Could Europe become a world leader in robotics?

0:01Hello, it's me, Amy, and welcome back to the Tifter podcast. On the menu today, robotics, a sector which has absolutely boomed over the past year and where many believe Europe could become a world leader if we play our cards right. Robots are being deployed in industry, healthcare, agriculture and defence. They're doing real jobs in warehouses, carrying out inspections, which are dangerous, expensive and difficult for humans to do. And as my colleague Aina Kelly who joins me on the podcast today found out recently they're even cooking crepes. But as always with Europe I'm wondering whether our robot makers are moving fast enough and with me and Aina to discuss that making his Sifted podcast debut is managing editor Tom Nugent.

0:46Aina welcome back to the show from is it sunny today in Paris? Very sunny. And Tom Welcome to the show from also sunny North London. Very sunny North London. Thank you for having me. So, gentlemen, let's set the scene with some numbers. In 2025, equity investment into European robotics startups more than doubled on the previous year, reaching€1.45 billion. And in the first quarter of this year, European robotics startups have already raised over€500 million. euros. What has triggered this interest from investors, would you say? There are a few things. I think there's acknowledgement that the developments in the AI world have allowed robots to do a lot more things.

1:36A lot of people I speak to, definitely researchers as well, seem to tell me that we're going from this era of robots being a thing you hard-coded to do one specific task over and over and over again, to an era where because of AI developments, you're able to get them just to do way more general stuff and to understand their environments much better than they used to be able to without the need to sort of hard code specific tasks over and over again. So I think there's like AI developments being one and two, just this comes up again over and over the labor gap that is coming in the industrial space in manufacturing, where you have this generation of workers who are retiring and this new generation of workers who just don't want to work in factories as much as their parents did.

2:26So you have this massive labour gap coming alongside this boom in artificial intelligence that has allowed robots to do a lot more things. One of those things being be way more efficient in a factory or a warehouse environment. And the robots that we're actually seeing, I guess, deployed already in industry or healthcare, are they still the kinds of robots that do one specific task? Or are we actually seeing those sort of next generation, cleverer AI enabled robots in the workplace already? Or is that what the investors are betting on happening soon, hence the influx of capital? They're definitely betting on that happening soon like you have a lot of the a lot of big companies like bmw mercedes-benz volvo they're like they're working on pilot programs or pilot tests with a lot of these these companies where their robots are doing maybe one thing over and over again on an assembly line that's definitely one area that hasn't maybe changed much but the efficiency and the speed at which those robots are able to do that has the other side of that bet that one day we'll be working alongside robots that look like us act like us do tasks like us that's like still a long way into the future i think we'll get into it a little bit later on on this episode but like that's the big bet that investors are taking that we will get there one day but we're not quite there yet and aina you You recently went to Europe's, I think we can call it Robot HQ in Zurich, and you witnessed the continent's first bot-on-bot boxing bout.

4:10How did that rank as a spectator sport? I loved it. Yes, I think everyone there was just very giddy to be seeing two robots punching each other. We're very simple creatures sometimes. uh yeah it was it was very cool it's true that these robots are quite limited still as tom says you know they are part human essentially you know these robots required two humans wearing headsets to direct the robots so you had two guys in the ring with the robots and they were just punching midair and yeah it was a lot of fun this was this was uh i should say this was the eth robotics club that hosted this event.

4:55And it was taking place in a large hangar outside of Zurich that belongs to the university, ETH Zurich University. And yeah, it didn't take long for the robot to malfunction and it was kind of game over. But what we did see was very exciting and a lot of fun. What else did you learn from the ETH robotics lot? What did it tell you about what robots are capable of? Often, as you say, where there may be many limitations still are. Well, I went there, I think it was three weeks ago now, and it was just a few days after a Swiss robotics company called River was bought by Amazon. And this company makes these kind of dog-like robots that help with deliveries.

5:43And yeah, so there was a lot of chatter about that, a lot of excitement. So that's one side of it. I did see a lot of robots that were just falling over, that were failing to do simple tasks. The robots that were being used, I should say, at this event were from a Chinese manufacturer called Unitary. And you've seen these robots on Twitter, I'm sure, running around racing tracks or dancing. They're quite amazing. But yeah, there's still a lot of things they can't do. But yeah, Switzerland is definitely having a moment. I mean, just looking at all the sponsors of this event, OpenAI, Tesla, NVIDIA, lots of VC firms that we know, General Capitalist, Rydalpah, and Founderful were there as well, Zurich VC.

6:31So there's a lot of interest. There's a lot of feeling that there's smart people coming out of ETH, Zurich University, and EPFL. I think Switzerland have figured out spin-outs a lot better than other European countries. I think that's fair to say. So yes, there's a lot of eyes on these universities and a lot of interest in their engineering pedigree. And Tom, you've been writing quite a bit, haven't you, about the sort of next big challenges to overcome in robotics, including, as you mentioned earlier, this conundrum of how you can get robots to actually take in data about the world around them and understand it.

7:13What are the different approaches to solving that? Yeah, it was actually really useful to go back through all of my notes to actually see how many different approaches there are. Yeah, I'll start with the most recent company I covered, which is an Oxford-based spin-out called Stateful Robotics, founded by someone called Kirstie Lloyd-Dukes, who was the CEO of an imitation learning company called Latent Logic, which was actually acquired by Waymo, the big US autonomous vehicle company. And she stayed on at Waymo and was there for a little bit as well, which I imagine helped her understand how you operate a company at scale that is working on these systems that help robots understand the world around us.

7:58So Stateful Robotics is building this software layer that you can apply to any robot, any form factor, any industry with a particular focus on memory and long-term planning. And this is something that Kirstie told me is a real problem at the moment is that you can have robots do tasks over and over again, but they don't remember doing those tasks and they don't have the foresight to know what tasks are upcoming. They're very much operating in the immediate presence. So Kirstie was like, this is our opportunity to do this. We're going after this area of the market because vision language action models have got really good.

8:34So the visual aspect of a robot is much better than it used to be. But this memory problem is holding that area back, as well as this long-term planning thing that they're going after. So that's one approach. And then you have companies like Genesis in Paris, founded by an ex-Mistral scientist, which is building this data engine to power AI models for robotics, using a lot of synthetic data. So like a simulated environment where you can train a digital version of a robot and then take that what that digital version of the robot has learned and put it into the real world so that's another another approach and then one that i think is is quite out there quite ambitious is neuro robotics in germany so i spoke to david reager the ceo last year and their big bet is on these robo gyms where if you imagine a huge warehouse instead of squat racks and bench press machines you have different stations where you might have a robot operating in a kitchen environment making you a cup of tea or making you a cup of coffee you might have a mock patient setup where there's like a hospital bed and a humanoid robot is practicing looking after a patient so you have loads of these stations where nura is able to not just train its own robots but also the plan is to rent out space to other companies that can train their robots in these warehouses in these robot gyms as well and the idea is that all this data that's been collected goes into this central AI model that Neura is developing.

10:04And David thinks that that's going to be the real revenue generator for the company is selling access to these gyms where other companies can train their data. And obviously, Neura benefits from developing this huge AI model from all of this. And they have a place in China. They have Tokyo and the US on their radar as well for these sites too. So really, really ambitious, ambitious play. It sounds kind of hilarious and ridiculous, doesn't it? That might actually be the best way to train a robot. But I guess it's good that there are these different approaches going on. What are the other big challenges for robotics as an industry?

10:46So there's kind of just the general understanding of the world around them. What else is holding robots or European robots in particular back? I think if I can jump in on this one, dexterity is another big one. So I think what I've come to appreciate the more I've got into this beat is just how crazy it is that you, Amy, or I, or Aina can walk around and just act and do things without really thinking about them. And then when you try to turn that behavior into code, it's incredibly difficult. So if you take the human hand, for example, robots have a really, really hard time feeling their way around the world.

11:31So like you or I might pick up an egg and know not to squeeze it so it doesn't crack. A robot doesn't really know that. So there are a couple of companies working on dexterity and working on humanoid hands in particular. This is a big problem for the industry. One of them is Mimic Robotics in Switzerland, which is developing humanoid hands as well as its own AI model. And it has humans wearing these lycaptic device type things, carrying out tasks with the hope that by carrying out these tasks using real human hands, you can then transfer that into robot form. and then there's physical robotics as well in Norway founded by one of the guys who was an early co-founder of OneX the sort of the big humanoid maker that moved its its HQ from Norway over to Palo Alto last year I think that was and similarly they're working specifically on dexterity and and the human the human hand which is I think is is really interesting.

12:31I feel like there's there's a commission here for Eina where we need to get you I don't know what we call the people who do the robot training, but I want to see Aina in a kind of robot suit going and picking up strawberries and bananas and eggs or something for your newsletter soon, Aina, if you'll take me up on that. That's a great idea. I'm up for that. When I was in Zurich, while Tom was speaking there, I was just reminded of some of the tasks that the robots were doing. the students had a showcase of they had they had a hackathon all weekends and they had 48 hours to come up with an interesting thing for robots to do and one of the tasks was for a robot to make a marriage proposal so to go down on one knee and everyone winced in the room because he went down with such a force I mean if a human did that it'd break their knee it was very funny.

13:26So on the customer side of things where which industries are we seeing most uptake for robotics at the moment and which you know how kind of willing are europe's big corporates to take a punt on these robot upstarts warehousing is one um there's a company called dexery in the uk um which has started to to generate revenue from from its platform which is basically these big vertical um robot pillars that scan a full warehouse while at the same time creating a digital twin of that warehouse so it gives warehouse operators a full picture of everything that is going on um in their warehouse down to like a product level so you know when you need to reorder you know what's going out you know what's coming in um and they've got a lot of traction they made they're not making loads of money they made around three and a half million pounds in the 12 months to march 25 according to their latest company's house filings um they are loss making but they raised a big hundred million dollar plus round last year so warehousing is definitely one that is getting traction robco is another that has been quite open about its um its revenue generation so this is a german company which builds robotic arms and a software suite for machines on like a factory floor floor um and it says it's doing tens of millions of dollars in sales and is approaching 100 million dollars in run rate um so again not not huge but there are factory operators who are buying its its kit um and i think it's deployed around 400 units across more than 100 sites bmw is is a customer affairs um so his it service provider t systems as well so so that's another another example according to some people like the humanoids are a bit of a waste of time uh they're still very far from being useful i mean tom had great examples there of robots that don't necessarily look like humans uh that are already functional and making money for companies but yeah there's a lot of critics of of the humanoid pursuits people who say well you don't actually need robots that look like humans it it might take too much to actually make them useful uh you know for example these guys who are trying to introduce robots to the house uh you know one x being a company that tom mentions the robot butler or housekeeper well actually a lot of the tasks that it could end up doing there's probably other ways to do them you know like washing dishes a dishwasher can do that pretty well it's not clear that a humanoid will be able to do it any better and yeah the hardware is still very expensive uh tom tom might tell me otherwise but i don't know if there's any european companies that are you know doing doing well on humanoids hardware the china is doing very well china has very good supply chains the costs are very low uh compared to what we're seeing in the us and europe for this hardware so yeah that's another kind of impediments for human rights i think yeah i think the fact that when you endure it the two robots you saw boxing were from china says a lot right like if if we had the capability in europe to to manufacture these things and use them um we would but but the reality is we don't and when you look at the the industrial deployments of humanoids china is just so far ahead than we are so how would you how would you stack uh europe in the global ranking is it is it china china number one us number two where where where's real innovation and real sort of use cases for robots taking off yeah i spoke to some people for a piece uh last year and there was a lot of skeptics uh they said there's a lot of expensive research projects underway that's perhaps don't look like they're going to be profitable anytime soon china seems to be very far ahead on humans japan has a very good uh history with robots the us as well uh i guess for europe software is looking more interesting and more likely to be uh the european edge for for startups uh tom do you want to tell me if that's right or wrong yeah i i think i think that is right i think it's if you look at what happened with with llms like the the foundational model makers in the us we obviously have mistral in in europe but it doesn't come close to to your open AI's, your Anthropics, or your Google's.

18:24So I think similarly in robotics, you have foundational model players like physical intelligence and skilled AI in the US, which are valued at both over, I think,$10 billion. So there's a similar sense that they're developing foundational models for robotics. There's a sense that the foundational model play in robotics won't be Europe's, but this application layer, this development of software on top of of hardware that can can do these tasks that they're required to in factories warehouses really really well is an area that that we can win but it's still very very early days um the companies we mentioned before like your stateful robotics physical robotics mimic the customers they're working with it's on pilot tests like they're nowhere near developing tens or hundreds of millions of dollars in revenue um it feels very much like because a lot of these a lot of these founders are researchers they're really smart people coming out of universities the companies that they've started are still at that like research phase a little bit they're still figuring out what exactly works and still gathering the data they need to prove their use cases so we're still a long way off i think from being able to tell who the who the real winners are so it feels like we're behind behind the us and china at the moment who's to say that won't change but at the moment it's it's early days it also feels a bit like um there's no one european company that's like we are europe's robot champion i mean i appreciate there's lots of different ways of approaching robotics and lots of different use cases but this doesn't seem to be you know there's no mistral of robots right there's lots of companies doing quite specific things for specific industries or you know with specific mechanisms um and and do we think that means that outcome is that lots of these companies get acquired by a bigger chinese or u.s competitor yeah i think it's a fair question i think we obviously had river um as aina mentioned before acquired by amazon last month and spoke to a number of investors when when that happened because i think there's a tendency in europe when when a company like that sells to a big u.s company it's very easy to to complain and say it's that the company by the way hadn't yet reached a series a so it raised the seed around a couple of years ago i think uh and then sold before it reached series a and there was there was a mixed reaction but a lot of reaction was yes this is another example of a european company that was quite hyped and came up in a lot of conversations that i had with people in the industry selling too early.

21:11But I think it's important to get perspective on that if you go back 10 or more years. Obviously, a lot of people still bring DeepMind selling to Google as an example of a European company selling too early. But ultimately, if you look at the impact that's then had on the ecosystem, 10, 12 years later, there are superstar researchers coming out of DeepMind starting their own companies that could be global winners one day. So I think the perspective needs to be, yes, this might have been a company that sold a little bit too early and we might see others that sell a little bit too early to big US players.

21:50But actually the impact that could have five, 10 years down the line is huge. And it's when that time comes, when we look at the actual outcomes of those sales, that's when we can sit down and get, okay, maybe this wasn't that bad after all. Because again, it's super early, right? We're super early in the development cycle of a lot of these companies. So I think it's okay if we don't have the global winners maybe in this cohort, but the next cohort that comes through, maybe that's when we see the global winners coming from Europe. I think there's also space for, you know, quote unquote, dumb robots as well.

22:29I was recently chatting to a guy, another Swiss roboticist called Robert Hennig from Basel. And he's come up with crap making robots and it's not cutting edge at all. It's a big hunk of metal. Took him a few months to build. and he was actually at this event in zurich this this robot boxing event and people loved it people were crowded around this robot it made about 300 creps in one weekend and there you go it was such a contrast with the humanoids it didn't look sleek at all but it was making great creps it's making money it's useful now it's in the workforce so maybe that's a sweet spot as well i'm all for more robot wars that's a reference for anyone else who grew up in the uk in the 90s to a brilliant tv show where parents and kids would make robots at home that would then fight each other so maybe we can bring a little bit of that back to europe's stuff i've seen as well on that note that's all we have time for so thank you so much aina and tom for joining us And if you want to hear more from Zifted, please sign up to both our daily newsletter and also our AI and deep check newsletter, which comes out weekly on a Monday.

23:52We'll drop the links in the episode description. And as always, please rate, review and share the podcast if you enjoyed it. And thank you to Maya Dorempel-Wambi, our wonderful producer.