Why the Last Mile Breaks Most Robotics Startups, with Roland Siegwart, Professor at ETH Zurich

these people have to understand that it's a long, long way. Typically, you say that from a proof of concept, you need additional 10 times more energy resources to get to something which is really, can be on the market. And this is typically the, we can also say, if you have a product which is 90 % ready, the last 10 % is at least as long and very often more. Robotics is having its big moment. Capital is flowing, humanoids are everywhere, and Europe is producing more deep tech startups than ever. But according to the latest reports and to people actually building robotics startups, something still isn't working. In this episode, I'm talking to Roland Seigward, professor of autonomous systems at ETH Zurich and one of the key figures behind Europe's robotics spin out ecosystem. We talk about why Switzerland succeeds where other European countries don't, why most robotic startups underestimate the jump from lab to market and what really determines whether companies scale or stall. We also get into capital strategy for robotics, how to choose a first market that actually pays, and we talk about when the actual moment comes that founders need to bring in experienced operators. We also touch on a topic that's rarely discussed openly, the growing overlap between robotics and dual use technologies. or deep tech.

and how founders should think about responsibility as these lines blur. As always, we ask Roland additional questions offline that you find in this week's newsletter under followthegradient.io. But now let's get into the conversation with Roland. Great to have you here, Roland. You've worked with robotics founders for decades through technological shifts, funding cycles and generational changes. Looking back, what do you understand about founders today that you didn't fully grasp maybe 15 or 20 years ago? Has something fundamentally changed in how people build companies today? Of course, what I learned by following up with a lot of these startups is that it's a long, very hard challenge and journey. And especially this in the field of deep tech, where you typically don't have enough income after five years. I think this is a learning which all the startups have also to do. There is probably a change that there is much more VC and... capital around there, especially in Switzerland for the early phase, the seed rounds, there is a lot of support, which on one side is nice. On the other side, it makes probably people also a little bit more too comfortable in the beginning. And then they all of a sudden realize at one point, they don't need only rewards, they need clients which are paying for the products. So it's really the speed and the sense of urgency that get distracted. So you talk about Switzerland and in preparation of our talk I dived a little bit deeper into a few tech reports and one was the Swiss Deep Tech Report 2025 and it shows that 60 % of all VC in Switzerland now goes into Deep Tech, the highest chair globally ahead of Israel and even the US. So that's remarkable for such a small country from inside ETH.

What culture or structural mechanisms do you see enabling this consistency? I think that, of course, ETH itself is a deep tech place. And we had a couple of structural changes, meaning that we try to support startups even before they are founded early on with some seed financing, which are still money within ETH, like the Pioneer Fellowships, where I was involved, I was initiating this. There is a lot of other elements also outside ETH which support this. This is one important element which was built up in last 15-20 years. And then there is of course also support from outside like Venturekick which support all this journey. And this makes that at one point you have much more role model and all of a sudden you have this self-running concept that young people actually early on think this is also probably a way to go in the future. When I was a student, I was actually only thinking about, yes, is it ABB, Sulzer or whatever? And today I think students think about, is it probably my own startup? And this makes a change and generates a new ecosystem, which was really nice to see. And it's exactly that ecosystem that probably helps and just that this could be a path for young students. And it also shows like, mean, ETH and EPFL are in the top five institution in Europe for spin out value creation. And ETH is number one in robotics. Is there one thing where you say this really says the why behind it? Like which decisions or environmental factors?

shape this outcome that we have so many spin outs? Is it also this ecosystem or is there something else that other universities listening to that could actually copy? Of course, this is in principle also support, but it's in the end also that you have a critical mass and you have professors and mainly professors which are also interesting to support young people to go this journey. And in case of robotics, looking back, I think the most important thing I did early on when I started at ETH 3 in 2006, a couple of years later, we had a specialized program, a master program in robotics. And these allowed us to attract outstanding students from within ETH, from within Europe, but also abroad. And this brings the talents to the place. And happily, we had also a lot of professors which not only think about deep dive research, but also want at one point to see their results in the society, really making added value. And this combination, I think, brought us where we are today. And we can proudly say that we are doing extremely well in this field of robotics. And for your own personal story, because you haven't been a founder, so to say, before you became a professor. Now you're involved in many, many startups. But was this also that you want to see your research turned into an actual product? Or what was it then that made you so entrepreneurial? In principle, after my studies, I never thought about making a PhD and never thought about doing research. I was actually always thinking about becoming an entrepreneur. So I was already starting to look around, but then I had a wonderful opportunity for a PhD. And I actually was involved with the outcome of my PhD was the basic foundation for a startup, which I was involved 89.

So this was my first startup, which I was involved in. And then later on as a professor, I had more opportunities to support people to go into the and also kind of live a little bit of the founder life. So to say as a professor. Yes, of course. Sometimes I think it's probably a little bit easier because you still have university, but it's still challenging and it's a lot of fun. A salary.

Yeah. Now we talk now a lot about also the good sides, so to say, but the European Deep Tech report in 2025 points out that Europe still faces a scale up bottleneck. More than 50 % of European Deep Tech growth capital comes from outside Europe. Now, when you look at ETH spin-outs like Anybot, Dex or 7Sense, what do you think Europe needs to unlock to consistently scale up Deep Tech companies into global players and not just invest early on? As you said, this is very, very strong. Of course, it needs money and it probably needs role models and people which actually did this once and can help the next generation of startups. And this, I think, will evolve step by step. But probably there is a difference also, for example, compared with Silicon Valley. We are speaking of deep tech startups, which are typically entire systems. which is much more complex than bringing up a company like Amazon, which is mainly a software platform, bringing up Google like a software platform. And this we should not underestimate. So we are probably not doing so badly, but we are mainly focusing on stuff which takes much longer to get it running. On the other side, think from scaling, we should really try to get more funding, which typically then goes for... out of Europe to Silicon Valley to stay, to keep it in Europe. But we have another story to tell, which is a more difficult story because we don't have the examples. How many examples are there with companies which are doing entire hardware system, which were successfully and are now mega companies? There is probably Tesla, one of these examples, Apple, but Apple took a long time. And so there is really a long runway. for this type of company.

But so you're positive if you look into the future. I'm positive and I think this is also the strength of Europe. I typically say Silicon Valley can do software, Asia can do hardware and we can do both. Yeah, that's a great sentence. When we come back to the students, you are also surrounded by one challenge that founders repeatedly mention is how capital incentive robotics is long R &D cycles, complex hardware and slow sales cycles. You also mentioned it. Now, from your experience across many spin outs, how should robotics founder think about their funding strategy, like especially the balance between speed, early dilution and the realities of hardware development because it takes very long. So how do you actually structure that to then become this big global players? Yes, this is probably really the biggest challenge. And I think you have to find the right balance. It depends a little bit on which field. In some fields, you can actually have early wins and where you probably it's not the extremely what you have were dreaming of, but there is applications. And I think it's important to really try to have these early wins to make all of the couple million turnover and then build up the next step. I think this is much easier.

for a story where you know that it takes a long term. You can sell visions. This is currently done in US with the human robots. think everybody which knows the field knows that this will only be really giving potential returns in 20 years from now. Some of these companies will probably have returns in a more restricted area, which is absolutely feasible there. But this is to see. And in Europe, think the mentality is not so easy to have people which really pour money, millions or billions into startup companies by just thinking about the dream that at one point we have human robots or robots which can do everything. So what do we do? Do we all go for an exchange in Silicon Valley to see how we do that and then bring it back to Europe? Because in the end, we also want to have this spirit here like where we are. I think I'm probably still too Swiss, too European. I think we should try to get inspiration from Silicon Valley or from US, but find our own way, which is probably a little bit more not pouring endless money at it, but doing one step after the other and once getting close to probably break even, then you go to a next step. I think in technology, which is very complex, which takes a long time, This is probably the more better way to really have success in long run.

And now, coming back to also the university, a pattern we've heard from ETH founders, but also other founders from universities, is that the jump from a research-grade prototype to a market-ready system is far bigger than they expected. What are the recurring obstacles that you see teams run into during that transition, like be technically, but also organizationally? think it's first really that these people have to understand that it's a long, long way. Typically you say that from a proof of concept, you need additional 10 times more energy resources to get to something which is really can be on the market. this is typically the, we can also say if you have a product which is 90 % ready, the last 10 % is at least as long and very often more. And young people have to be ready for doing this and probably also then learn that probably they should reduce little bit their expectations and go for the early reaches which you can do, which is probably not what you're dreaming of, but you can already do revenue in it because it's less complex. So it's basically focusing also on the boring stuff and not the super fancy that looks amazing but maybe is only like in a couple of years down the line something that then turns into revenue as you said. Yeah, absolutely. And I think this is not so easy because as researchers, these young people dream something and what we do at the university, and this is exactly what we should do, is actually try to approach these very complex problems to contribute to one way. On the other side, as an entrepreneur, you should dream about having something on the market which solves an issue of the society. And if you are not able to switch this dream, you are in trouble.

Yeah. And what, like personally, because you have interacted with so many teams that then turned into successful startups, what early signals give you confidence that they have the mindset and also the resilience to actually navigate this phase and become successful and kind of change this identity, so to say? Very often I'm involved in startups where I know the core people, it's been a while. And you see some of these people already during a PhD that they are really also eager to do something out of what they learned and they discovered during their PhD. But the second thing is also you have to team them up with other minds. And typically what I do is always try to bring in an older guy. Unfortunately, it's not so many ladies, which I hopefully in the future will be better, which have experience, which have experience also to focus on the market. These are typically people which are not in full time executive, but they are somewhat the personal at 60, which is now extremely interested, eager to with people. for them, typically it's a win-win because they are really living a new life. And most of them, which I brought in, in companies like antibiotics or Bodeuro. They really enjoy it lot and they are extremely helpful to help the young people to get a focus on the application and the client. Yeah, and maybe on that point, because you also in other interviews, you often emphasized how decisive the first market choice is. And ideally it should be a use case with a very acute, almost painful need for the customer that he or she is also willing to pay. So from your experience and also based on the experience of this senior experts, what distinguishes also teams that pick the right first application from those that spread themselves too thin? Like how do you do that? Because I think it's one of the

most important but also the most difficult decisions to take and also how to figure out what the right market and the right focus is. Yes, this is, think, a very important element in the beginning. And some people always say focus, focus, focus. I say something a little bit different. In the first phase, which is probably the first two years, you can and you should not focus. You have to first understand, you have to speak to clients to really understand where are the pain points. And a lot of these technologies are disruptive technologies where the client can even not imagine what they can do with it. And you have to find places where they take the risk because they see that they can gain so much. So you need really the pinpoint. But even if in other fields, you can also contribute something, but they will not do it because a lot of companies are resistant against novelty. So if they have a business which is okay running, they continue. And they have to have a pinpoint and to have a strong benefit. Once you are there, then it will... actually evolve and everybody will probably accept these new technologies. So it's really basically that founders should always be out in the market and talk with as many potential industries and clients as possible to actually figure out which one is actually willing to also go with them and which one is maybe just for like having a fancy pilot and then kind of disappears.

Absolutely. I think the contact with the client and the client will not tell you what they need. You have to find it out when you discuss with the client because they cannot imagine what you can do probably with your device. So it's a very complicated process and I tell all of my people which I'm involved as advisor or directly that they should probably go for one or two award which gives visibility. It's also good for investment, but then the awards are the clients. Each client is much more important to have contract signs than 100,000 from an award. Yeah, and still, like when you say, like sometimes they don't even know what they want, it reminds me of this, think Henry Ford said, like, if I had asked the people, they would ask for faster horses and not for a car. And I think so as a founder, you also have the obligation to also then again, think bigger and beyond what is already here, because otherwise we just end up in incremental optimizations, but not in actual innovation, I guess. Yes, absolutely. And I think this is a wonderful example because people cannot imagine what they can do with something which they don't know. And so this is what startups have to do. They have to find it out, to have to test it with the client, but it's not the client telling you what they want. You have to find out what they need. Real pain point, yeah, absolutely. Now, when we go beyond the pilot phase, you mentioned antibiotics, there is also Bluebotics, 7Sense and others. Like when you reflect on the ones that managed to scale beyond this actual pilot phase, what patterns or strategic decisions do you think made the difference that you can also maybe apply for others who are in similar stages?

Of course, in some ways, sometimes a little bit difficult to say because most of these companies, because it takes a long time, are not breakeven. But they also were moving or starting in another time. Now Blue Botix is breakeven, but they have also been acquired. And they had a pattern, which was the right pattern at this time, where they actually had a lot of smaller projects and we just tried to step by step. Yeah.

improve. took quite a long time. Now, antibiotics is probably the other extreme. There is a big dream, there is a potential big market, but there is still a way, long way to go. I personally always think we should go for the middle way, where you try to really as soon as possible to have reasonable revenue. And then from this basis, because then you need to know the customer value, so you define the next step. This would also be the investment step. So try to go with the first couple million to a point where you at least close to break even and then define the next step. And then you can also easier convince actually the investor and they say this be all it reach. This gives confidence. And then you have a story how you want to go for the next step, which is probably even much bigger because you have shown there is the market. I know the client and this is what they need as a next step. And when it comes to this long way ahead of you as a deep tech startup, it also means that you need a lot of different people at different stages. And robotic startups place unusually high demands on team composition like hardware, software, controls, machine learning, someone thinking commercially. In your experience, at what point do founders need to broaden the team beyond just the technical talent, which also ETH and EPFL are very well known for? by really bringing in experienced operators, but also commercial leaders or industry veterans. mean, you said veterans you basically bring in early so they can kind of be a mentor, but what about the others? At what point should you actually also team up with the business side, so to say? I think they should very early on team up with people more from the business side, which have another competence and the drive also. This is by the way, also why we initiated TalentKick, which is early on to bring techies together with people from economics. Now, the next step is very difficult. So adding these veterans, this is easy and this works very nicely.

But adding a person who has probably 20 years of experience in a startup environment, it's very difficult because it's a wrong match, very often wrong expectation. Salary wise, it's difficult and whatever. I'm personally always trying to opt really for getting the best young people and train them so as fast as possible. I'm a strong believer that people can learn extremely fast if they are brilliant. And we have a lot of examples. On the other side, you have also to accept in startups and this is I tell all the startups even before they start that they have to accept that the founders are not automatically at the end, the CEO, the CEO, whatever. They should find the place where they are best at. Even with thinking about if they have a lot of shares, they should actually act for the company and probably saying, yes, I'm probably a very great, brilliant tech, technical guy, but I'm not a leader. So I'm not going to be the CEO, but I'm doing special projects. And this is sometimes painful for these people. I, this is reason why I always prepare them early on to think about this and to, to prepare them so that they, they know at one point they probably, they should step out of the way and have another person, which is technology wise, probably even less good as, but they have the leadership. Exactly, because then you bring in better people and that have to lead. So like we had quite a few examples on our show where also like engineers like were telling me about this moment where they had to step back and let others run it. And it has a lot to do with ego, your identification as a founder, which is also important. But at one point you're basically blocking yourself and you become your own bottleneck. Yes, my experience also tells me at one point you're somewhat in the wrong place and you personally feel it and you're not happy there.

Yeah, that's the other thing. Yeah. So you need to be reflective. I now want to also talk a little bit more about also the mindset because we already talked a little bit about it, but European founders and like including Swiss founders are often described as more methodological and cautious than their US counterparts. In robotics, where position and safety matter a lot, this can be a strength. But there is the saying that the strength can also be your weakness. So where does this mindset help teams and where do you see it creating friction or slowing momentum and what can we do about it? Yeah, I think it's really important, but this depends also from the product. Absolutely. I always say that if I do a search and Google, I get an answer. Yeah. But nobody knows if it's the best answer. Exactly. I just get an answer and this is okay. I'm accepting this. But if I get a car, which is not working half of the time, this is catastrophic. Robots are much closer to the car. They have to have a reliability which is extremely high. It's probably good if we have this approach and if we sometimes break a little bit and make the products really good before we release it, if the client is then having only problem with the systems. Yeah, and so you need to find a middle way. And I think that also requests that you as a founder or as a team kind of can balance these two things. Because in the end, as we all know, we also still need to have this ambition and be very big thinking. So based on decades of mentoring teams, if you could just name the qualities that tend to define a strong deep tech founder today.

Can you name a little bit some of the characteristics that you see that really help a founder to go through this long period and then kind of become and stay a successful one? Of course, the first thing is still that they should technology wise be really also themselves deep. They have to have a good understanding what they're doing. Ideally, they really did some work on this or PhD ideally, whatever. But then of course, the second even more important thing is leadership. Leadership means not that you just are on the command and they always have to follow. You have to convince the others that they follow you. And there we have... wonderful examples where the leadership is so clear from the beginning and this runs very smoothly. We have other startups where probably the technological leader has an even the biggest part of the share, but it's not the person who can lead the company. And this already gives the first friction that you have to change more or less leadership. And this is painful. This can lay back the company. It's also very difficult to tell them the the VCs or investor that you have to change. So you have to organize this early on that the team really is well aligned. And this is by the way, coming back to TalentKick is also the main element that what the young people learn how to reflect on the team and then probably do the best out of the team. And this is very often not really done. And you have to do this at one point. And this is also where, like for example, I was also involved in the Founders Agreement where you actually go through many, of these questions that are super hard to answer, but they're so crucial to actually like learn about the other, like what is your expectation, why are you here and everything in order then to down the line also be aware of what is coming, so to say. So I find that such an important discussion and many, founders don't do it, only when it's too late and you're already in the trouble.

Yeah, sometimes I have always also encountered a situation where the CEO all of a sudden had a life plan, for my view was not compatible with a startup. It's bad, but I think you cannot have found a family where you have then only an 80 % job in the early phase of a startup. This is unfortunately not possible, I think. probably in very specific cases. Later on it's absolutely possible I think, but not in the early phase. I think there it's exactly important that you actually talk about that and then also find maybe an individual contributor role where it is possible and then maybe someone else like takes over the leadership position for it. Absolutely, because it's clearly not that you cannot be part of a startup, but you cannot be the CEO which has to drive everything and which has to be ready 24 hours if there is something. And what role do you also see, because you are a professor and with a very entrepreneurial mindset, what is the role of professors, especially in deep tech startups, that you see work very well for them?

I think the role in this is that you prepare the teams early on as long as they're still around. there is this example. So on one side, PhD students, which already start to think about, typically they start to think about before they finish a PhD. So you can already shape the whole situation. And especially we had a lot of student projects from the bachelor, these focus projects, and some of them turned into startups. And there it's a wonderful situation because then these students can actually do it. do during their master project, their project on their future goal. So they're much more ready. They already have the first test with potential clients during the master project. And this turned out to be extremely valuable. And I have fun with this. I would love to have even more professors at ETH which enjoy this because it's You support it because in the end you also give them some slack that others wouldn't give because they want to like that they focus on the paper and everything and not on commercial use. Exactly. And I think we have a lot of professors which do this. And I think most of them which did it once, they will do it again. But I would like to see a little bit more which tried first and then they will see that it's actually extremely rewarding if your research is ending up in a product which has a value for society. Yeah, so much. Now, robotics is increasingly intersecting with areas like infrastructure, protection, autonomy in complex environments, and also in some cases dual use. Founder asked about this more and more, because it's just like a fact also where we are in geopolitical terms. When young teams navigate these gray areas, what principles or maybe also considerations do you encourage them to keep in mind when thinking about these areas?

So especially dual use is in robotics especially more and more a question. And I think at the end it's each person has to decide for herself or himself to what to do and how to handle this. But I think it's important to discuss this openly within the teams because at one point you probably can even not change because you will lose more than 50 % of the teams. This is quite important. On the other side, think it's also for me, in most cases, a radical decision. If you go into defense, it's a other business. And I think you cannot do, especially as a smaller company, both. Either you go for a civil or you go for defense because it's totally different. In civil, you have a different type of client. And then in military, it's a lot of timing cycles and so on. And if you think you can do as a smaller company both, I think it's wrong. And do you have an opinion, like, because in the end, like we see that also with like startups, for example, in Israel, where they have, they also have a lot of success focusing on this area. Do you think that is like something where Europe should actually play a huge role in? Or do you think like, do you have an opinion about that? Yes, and I can admit that my opinion changed a little bit. I think most of the people which I work with from a peer system and so on, they have also changed. And of course, the war in Ukraine has opened our eyes. I think as long as we can contribute to defend our country or the whole Europe in a way, we should consider to do this. Then of course, it's always a question on a

You can do fundamental research, which is anyway typically dual use. But then you have to think about the next step. And then it's at the end, the question of the people, if they want to go so far and really have arms on drones or whatever, because there is also a lot of other fields like protecting airports in a civil environment, which is a little bit different where I think most people would agree on. That is important. armed robot. is a step more which I can easily understand that most people don't want to work on this. Yeah, but as you say, think that the discussion needs to happen. Like we cannot close our eyes in front of what is happening around us, so to say. Now, you mentioned it at the very beginning, that there is also a huge attention on humanoid robots right now, while more application-specific robots quietly scale as well. But like it's just more fancy to talk about the humanoids. Which areas of robotics do you personally expect to create substantial real-world value over the next few years? I think there is a lot of jobs where first, where there is an urgency actually to have robots in construction, example. We don't find people in construction anymore. Agriculture, it's very difficult to find people to collect the grapes or to maintain the fields. There, I think there is a potential and a need and we should do something about it because this is also very linked with our society. We need a good sustainable... food production and this can be enhanced by robotics and then there is probably the more even more extreme cases in mines on the ground where actually for humans is really not a place they should be and these people working there they have a life expectancy which is typically below 60 or 50 and we should get these people out of their environment because they and their robots can do today already good job because it's less complex than cleaning up a kitchen

Yeah, So exciting. I'm sure like in a couple of years we can sit down and talk about all the incredible cases that have spun out of the universities. Now I want to end with a few rapid fire questions where I just ask a question and you can answer it in one or two sentences. One misconception about robotics you wish would disappear. the hope that they will solve everything. A moment in your career where a student taught you something important. I think this is every day more or less. I'm extremely proud that I have a lot of students which are much better than I am and this is wonderful to see. A robotics problem you hope someone solves in the next five years.

the robot hand. This is a wonderful element and we are still pretty far from having a human, similar to a human hand in robots. So we had Mimix on the show. So I already know one startup that is working on that. And last question, if you weren't a professor, what would you be doing today? I hope I would be a serial entrepreneur because this was my main dream, which I think I was to a certain extent by being a professor, but I would probably still be in this environment. Thank you so much for being on the show. I really appreciated it. And yeah, talk to you soon. Thanks very much. was a pleasure.

We would love to hear from you. How did you like it? What should we Questions, ideas or dream podcast guests? Please let us know. see you again next week. Do you have any such

Thank you. Week.