E345 | Sam Cash (Entropy Industrial Capital)⁠, Eric Truebenbach (Teradyne Robotics Ventures) & Claude Ritter (Cavalry Ventures): GP/CVC Roundtable on AI in the Physical World

33:47So I mean if the question is do I think there are enough investors in Europe who are willing to invest in these types of companies. I don't. I think the market currently is massively underserved. I think kind of zooming out and thinking about this trend, you know, the thesis that I have and the view that I have is that technology's impact on the physical world is going to be an order of magnitude larger than that of the virtual world. Sorry, let me say that again. My specific thesis is that AI's impact on the physical world is going to be an order of magnitude larger than that of the virtual world.

34:34If you believe that and you believe that this current trend of AI is real, then what I'm seeing kind of on the ground in terms of people who are willing to deploy capital really early on in companies that are in robotics or using AI for some kind of physical world processes, there is a dearth of coverage for sure. And I think that to some extent is being because, one, a lot of generalist firms in Europe have done very well from things which are not robotics. Number one, whether that's SaaS or fintech or otherwise. And two, I think we're only just starting to see some of these big, big success stories in Europe.

35:23You know, whether it's Wave, who just announced a billion dollar funding today, whether it's Exotech, whether it's Autostore. So we're only just getting to this inflection point where I think we're having, A, enough case studies that people can get comfortable that this is a market worth investing in. be enough talent coming from the deep mines and waves of this world to build some of these companies from day zero. And I think the remaining C is having a depth of institutional investor at all stages pre-C to IPO that are willing to underwrite some of this risk and really understand what it takes to build these companies.

36:07I think that is changing. The narrative historically for these companies in Europe has been sort of okay to neutral. I think we're at this point now where people are really starting to pay more attention to some of these companies and some of these founders. So I think generally in Europe, it's undercover, but I'm hopeful that over the next five to 10 years, there's going to be far more people willing to back these types of companies. Claude, I'd love to ask you as a generalist investor who have made some bets in this space, but I'm sure you've also seen more deals in this space that you're like, I think there's something there, but it can't get comfortable.

36:51Whereas if it had been a pure digital play or something that you've done a bit more of, you would have maybe been willing to make the bet. meaning and that's of course that's the critique that we're giving when we're saying that there's there's missing investors willing to put their bets here it means that basically we're saying that the investors it's you know the problem of funding is with the investors it's not with the startups because if there's lack of quality and lack of viable business models and so on so forth then then it's it's not a matter of of the investor side but when we say that there's missing capital, it's because the investors are not comfortable making the bets.

37:33I would, I would, I would, I mean, I, I tend to agree. I think there's obviously reasons for that. I mean, if you look at a business that, that let's say produces some sort of physical device or a robot, you know, and then you talk to a generalist investor that, that typically or historically has only been doing software like that person, you know, probably doesn't have a business MBA and has no idea how the P &L of a hardware company like 10 years down the road looks like you know what I mean and so it's very difficult to sort of underwrite something like this because you have no intuitive feeling you know not from academia not from sort of doing it for years for how these companies you know scale what to look out for and so on so I think that definitely makes it more tricky to sort of look at opportunities that involve, you know, hardware.

38:28You know, I think for me, though, personally, that's not sort of the biggest issue I have. I think for me, I'm, you know, I'm happy to sort of engage with anything that I find interesting. And I'm happy to sort of, you know, spend days trying to sort of get up to speed with with something that I'm not an expert in. But for me, and that's more broader AI thing, I find it increasingly difficult to sort of, you know, really figure out what is something that might be something that lasts and what is something that, you know, might just be gone in a year or two from now. And that sort of transcends the digital software world to the hardware world, where there I even obviously have even less sort of intuition, but it's pretty much the same thing sort of across, you know, across hardware and on software.

39:30I want to add something to what Sam said. And I think this is really something that I've been thinking about a lot. I'm not sort of done thinking about it, but I think I'm kind of getting the hang of it for myself, which is, you know, if you look at some sense, sort of the impact on the physical world will be order of magnitude higher than the sort of non-physical world, digital world. I'm not sure if I fully agree with this, but if you like broadly just think about sort of the world's GDP, you have sort of, I think, don't quote me on the numbers perfectly, I think you have like 60 % in like white color services type work.

40:11And then you have like 40 % in sort of, you know, blue color, non-services, more manufacturing, logistics type work. And I think for both of these sort of buckets, we're far from sort of reaching any point of saturation with regard to sort of AI being deployed at scale, right? Like we tend, we technologists tend to think it's over. Or like, you know, I can tell your average German company has not heard of it, right? But when we think it's over, your average German company has not even heard of it. So I think, you know, there's a long way to go in both. But I do think that because of a lot of the, you know, technological innovations that both Eric and Sam have been mentioning at speed, like you can feel it, that the speed of iteration is sort of increasing.

40:57I think we will see a lot of sort of, yeah, innovation in this sort of 40 % blue collar work bucket. And that will, I think, have, in terms of impact, to come back to what Sam said, you'll have a massive impact because it will be an impact that we feel in our daily lives. And so it's not just like a monetary impact. It's also a visceral impact in terms of it really being visible in our daily lives. I think Eric also said he's not into home robotics stuff, but imagine you see home robots, you see those things appear and all of that. I think that will be part of this order of magnitude higher impact that I would say is included in what Sam mentioned.

41:49I mean, I think BMW said they want to have humanoid robots in the US in 25, right? So, and I think that's not too far away. So I think, you know, you will start to see that portion of the impact pretty soon. What Claude and Sam both said, I think, is something that is one of my great fears. AI is changing so quickly that any investment I make now will probably be obsolete by the time they get to market. And if it's a physical application, that's what takes a while. You might have the model all done, and then you have to make a product out of it and make the product work. And so there's always going to be a lag between AI, not always, but for the foreseeable future, there will be a lag between AI development and AI deployment.

42:36And it's going to be a difficult dance for as an investor to find the right places to go. How do you, Eric, because you're helping on it, right? You've made the commitment. So how do you kind of then make that jump? What's the litmus test that you used to say? Because I think many are thinking exactly what Claude said. Will this be sustainable or will it be like two years down the line? We're on to the next thing. If I had the answer, I wouldn't be afraid, right? I said I was afraid of this. And I don't have the answer. All I can say is what I'm looking at now is companies that are using AI as a key part of a product, but it is not the product.

43:23Because I think that could fall down fairly quickly. um yeah in fact there's one company that i've been following for years that it was the product and i'm just seeing them slipping to be honest so um if you have it as part of the product you can incrementally improve it without having to redo everything else i broadly agree um i think we consistently see in technology these hype cycles if anything the hype cycle seems to get more pronounced and of higher volatility as time goes by. And I think whilst we're obviously in a period of massive excitement around AI, I think most people are of the view that it is a paradigm shift in terms of a lot of the various applications that can come from it.

44:17Now, whether we will be excited about AI in its current instantiation in two years' time or three years' time or 10 years' time, I don't really know. I think what we've seen historically, if you look at some of the big sort of meta trends that we have in technology, so from the PC era to the client server era to mobile, cloud, SaaS, maybe AI is kind of the next transformational wave. And the thing I'd say is looking at those trends, whilst on any given year, people probably had varying levels of excitement from those, what you tend to see is the ability to invest for really prolonged periods into those trends.

45:05So 5, 10, 15-year periods. I mean, people are still investing in mobile and cloud and with a great degree of success today. So, you know, whether we'll be doing this podcast in two years time and talking about the same things, I don't really know. But I have absolute certainty and clarity that there's going to be fundamentally a lot of incredibly interesting investments and technologies being built over the next five to 10 years. Yeah, and I think that that's the perfect pivot to the next question I wanted to ask you guys, which is where are the largest opportunities you're seeing right now? What are the spaces where you are seeing like, wow, here it's really, really just like we're at the right point of the inflection curve right now.

45:56and we're actively looking for the next thing to invest in this. And maybe, Sam, I think I want to start with you, actually, because you have a very developed thesis on three larger areas. So I think that you're also one of the more thesis-driven investors here. I think fundamentally, if you take this meta trend, which is AI in the physical world, and you look at the sub-segments below that where there have been big, large-scale macro shifts, I think there's probably a few. There's one, a lot in energy and climate. There's two, a lot in supply chain and manufacturing. And increasingly in Europe, especially, a lot in defense.

46:47And to try and tackle those individually, where we've seen AI tackling problems of significance in energy and climate. For example, DeepMind's GraphCast model, which came out relatively recently, this is an AI model which massively improves our ability to predict weather or chaotic and relatively random scenarios. Why that's important is it just means that we have a much better understanding of climate. Climate today has been sort of a really hard to understand complex system. So we're seeing applications that are similar to that. We're also seeing a lot in materials discovery, whether it's on the chemical side.

47:34There was a big open source dump of sorbents that Meta came out with, which are sort of the key ingredient needed for direct air capture. So there's a lot in these types of fields where AI is basically increasing our ability to search and discover really complex things within the physical world and otherwise. I think looking in supply chain and manufacturing, to Eric's earlier point, I think there's a lot of interest in humanoids. That's maybe an area that is potentially overhyped. But a lot of new companies that are applying foundational models to really difficult problems within supply chain, within manufacturing, within warehouse logistics.

48:27And I think whilst a lot of those businesses are not net new. They are a massive increase and improvement on a lot of the historical state of the art. And then maybe lastly, in terms of defense, this is one that I guess is typically a bit more sensitive for people. But the reality of a lot of what is happening in Europe is a lot of countries and sovereign states need to become more technologically savvy and need to requiring new technologies within the realm of defense. And there we're seeing a lot. We've seen the market shift from effectively defense forces and departments of defense having what they describe as mass, which is these big, large systems, tanks, aircraft carriers, to actually what we've seen in Ukraine and in other places, which is kind of this new era of asymmetric defense, which are typically people using drones or unmanned subs.

49:36And there, there's a lot of AI being applied where they're making a lot of these robotic systems, whether it's a flying drone or a boat, specifically unmanned and autonomous. So there's lots of interesting things happening. I think in terms of how I think about it, I tend to get led by where the most compelling and most qualified technical founders are building. But really, there's lots of new interesting spaces in Europe and in the US that are worthy of attention. I am personally very much looking forward to the first defense fund investment from our side. I think this is a space where an emerging manager is incredibly well positioned, given that we have many legacy funds or large funds, established funds.

50:28I'm using the word legacy today, sir, for that's all the ones that feel offended by. We have many of the established funds that just do not have it in their mandate, which means that for probably a two to three year period, we will have some emerging managers that are the only ones that can really capture these amazing opportunities. Eric, let's go to you and ask you the same question that Sam just replied to, which is, where are you most excited about the recent developments? Well, what I'm most excited about and what's the most likely, maybe two different things. It's always been a question of training data.

51:06And so one of my annoyances about AI in the past has been there's been way too much research that was based around the availability of something like ImageNet. So I feel like that was limiting AI development because in academia, at least, they couldn't afford to create a new training set. Now the foundational models are starting to train on not the exact training data, but something close to it. And they figure they can tune it from there. So, where I think it's likely to see the next things are where large training sets are available or can be easily generated from the existing process. You instrument up an existing process and then see if you can improve it or extend it or do something new with it from there.

51:56And in academia, there are some, but they're generally simulator generated and that's still not good enough for the physical world most of the time. So I'm going to leave it vague around the training sets because I'm still looking into where those might be. I have some ideas, but I'm not sure I want to disclose them right now. What I'm most excited about is trying to bring collaborative robots to the next level. They were easier to use. They were safer. They were a bit more accessible to the common, the first-time user. but it's still too hard. There's an ecosystem that seems to actively get in the way of making it easy because of tradition of experts selling to experts in robotics.

52:46And there's just a lot of details that are really hard for a beginner to understand. And it's that place where I really like to see the application. And there may be training sets available on that where you can say, look, here we go. I need to do this. Is there anything that's been like that before? Can I use text input and say, design me something like that? And they're looking at the result and say, well, could you make the footprint a little smaller? I mean, you can do that in chat GPT now. It just doesn't have a physical output, but it's certainly possible. You could start developing that today.

53:25So, and then generating training sets. There are companies out there today that will follow a human around and determine whether they're doing something that's ergonomically correct. And I'm talking, this is AI based, where it's all video input and they track humans and they either verify that they're doing the right motions or that they aren't making any harmful motions or that they're being efficient with time and space. I'd love to take that training data and then take it and say, can a robot do that too? And the answer could be, yes, it could. Or only if you did this, or it'd be 10 % slower or 100 % faster.

54:09That's starting to come. And it's starting to come from an ancillary application of watching humans. But now you can start trying to generate automation around that. That would really excite me, but I don't see it happening just yet. But I see the pieces falling into place. Eric, what you just said is literally what my notes say. Like, I have a line that says, so the first thing you said, what is likely or what you see, it says for me, synthetic data and simulation for physical spaces. So it's really, and then the second thing you say, I briefly mentioned one of our companies called Deltia, the vision-based system for sort of shop floor insights, right?

54:50And so if you think about it, obviously today you watch someone manually assemble whatever, a vacuum cleaner, right? And you can, from that, you can derive things with regard to efficiency, quality, but then obviously also by watching humans, You obviously, you know, collect data around sort of how, you know, things should be done. And eventually, you, at least that is the idea, you can think about sort of moving into something like collaborative robots and eventually automation. And so this is definitely something that, you know, we are, me personally, I'm very excited about. I've just seen a few companies that have tried to sort of skip two steps and just, you know, ended up in a place where it wouldn't work or customers wouldn't accept it.

55:47Right. And so I think an approach where you start with something that is a bit less invasive, that delivers sort of value immediately. and then sort of you transition from there and build up your capabilities along the way while you sort of learn more and are able to train your own models, I think is something that I'm very excited about. And we have two or three companies in our portfolio that are sort of on a similar trajectory in sort of different areas. And I think that is something that I like a lot because obviously as an investor, if we invest today, you know, you have to think about the next 18 months or 24 months, like, you know, how can you show something as such that, you know, someone else will invest in 18 or 24 months from now, right?

56:36And so the most beautiful, and I think it goes to Eric's, you know, finance, like finance point is like the most beautiful vision of the future 10 years down the road will not help you in most cases, unless you're the most amazing salesman, you know, it will not help you sort of to get there right you have to sort of show something along the way you have to show value you have to build distribution land with customers learn and then sort of move from there and i think that approach that this particular company that i mentioned delta is taking is something that i find really exciting you know whether they end up in a situation where you know they will sort of reach this end goal of being able to sort of help customers automate is TBD.

57:21But I think that is an approach that I'm really excited about. And now to take us out of this, I just want to end on a cheery note of asking you, where do you think that Europe is well positioned and why? I guess sort of a bit of a narrative violation is I think London is actually really well placed to be a new and exciting robotics hub. so it's timely that today is the day that wave which is an end-to-end autonomous vehicle company announced their billion dollar raise but i think if you look at a company like wave there's also deep mine who put out rt1 which is a almost seminal research um paper about 12 months ago that really changed a lot of what's happening in robotics.

58:16Both of those companies are based in London. And I think what we're starting to see at the moment is a lot of robotics, regardless of the use case, a lot of robotics has moved from a lot of the innovation being predominantly mechanical to a lot of the innovation now being around vision systems and AI more broadly. And we're starting to see in London in particular, a lot of that talent leave some of those larger scale companies and start new companies. So if I was to have one prediction for the next 10 years, and maybe this is a bit hopeful given I'm from London, is I think we will start to see a lot of that really high quality, qualified talent starting new companies.

59:03And London will start to emerge as a bit of a hug for a lot of European robots. exciting Eric you come from the States so you've looked at all across Europe and decided that Odense my hometown is the best place well that's putting the cart before the horse I actually used to live in Europe so I'm not entirely an ignorant American in the Netherlands yes so I'm still forming my thesis and I'm still asking people about the investing climate in Europe. And I think the answer may be different in different countries. I certainly have heard, as Sam said, that the climate in the UK is much more similar to the US.

59:50So I expect more than half of my investments to actually be in Europe based upon all the startups I've met along the way the past 10 years. I think there's a strong focus on industrial applications. There's government funding that actually is trying to support manufacturing and commercial applications. Whereas in the U.S., if you want government funding, it either has to be educational or help the elderly. It has to be some sort of something that tugs at your heart, I guess. So I like the practicality. The heart of voters. So I like the practicality in Europe, and I like the focus on commercial applications.

1:00:32On the other hand, I'm not seeing a good geographic center. Sam thinks it's London or will be London. I haven't really found it. Maybe Berlin. I just, I'm not sure. And that makes things difficult. It makes it much harder to find things. And then just the funding environment, right? In the US, you could say, hey, I'm going to build a humanoid and people will line up to hand you money. And I'm a little arrogant about this because I think I understand the technology a bit better than the average investor and that I can make better investments. But the amount of funding available in Silicon Valley is just mind boggling.

1:01:13And it's a much harder road to go down in Europe, but I would like to try to make a difference there. Eric and Sam are both much more well-versed in the intricacies of the robot landscape and have more well-formed pieces around this. For me, the way I look at it is pretty macro, to be honest. I think in Europe, we have to do something. So it's a necessity that stuff will happen in this space. it's a bit like betting against the energy transition. Like you probably shouldn't do that. And so we're sort of AI in the physical space, robotics. It feels a bit the same to me in a way where we have demographic change.

1:02:06People are aging out of the workforce. We have sort of pressure on Europe's sovereignty and all of these sort of macro factors, I think, will lead to an environment that has to be innovative. And I think, particularly in Germany, if you look at the structure of the country, it's a bit different than other countries where there's not this one single city like London and the UK where everything gravitates towards. You have Munich, you have Berlin, you have brilliant universities in Darmstadt and so on where a lot of interesting stuff is happening. And then also you have these Mittelstand manufacturing companies spread across mostly southern Germany, to be fair.

1:02:55So I think generally the ecosystem and the experience and heritage, if you want, in the built world is here. And so there's a lot to be done. There's customers that will feel the pressure to move. And I think I'm with Eric on that one. You have to sort of work with the customers to really sort of get these things deployed. And I think here we're in a really good position to do so. The financing thing is obviously without an issue. I think a few weeks ago, I sort of looked at, I think, the combined market cap of the top three or five European stock exchanges. And it was like half of NASDAQ. So I think, you know, so I think just like from a high level, right, there's clearly a bit of a financing issue.

1:03:41But I think a lot of American capital, obviously part of Eric's capital, will be moving to Europe as well. So I think that the capital issue will be solved to a certain degree, at least enough that will allow us to sort of build interesting companies. Right. And then so I think this sort of combination really gets me excited. obviously lots of issues and so on and so forth, but I think it's a super exciting opportunity, particularly in a place like Germany where you have so much manufacturing. Amazing gentlemen. I am so thankful that you joined me for this session because you scratched an itch for me that I've had for a while to talk about this topic with some of the leading investors that we have in this space.

1:04:27So thanks so much for joining. I really hope that you will be joining us in Odense, Eric and Sam, Claude, I'll think of you when we're there. To everyone tuning in today, thank you so much. If you enjoyed this episode of the European BC podcast, do make sure to go in on eu.bc and subscribe, where you'll also be able to find the show notes of this conversation, which will hopefully enlighten you much further. Thank you.

1:05:19instead of competing. Where the ecosystem is capable of commercializing startups and new technologies. Where you will find an international test center for drones. And where the university is so good it won the unofficial world championship in robotics. That place is the Danish city of Wurmser.

1:05:49It all started at a Williams-Steele shipyard, brought to life by A.P. Muller, the founder of Maersk. Academia and highly educated people were not dominant. It was a production city. To be able to compete with low-cost shipyards, they had to rethink their manufacturing process. So the owner of the shipyard decided to donate some money to the local university to stimulate more robot research. It was so successful that actual working robots for shipyard welding were produced. Despite their many efforts, the steel shipyard had to close. They were not competitive because of the high wages and the technology at that time was not ready yet.

1:06:29Shipyard was not the only one that was closing at that time. There was really a point where there was a lot of the last companies that was closing. So there was some kind of bad years. It was a major blow for the city. I'm happy to say that there were quite a number of people who were visionary. So there was a new hope. A new team of innovative people had some great ideas. Maybe you've heard of something called Universal Robots. We wanted to make a robot that's more like a tool that helps people do their work, more than this big machine behind the fence that deals with people's jobs. The team created what today is known as collaborative robots or cobots.

1:07:09In 2015, everything changed. Universal Robots was acquired for$285 million. It actually opened the eyes for the world at large. Everybody would stop talking about the future is collaborative robots. Prior to UR, there was a lot of talk about all the things that never worked in Odense. And now people, they talk about all the things that are possible. So everybody suddenly realized there's something special going on in Odense. And this was only the beginning because it wasn't just universal robots. There's a whole string of equally great companies in the making. Companies that are also growing at phenomenal rates of more than 50 % a year.

1:07:50Ranging from startups to large corporations, Woonz and Our Houses dozens of robotics companies and a total of several thousand employees. I have not seen any denser cluster of robot technology anywhere else. We have the university with the world-class research. We have Technological Institute, which brings technology out to the industries. And then we have the local government, who is all for robotics and all for robots. People see each other as colleagues more than competitors. Companies are building their own business on other companies also. And they need each other, so they exchange information, they even exchange people.

1:08:26We have some extremely exciting jobs here that you cannot find in other cities. The limiting factor on high-tech industries is usually talent. And talent is attracted by talent. And the critical mass of talent is in Odense. There's a huge interest in investing in Odense now. And it's been a lot easier for the startups to raise funding. When you come out and say that you are part of the Odense robot cluster, you get an ear with the customers. But Odense needs more. It needs you. We are already one of the leading cities in the world, but I think that Odense as a robotic city can grow even further.

1:09:08It's probably one of the best places in the world to start building robot businesses, and that's why Odense has become, let's say, a rising star in the robot scene in the world. If you are able to succeed in robotics, then it's here.

1:09:54Thank you. Europe is a story of new beginnings. New beginnings. Let's start acting.