's a challenge and it goes to the culture comment I made before to keep the teams aligned while changing everything all the time, right? And people hate change, but still as a company, when you're scaling, you have to think about the architecture of your company. So there I I saw it right when you were s when you were restructuring teams. Say, okay, guys, what we're doing now it just doesn't work anymore. Let's restructure the team, bring in a different team lead, reorganize the competencies, the mission of the team. Those are always very challenging because it's uncertain, there are discussions going back and forth, everyone is nervous about where they're gonna end up, how it's gonna do, everyone has opinions. So that's the exciting part.
speaker-1: Yeah, and I think that's probably also one of the toughest part, changing teams, because what worked for selling and deploying maybe 100 robots doesn't necessarily work when you grow.
speaker-0: If you're deploying a hundred robots, you're already far in the game, I would say.
speaker-1: It's good range, yeah. But I think just generally what worked maybe to a certain, let's say to 1 million ARR doesn't work, then to reach the next 10 million or 100 million.
speaker-0: Yes. And at some point you also have to to switch from an R and D organization into a sales and operations organization, right? Especially young startups, they usually they start with with some piece of IP intellectual property, right? It's an idea, it's a way of doing things differently, but it's very especially in robotics, technology driven, right? So new technologies enable new approaches to things, whether it's a product or an execution. But over time when you mature this idea. Execution gets more and more important,
right? You have to make sure that you get the things out of the door, especially in robotics, it's it's a product. You have to make sure that the product is sharp and that you can also sell the product, that the customers are understanding how the product works. So it's a shift away from crazy ideas, technology-driven approaches, university spin-outs, into an organization that executes on this product, right? And this means means much more sales, much more partnerships. Much more contractual and legal topics, go to market think it's a completely different thinking that you have to bootstrap into your organization, into your culture. And that's also one of these inflection points, right? That you that you always have to think about consciously.
speaker-1: Yeah, it feels first like you're proving the technology and then you need to prove the product and eventually the business, company.
speaker-0: Yes. And I think one of the very hard things in robotics is good product management. Is really having and in the beginning of a robotics company it's very often founder driven, right? The founders they have a feeling of what the product should look like, how what it should do. But over time you need to institutionalize this product management thinking. You need to start having people that Bring this to a roadmap, that execute on this, that structure ideas, bring ideas, bring the learnings, and especially decide on what the product should be and especially should not be. And also when you prioritize those, the product management has to prioritize. And that means also deprioritize certain things. And this is always very hard because the startup goes from a founder driven product organization into an organization where product management is more and more institutionalized. But I think this is always a very crucial transition because it helps to make sure that the product stays relevant and that the product changes. And yes, there might be changes needed, but enacting these changes through structural, organizational, smart product management design is very, very hard. Because I think it's very hard to find good product managers. And somehow in Europe, product management thinking in our culture, in the way We have done companies and thought about this. It's very immature, for example, if you compare it to the US. Somehow companies in the US have a much more mature product thinking. They really think about what the product is and what it is not, and the hardware behind it is just a means to an end. And in Europe, this I I always struggled in my experience, and even today still to find good product managers that have this breadth that understand the technology but also the business case. But I think for a startup this is crucial to have a conscious product management mindset.
speaker-1: Yep. And I would say for product management in robotics, it's different than ZAS because you have all these different components, right? You have a robot yes, you have the hardware part, your whole navigation, but then you have probably an interface for your customers. How do they interact with it? Then all the integration parts, how is this integrating to the other tech stack or fleet of the companies having so many components that make it quite complex. And then you also mentioned the industry knowledge, things I can only double down on. They need to understand the industry. and especially where it's heading. Do understand where are we going? Are we going to autonomous warehouses here and how would our product function in this ecosystem?
speaker-0: Yes, absolutely. And it's it's astonishing, right? Seeing this journey even for for very fast moving robotics companies that that we see, it still takes them like three years to to scale, right? And and even if they have a very sharp product understanding, we're still talking years in hardware iteration cycles. And the investment is massive. The you cannot quickly change hardware. Yes, you can to a certain extent. But still it takes much, much, much more longer than for software.
speaker-1: Yeah. So next we want to come to a bit more of a controversial topic, right? You mentioned before partnerships are important and the big question is what do you make yourself and what do you buy? And here we see really completely different approaches. I recently see a lot of startups developing their own parts, right? Their own... Accenturators, own like hands, they even start collecting data themselves. So how do you as a young robotics startup, how do you pick the right approach?
speaker-0: Yes, it's it's make versus buy in hardware, right? And partnershipping and partnershipping decisions. And those are always very, very hard decisions, even for mature companies. And I think the the recipe behind it is something like you have to be conscious about where the value lies that you bring to the table. And it can be perfectly okay to spin your own hardware. if you need it because if it's linked to the competency or to this intellectual property or that thing that makes you unique as a startup, if it has a direct link to hardware, it can be perfectly fine to do hardware yourselves. Now it depends a lot on what what you mean by hardware. And there the devil is also in the details. So for example, now speaking from my perspective, right, we if if you think you can do a massively different electric motor, or if you can do Different gear type or actuator architecture, this is a very hard problem because these things have been around for 100 years and the trade-offs are more or less known. Maybe not to young people, but if we have a lot of people coming to us pitching ideas, hey, I have this idea of a crazy new actuator topology. And then we look into it, we we have a partnership discussion, but we quickly realize: yeah, look, it's it's a known thing already, and here, here, here are the trade-offs, right? So Innovation or IP on a low, let's say low level hardware stack is very, very hard
because the basic building blocks of electric motors and gears, robotic actuators, now speaking from my perspective, have been around for a long time and it's it's more or less known. What can change the game is manufacturing technologies, material science technologies, where you might have a fundamental advantage in cost, for example, or in speed, how you build certain things. But even there, going back to commoditization, you really have to be Conscious, is it worth the investments and is the value add that you bring sufficient for, let's say, a component like an electric motor? You could, if you if you are a little bit controversial, you could say, who cares about the electric motor, right? As long as the actuator or the robot in itself brings sufficient performance. You could argue that it doesn't matter if the electric motor is 20% more or less efficient or power dense, as long as the robot still works, right? So these discussions. yeah, you you have to be conscious about where your IP lies, right? If your IP or in your intellectual property is in a system or in a product that lives in a more complex context where software plays a role, the interplay between software and hardware, there the the the the buy, the partnership discussion becomes much more relevant, right? The pitch I make, for example, to our customers is Maxon is good with hardware development and with manufacturing, right? We want that our customers, the robotics customers, can focus on building their product, their robot, the means to the end of their value proposition to their customers, right? They shouldn't worry about all these nitty-gritty details in actuators, right? Gear lubrication and manufacturing, scaling and all these very complicated topics. We claim or we hope that we can offer this. To our customers, such that they can focus on their complexities, which in robotics is already enough, and we can help them executing on the hardware. But again, this is this is still very hard to get into because the our customers they have to trust us that we we really think in their best interest that we are not too slow, right? As a startup, you want to be fast as well. And and yes, Maxon or large partners can be much slower than startups. But on the other hand, they can be much better in scaling up hardware or manufacturing because they have a lot of experience already. So, yes, it's it's very hard decisions and discussions. I think what you can do as a startup is keep this path for scalability in mind, no matter who the hardware partner is, right? Have regular touch points and think about when a good inflection point would be as a company to. start something with a partner. And maybe there are ways to start something while at the same time also being fast. For example, no no customizations or just taking stuff off the shelf and then knowing that you can evolve into something more elaborate. But it's it's very hard discussions always. And we see this with early stage startups,
but as well as large enterprises as customers.
speaker-1: Now, Maxon is a Swiss company. Let's dive a little bit into why should they work with you versus maybe some, let's say, Chinese competition that says, hey, we can do the same, but at different scale, at different like price. Why would their companies usually choose you?
speaker-0: Yes. So Maxon is headquartered in Switzerland, yes. We have about three thousand employees, but we are a very much a global company. So we have production in Germany, Hungary, South Korea, Thailand, and sales organization in every major jurisdiction. so we think very much globally. Yes, our R and D and engineering is in a high cost country, in Switzerland or in Europe, but still the we have the flexibility for our customers to Build them something that is competitive, I would argue. And even more as we look forward into the future, automation is becoming more and more important. This has implications on costs on a large basis. What we can do as Maxon, and yes, others can do that as well, but again, our pitch is we are deeply vertically integrated, and this allows us to take complexity out of the systems. And when you have a low complexity system, then you can also scale it. Much easier, and you have attractive costs from the get-go. So we do have customers where we produce a million units a year, complex microtronic systems, cost competitively into even automotive sectors. So that's perfectly achievable. It's not an easy path to get there. But we have this experience and this deep vertical integration that allows us really to take complexity out, to remove things in a product, and to do this together with customers. Yeah, I think our Swiss heritage boils down to trustworthiness somewhat, right? We are a an approachable, trustworthy partner. Yes, others can do that as well. yeah, we have six years of experience, others have similar experiences and ecosystems as well. If you have a good product and a good company, there is always competition around it. That's that's perfectly okay. but I think we can be proud of our customer base and the achievements that we are doing. having as a company and as well in robotics. And obviously we want to have more of this, right? We see robotics as an important global mega trend. And yeah, electric motors play a very, very large part in this. So that's it's our mission to accelerate robotics on a global scale.
speaker-1: Yeah. Now you started out as a CTO at a startup at Anybotics. Now you're on the supplier side. What's something you changed your mind about since you're on the other side now?
speaker-0: Well, I didn't start as a CTO at antibiotics, right? I started as a as a developer and then took over more and more responsibility and and yeah, grew with the company with antibiotics from like twenty to two hundred people. At Maxon, I it's f it's my first corporate job, if you will, right? But I'm lucky that it's still a fast moving entity inside Maxon where we where robotics we can help push robotics forwards. what changed my mind or My key learning working in a global company is maybe that what we discussed before is very relevant for global companies as well, this product mindset, right? Especially if you are in a fast-moving market and you have to decide where do you invest your RD resources. There's opportunity costs behind this, right? How do you now figure out which products to develop, which customers to prioritize? Because we We are also investing our time and resources into customers, right? And sometimes we have to decide which approaches to prioritize. So this product management mindset is not a given, I would say, in companies. Some companies are it very it's non-existing and they're simply lucky that they that they are successful. And with some companies it's very strong. And that's driving this forward is is very interesting to see. And of course, you have you You have all the operational it's very cool to see the whole operations inside a company like Maxon, right? We have high makes low volume is a large part of our operations. And if you order something in the web shop, for example, it immediately creates in our ERP a shop order that lands fully automatically in the shop and then gets produced. And seeing this whole machinery running at scale is super interesting. We produce about five million actuators, motors, gear combinations a year. And most of this is just processed automatically. So that's a cool learning, how you scale a hardware company like Maxon.
speaker-1: Yeah, I'd be also interested. You said you get so many different requests, you serve so many different industries, you said sometimes it's tough to decide what probably areas in robotics you should also focus on with Maxon. How do you decide? What are the things you're looking into when you think about for which segment use case should you now invest into new products?
speaker-0: It's great question, yes. So on the one hand, we we're always close with our customers. We make sure that the customers or we try to make sure or help the customer that they they follow the right approach in terms of also their maturity, right? Sometimes the customer has to prove themselves to us in order for us to also invest resources and time into their problem, right? It's always a tricky balance. So that's on the customer on the resources side of things. Where yes, resources are limited, but we have to execute on a lot of different projects. And if you have an agile mindset, you can you can always find a solution, right? You can put the puzzle pieces, the staging of the different problems that you're tackling into the right order, such that in the end everyone is happy. Critically, you could also say that everyone is equally unhappy, but that's that's how you look at it, right? I would tend to think that we can usually achieve a good balance, but it's not always easy. And then on the other hand, you have the technology, right? You have to do greenfield
research sometimes. You have to decide, and sometimes you have to even in a mature company like Maxon, we are trying out bold new approaches where we're like, hey, should we try this? Let's just build it and see what happens. Let's build it and see if it's a good idea. Of course, it always has a link to the market, right? We're not just doing willy nilly things. We are always thinking about could this technology building block? create value for future customer problems. And yes, this can be guesswork. It's a lot of exposing yourself to thoughts in industry. So we're attending a lot of conferences, we we have an open mind. But still we have to decide and we have to and sometimes you have to take some risks, right? And there again, a good product management culture, a good innovation culture, an operating system behind innovation and product management. where we have roadmap reviews, where we have a diverse set of stakeholders deciding and looking at a problem, that helps a lot.
speaker-1: Yeah. Now, when we look into the news and the media, right, a lot of now stories about humanoids. What area do you think is most underrated? Like is there any area of thinking, hey, like applying robots here right now, huge potential, but it's not that obvious. It's underrated.
speaker-0: Yes, I would maybe not say. Underrated, but what we're seeing in intralogistics is interesting, right? So intro-logistics means warehouse automation, automatic storage and retrieval systems, stuff like that. Imagine you have a shuttle, an HEV or an AMR robot driving around in a warehouse bringing boxes from A to B. The trend we're seeing there is that this gets this fuses more and more together with mobile autonomous robots in unstructured environments. So these HEVs start having arms, they they don't follow a strict path anymore. Interact with unstructured environments, box picking is an example, right? Where you start seeing modern robotics approaches inspired by what's going on in humanoids and with dexterous manipulation being applied to solving real logistics problems at scale. So that's I wouldn't say it's underrated, but it's people are maybe not aware that this is happening, right? otherwise underrated. I mean, there's gonna be massive, massive. Massive things coming, right? I think a trend that we will see more and more is and I think we we see this already is robots in public spaces, right? If you go to an airport right now, there are robots, cleaning robots driving around, delivery robots, and it's like, ooh, okay, robot. Even in our hallways, right, in production, there are robots shuttling boxes from A to B. And this will slowly get more and more. Suddenly we'll have robots delivering our packages to our front doors, right? And it It will just appear overnight, as maybe autonomous cars will just appear overnight. Suddenly you have a company like Waymo deploying cars in in Zurich or in Switzerland and here they are. And this is also gonna happen in robotics, I think,
over time. Safety will get addressed for sure. This is a very solvable problem. So we will experience a lot more robots in the next decade. And it a lot of them are suddenly just gonna pop up. Yes, you won't have a humanoid at home yet, but maybe your vacuum cleaning robot can suddenly easily climb stairs. And then as a next step, yes, it puts away the toys, or maybe then starts cleaning out the dishwasher, stuff like that. It it'll come in steps, but those steps I think will be surprisingly quick. Now, this is in terms of autonomous robots, right? And yeah, I mean the trends we're seeing are the right ones. The dexterous manipulation is here to stay. Will it be a 30-doveff, 20-dOF hand? It depends on the application. Some yes, some no. Will it be a humanoid form factor? Some yes, some no. It'll diffuse into very different form factors, but for sure it's it's it's here to stay.
speaker-1: Yeah, now let's zoom out a little bit and look a bit more into the future right now, assuming that robotics will develop the same way as AI right now, especially with the LLMs. Suddenly with the LLMs, saw bottlenecks like energy is a huge one, GPUs, data centers. Looking into the future, what potential bottlenecks do you see for robotics that we don't see right now?
speaker-0: Yes, you're talking about scaling a diverse set of robots up into very high volumes, right? What will happen is essentially the question. Ooh. And I'm thinking like ten ish years ahead now. I think one huge challenge will be accelerating all the supply chains behind it. For example, data centers today, yes, we see it already there. Compute is limited, right? Everyone is screaming for compute, is screaming for data centers. And by the way, it's also super interesting to see limitations already there, right? For example, cooling systems for data centers are a limitation. Transformers for data centers are a limitation because if you build a gigawatt-scale data center, you need power supplies for it. And those transformer companies are not ready to put all the copper into it that is needed. So you also see innovative approaches like solid-state transformers or new cooling system approaches on that end. And I think the same happens in robotics. What I want to say is Accelerating all these supply chains and the vision of a guy like Elon Musk that there will be millions, hundreds of millions of robots is not wrong, but creating the supply chains and you need very diverse supply chains: compute sensors, batteries, safety sensors, actuators, gears, electric motors, tactile sensors, and so forth, cameras. You really And this is automotive scale complexity, right? It took automotive companies a long time to produce 20 million cars a year. Robotics will also get there, but bringing this entire supply chain up to speed will take some time, I think. And I think that it could be a bottleneck, as you say, yes, because the development of such robots will accelerate. What we're seeing with, for example, with Austra today is is very impressive and it's also only the very beginning. So Austra can already lay out PCBs, right? A lot of hardware design will get commoditized by modern software. So we can develop hardware faster. But the bottleneck will be production and deployment of this hardware thing, yes.
speaker-1: Yeah, I think that's where we can learn so much from the automotive industry, Also talking to a of suppliers that historically focus heavily on automotive and now they're redirecting to the robotics, which I think totally makes sense. And I think also robotics, need those learnings from automotive.
speaker-0: Yes, absolutely. Yes, yes. But it's very exciting and this stuff I'm convinced will happen in one form or another, whether it takes ten or twenty years is to be seen. But as I said, it's it's here to stay these technologies.
speaker-1: Yeah. Well, thank you so much, Mario. We're already at the end of it. Thanks for this super interesting discussion from the tough decisions to take, the inflection moments, especially around, yeah, when to buy and what to make and already looking a little bit into the future. So that was already it of today's episode. Again, you'll find more information in the show notes and then until next time. Thank you very much. Thank you.