Patrick
I grew up watching the Terminator films. They were stories about machines that were rising and humanity worried about what it had created.
Today, humanoid robots aren't sci fi anymore. They're real. They're arriving fast, and they're starting to shape global conversation and the sun to shape global conversations. Case in point, being the Nvidia CEO just the other day, he said that the ChatGPT moment for robotics is here.
And 2026 is the year of physical AI.
I'm Patrick Coffey, this is the Barclays Brief, and my guest today is Zornitsa Todorova, Head of Thematic FICC Research. Zornitsa is on her way to Davos to present her work on humanoid robotics. Welcome back to the podcast.
Zornitsa
Thanks, Patrick. Thanks for having me again on the podcast. It's been a very busy week talking about robotics, and I think it's fair to say that AI is getting very, very physical.
Patrick
Okay. So let me give you the easiest question you've got this weekend. What is a humanoid robot? And why does it need to look like a human?
Zornitsa
A humanoid robot is a robot that looks like a human, and it needs to be in human form, because the world is built by, and for humans. And that means that there's so many unstructured tasks that robots need to do and need to execute to be useful and to be and to be scalable.
Think about a manufacturing floor. It has so many different tasks, and it can be a very chaotic environment. There’s the assembly line, there are doorknobs, there's stairs. And all that means that if a humanoid robot is designed well, then it could easily do all these tasks in one go without being programed every single time.
Patrick
Okay, so, you put out this note. It's incredibly well read. But when people hear humanoid robots for the first time, what do you think they tend to get wrong? What's the biggest misconception investors have about this?
Zornitsa
The number one misconception is that humanoid robots are sci fi, that they are a project that only exists in the research lab, but that's absolutely not the case. Humanoid robots are getting more real by the day, and in fact, they're stepping out of the research lab and they're moving into the real economy, into the physical world. The reason why this is happening is that we've seen a lot of improvements, a lot of breakthroughs in AI models, which mean that humanoid robots are now able to interact and contextualize the environment.
And this historically was always a big problem for humanoid robots because they couldn't read context. And in the real world, that really meant comedy gold, like a humanoid robot watering a person instead of a plant, just because it couldn't make the difference between the two. Now all of that is changing, and that is changing the landscape, expanding the use cases and creating investment opportunities as well.
Patrick
Definitely. I'm sure our listeners can find some other comedy gold moments as well. If they, if they Google it. But let's just grind this in the numbers for a moment. How big is the humanoid market today, and how do you see that growing over the next few years?
Zornitsa
I think we're just scratching the surface of what physical AI can do and can achieve. And at the moment, the humanoid market is relatively tiny, about $2-$3 billion. But I could easily see how this is going to change in the next 5 to 10 years. And according to a forecast, the global humanoid market can easily reach $200 billion by 2035, which again, to put in context, is about 30 times higher than where the market is currently valued at.
Patrick
Okay, so you're talking about a growth that could reshape entire industries here. What are the core building blocks? What are the real foundations of the supply chain for humanoids? And how should we kind of contextualize this?
Zornitsa
Humanoid robots are part of a very complex ecosystem, and there's so many moving parts that need to click seamlessly together in order for that technology to really take off and to scale. We looked at the building blocks behind humanoid robots, and we discovered three and we call them the three ‘Bs’ - brains, brawn and batteries.
Brains is the easiest one, as the name suggests, is the brain of the robot, the AI model, the compute, the software stack, brawn on the other hand, is the body of the robot. It's the connection to the real world, to the physical world that is driven by the laws of physics and mechanics. It's the small bolts, the screws, the motion systems, the actuators that enable the robot to move in the physical world and to interact with it. And finally, we all need energy. And this is where the batteries come into the picture. And when we looked at these three Bs, brains, brawn and battery, we found out that the brawn component is responsible for about 50% of the unit production cost, which was quite surprising, at least to me.And I think something that the market has not yet factored in and fully digested.
Patrick
Why do you think that is?
Zornitsa
I think because the focus has been on algorithms and software and the challenge of making robot interactions and reason in the real world. But the reality is that what makes physical AI so different from any other frontier of AI is that we need to make, the link to the physical world, because these robots are going to be put on the factory floor.
They need to be able to move and to interact and that's incredibly challenging to achieve. And this is where the brawn component comes in with the 30% of the total bill cost.
Patrick
Understood. I want to go back to the batteries for a second. I remember reading a book by Matthew Walker called ‘Why We Sleep’. And in essence, humans need to sleep at least seven hours a night. What about robots? You know, how long do they need to be charged for so they can keep going and get back to the back? So how long do they need to be charged for so they can keep going and get back to the factory floor?
Zornitsa
It depends on the specification and the model, but it could be anything between 2 to 3 hours to five, six hours. But I think the key message here is that robots can work 24/7. And I as it to your point, humans need breaks. They need to go on holiday. They need to take a lunch break. They need to grab a sandwich. Humanoids don't have they don't have these issues. They can work all the time, even if at the moment they're a bit less efficient than humans.
Because let's not forget, the human body is incredibly efficient. But humanoids can work, continuously. And this could increase productivity and boost efficiency in some cases up to 1.5 times, which is quite a big number. If you come to think of it.
Patrick
Yeah, I mean, what we are talking about here is just a massively transformative technology, that could impact societies, in the future. Cars transformed the 20th century. You're saying that humanoids could do the same for the 21st century. And, because in the note, you talk about how humanoids are cars in miniature, can you also unpack that for me?
Zornitsa
There are lots of complementarities between cars and robots, which might not be obvious at first glance, but let me give you an example. Take the robotic hand. It is by far the most complex part of the humanoid body, and that's because we need a lot of dexterity in that robotic hand. For us humans, it comes naturally to use our hands to do dexterous work, but to replicate that in a humanoid is incredibly challenging because the technology is quite complicated.
The humanoid hand has about 50 to 60 actuators and actuators. Was basically there's this mechanical component that makes the hand move and twist and turn. So that's what create the degrees of freedom. So basically you'll be able to pick up a glass and not break it, which is quite important. And this means that there are thousands of smaller particles that need to be assembled and put together.
And when I think about this problem, simply conceptually, I think it sounds very much like the challenges that the automotive sector needs to solve and the way motor supply chains are set up, which makes me think that we can use that automotive DNA to, to repurpose some of that technology for humanoids production. But we don't need to start from scratch. We already have the building blocks out there.
Patrick
I wonder how many of our listeners are moving their hands around in all the different planes like I am right now. Okay, so, you know, China's leading the way in terms of physical AI right now. Back on episode five, Kaan Singh was talking about this in China’s five year plan around AI.
But as humanoids move from prototypes to real mass production, does Europe, especially given our automotive expertise here, have a genuine chance to lead in physical AI?
Zornitsa
I think so, and it all comes down to the three beasts, the brains, the brawn, the batteries. So brawn takes about 50% of the unit production cost. So to understand when Europe, where the Europe has an edge or not, we need to figure out where does the brawn capability comes from. And when you map that, you see that Europe lights up because more than 30% of the global supply of high precision components and mechanics comes out of Europe, and 15% comes out of Germany. So there's definitely a lot of DNA embedded in European manufacturing systems that could be used to, to scale humanized production and ultimately give a chance of for Europe to lead on that.
Patrick
Understood. Let's zoom out and think about sectors that are going to be impacted because we talked about supply chain there. In terms of sectors, you've talked a lot about industrials in the note. Do you want to just explain that a bit more. Because as humanoids scale, we're expecting to see them kind of pop up in different places and different sectors at different speeds. Do you want to explain that for our listeners.
Zornitsa
Right. So I think are two components of that. A story about the tailwind for industrials. And the first one is I think it's going to impact companies and sectors that are directly involved in the production of humanoids.
Going back to the three B’s, those that produce, AI model software stocks, the brawn, the batteries, I think these are going to be directly impacted. But also, it's going to be very transformative, I think, for other industries that are looking to deploy humanoid robots on their factory, floors and in their processes. And I could think of manufacturing, logistics, agriculture as the first kind of industries where humanoids can be deployed at scale. And I think this is a natural place to start because the tasks they're more structured, they're easier to learn and to define for humanoid robots, and ultimately down the road when the technology matures and it's more proven. I see a lot of potential in health care and even elderly care.
Patrick
Ok so talk to me about that. You and I sit quite close to each other in the office, and we talked about demographics lots together. How are demographics going to shape the rise of, humanoid robotics?
Zornitsa
Well, quite frankly, I think this is the number one driver and demand for humanoid robots, because the reality is that the world population is aging. By 2050, the share of people aged above 65 is going to double from 10% to 16%. And when you come to think of it, this means fewer workers available to work in factories. Also, humans are increasingly drawn to cities. They don't want to live in rural areas. And as urbanization accelerates, the factories and the manufacturing facilities still remain outside of city boundaries, which creates a bit of a structural mismatch between where the workers are and where we need them to be. And finally, let's not forget that workers preferences are changing and not always necessarily due to pay. And so when I combine the three factors together, the result is that there might be some essential jobs, yet undesirable jobs which humans don't want to take. And I think this is precisely where humanoid robots enter the picture, because they could take on these repetitive, dull, dirty, even potentially dangerous jobs that humans don't want to have. Leaving humans in control, in charge with more capacity to do value adding activity and creative work.
Patrick
So interesting. Zorntisa thanks a lot for joining. Enjoy, Davos - I hope it goes really well. And thanks for coming back onto the podcast.
Zornitsa
Thanks for having me, Patrick. See you soon.
Patrick
It's funny as I mentioned at the start of the podcast when I was growing up, those Terminator films taught us to fear the rise of the machines. But today's conversation shows the real question isn't whether the machines rise - that's happening. It's how thoughtfully we choose to build them. And while the future isn't set, as John Connor reminds us, we do know this.
Firstly, humanoid robots are increasingly well suited for real work in manufacturing and logistics, agriculture, healthcare, eldercare, and even construction.
Secondly, they have the potential to augment labour forces in aging economies and help offset demographic headwinds.And finally, and possibly most interestingly, while China leads in production and deployment today, that could well evolve as humanoids move to mass production.
If you enjoyed the conversation with Zornitsa today, hit subscribe wherever you're listening.
