Investment Insights

Tune in for a discussion on the Robotics Evolution and Revolution with Dr. Henrik I. Christensen, Chancellor Chair of Robot Systems for Qualcomm, Professor of Computer Science at Department of Computer Science and Engineering at University of California San Diego, and Director of the Institute for Contextual Robotics.

Webinar Transcript:

Jeremie Capron:

My name is Jeremie Capron. I'm the director of research here at ROBO Global. As many of you know, we are the creators of research-driven index portfolios that are designed to help investors capture the growth presented by the technology revolution of robotics, automation, and AI. And this is what we'll be talking about today with a very special guest. I'm really thrilled to be joined today by Professor Henrik Christensen. He is the director of robotics at UC San Diego. And Henrik is also the main editor of the US National Robotics Strategy and an advisor to companies, governments, and investors around the world. Henry is also a co-founder of ROBO Global, and he's been helping us steer our investment research efforts over the past seven years. And for that, we're very grateful. Henrik, thank you, and welcome.

Henrik Christensen:

Thank you.

Jeremie Capron:

So, today we're going to discuss some of the most exciting areas of robotics, automation, AI, and how investors can capitalize on these trends. This is a very important topic to us, to investors. Robotics and AI, we see as a set of general-purpose technologies that can be applied to every industry very much like the internet or electricity at the beginning of the 20th century, and this is offering tremendous investment opportunities. So we're very fortunate here at ROBO global to be working with some of the most prominent thought leaders in robotics and AI; academics and entrepreneurs like Henrik and several others that you can see here on a strategic advisory board. So we're really proud to combine a financial and investment analysis approach with technology research and expertise that's coming from people like Henrik. And today there's more than $4 billion in funds tracking our strategies. Here on the New York Stock Exchange, as well as in Europe and in Asia.

And the most notable is probably ROBO. ROBO was the first robotics and automation ETF. It launched around seven years ago. It's returned more than 50% over the 12 months to the end of the second quarter, and an annualized 22% over the past five years. So, we also run THNQ. That's the artificial intelligence index, which we launched three years ago, and Henrik has been instrumental in helping us research and understand the world of AI machine learning. And HTEC; H-T-E-C, that's the healthcare technology and innovation index, and you can see the three strategies have significantly outperformed global equities since inception. You can find more information on our website at roboglobal.com. So with that, I would like to invite Henrik to the virtual stage. Give us a brief presentation on robotics evolution and opportunities. And for all participants, we'll be addressing your questions after that. So please use the Q and A function at the bottom of your screen to log your questions. All right, Henrik.

Henrik Christensen:

Hey, thank you, Jeremie. Let me see if I can get the sharing going here. There we go. So, thank you, Jeremie. Thank you for this opportunity to also talk about some of the things I'm very excited about. So, one of the things I've been doing, as Jeremie mentioned, is that I'm the main editor of the US National Robotics Roadmap, which we publish every four years. So 2009, it was, of course, the economic downturn. It was collaborative robotics that was the main stage. 2013, after the terrible disaster in Fukushima, it was robots for rescue and those kinds of things. 2016, it was globalization. How do we then get high mix, low volume? And now the roadmap that's just been published, we're looking at what is the post-pandemic world that we're looking at. So we look at some of the mega trends.

It is how do we get to a society where we are post COVID? What are the impacts there? We're seeing still changing trade dynamics. We're seeing mass customization, questions about urbanization; are we all moving back to the big cities, or are we going to stay away in sort of a work-from-home kind of scenario? And of course, the aging scenario or aging society is still a pretty big deal. If we look at some of the COVID implications, it's very clear that logistics' last-mile deliveries, those kinds of aspects, are playing a big role. It's grown tremendously. We had 18 months or something like that, where most people would take home deliveries rather than going to grocery stores. We saw somewhere between 40 and 60% growth in last-mile logistics. That will continue. We see that staying on. But we also seeing Tele-robotics, where there's a lot of interest for doctors and others to not have to see patients face to face, but still conducting virtual visits.

We're seeing, of course, screening. We're still doing COVID testing on a very large scale. We're seeing food safety. How can we guarantee that some food was not touched by humans? Cleaning, disinfection. And of course, we also seeing a lack of a migrant workforce. It used to be that you would have a lot of people moving around because of COVID and because of other aspects, we don't have that same size of a migrant workforce, and that impacts how we would do things, and it motivates a higher degree of automation than we've seen before. In terms of the trade dynamics, we are seeing manufacturing moving in areas like automotive. 30% of the cars are still made in China, but we are seeing a movement towards India, towards Vietnam, towards other areas that are impacting this. Salary is still a big factor, and for that reason, some of that manufacturing is not moving back to the US, and that motivates a higher degree of using automation as we see a shuffle, just like we did in 2009.

So that's important to keep in mind, but clearly, the US-China is impacting how we look at this for the future. Some will come back to the US. Some will go elsewhere. Mass customization is and will continue to be a major trend. Everybody wants a one-off. So if you look at BMW X5 that you're seeing in this slide here, it's available in four million different configurations. Everybody wants their own kind of seats, their own kinds of audio system, their own different kinds of engines. All of that implies that these cars are available in four million different configurations. We only make the same car every two and a half week, which imply that you need to have a highly streamlined supply chain and manufacturing process, and you can only really do this if you have a very high degree of automation.

We're seeing the same thing for individual products like sneakers. Everybody wants their own color, their own customization on it. 3D printing and other technologies are enabling us to do this to a degree we've never seen before, which is really cool. My long-term vision is still that you'll have an autonomous UPS truck driving around in your neighborhood, and it will be 3D printing a product for you. And so you order it, five minutes later, we'll ring on the door and we'll say, here you go. Aging society is still playing into this post COVID or pre COVID. We had eight hours of increase in the age of the world population every day. We've seen a setback by two years during COVID, so the overall average age of the world population went down by two years.

This is starting to pick up again. So again, you should expect to see eight hours a day. This implies that we will live longer. Unfortunately, we will get various kinds of challenges as we grow older, both in terms of mobility and in terms of other things that will come in and impact this. We not only have societal drivers, but we also have technology drivers. So, new materials, as an example, are fundamentally changing how we do this. We can build grippers that are fully customized to products. So, as an example, if you do strawberry picking, we can actually build a gripper that morning that can be used for the size of strawberries we're going to pick today. And tomorrow, it might be a different size for strawberries. We can literally customize it to this using a combination of beta signs, 3D materials, and soon we will also have a much better hybrid sensing building, which fundamentally changes how we do this.

So we can do print to order, and we can add sensors that we've never seen before. The same thing with sensors, we've seen camera technology coming out. Here, the gray area is traditional old-style cameras, blue is your standard travel cameras, Orange is your cell phone technology. So, you're having more of these come out that we produced any other kind of camera. I live in San Diego. So Qualcomm would tell me that you can buy cameras for these in volume at $1, and you can buy the processing power to do it for $9. So suddenly we are at a level where we can use technologies that we've never imagined before, which is really cool. We also seeing computing. Computing is becoming very cheap, so we can now do trillions of operations per second, including deep learning, and all of this, we can integrate with eight to 10 cameras into a single processor, and get that processing power.

So we getting to a point where we've commoditized computing, AI, and deep learning to a level that we have never imagined before, which is really cool. The fact that I can do all of this and put 10 bucks at volume makes a difference in how we think about things. Some other societal drivers, if we go briefly through some of the verticals, will give you a sense of where we're going. So in manufacturing, it's still cost, it's still flexibility, but it's also this that we can do very high volume but also high mix so that we can get almost one-off manufacturing. Same time we get into where we have codeless programming. So you can imagine you can more or less show up and you can use a very simple interface to be able to do very easy programming of this.

It used to be that it's required you to have weeks of training to be able to operate a robot. Here, you're sort of seeing an iPad-like interface that makes it very simple for people to actually go in and do programming of this. So we're seeing READY Robotic, we're seeing Intrinsic, we're seeing a number of these companies that are coming out with very simple interfaces that radically change the cost of deploying these systems in everyday scenarios. Logistics driven by Alibaba, Baidu, Amazon, or Walmart, but also driven by the fact that e-commerce is really driving this forward. You will see much more flexible distribution centers than you've ever seen before that allow us to go in, put material and really transport it to you in a minimum amount of time.

If you go and ask UPS, UPS will say, when they send the package to you, three hours of the time, it sits still, the remaining amount of the time, it is actually moving towards your destination. So the fact that we getting this level of agility makes a huge difference for us. And especially as we see a change in a post-COVID world towards much more e-commerce, much more delivery of food, whatever materials, these are going to be the enabling technologies to make that possible. Healthcare, we also see, where there's a lot of interest in individualizing it, making sure we can do it. How can we make sure the patients, the families have access, the integration? Here, we have sort of the interblade between robotics, AI and healthcare technologies that we're seeing in ROBO in terms of how we can do this. Because we see an aging population, this is not only going to be what you see in the hospitals, but it's also going to be, what can you do in the home? How can you do rehabilitation? How can you help people that are isolated to feel less isolated than they feel today?

So here's an example. When COVID started, a number of doctors showed up and said, I don't want to be right next to a patient because I'm worried about potential COVID infection. So having a teleoperated robot where you can stand even 10 feet away and do a teleconsultation with your patient would make a huge difference. So we've seen a fundamental change in technology is here to give remote access, and remote access could be anywhere from 10 feet to hundreds of miles, where you provide that level of interface. This is in place now. We're seeing the same thing for being able to do remote surgeries. We've done experiments recently where you can be on an aircraft carrier in the middle of the Pacific, and a surgeon can dial in and assist with doing surgery for this without having to actually be at the aircraft carrier.

So we're revolutionizing healthcare and access to healthcare at a level we have not really seen before. We're seeing new kinds of mechanisms. Digit is a robot that was developed in Oregon that... So the problem is with a lot of the last-mile deliveries, you can only get to the staircase. We are seeing innovation in new mechanisms that can walk up the staircase that can get you all the way to the front door. This is a technology that is being tested by multiple companies right now to actually make this possible. It's not only a question of, can I make it possible, but it's also a question of, can I make it accessible in terms of economic costs? And here, we're seeing these robots coming out and doing this in a way that is very, very flexible. So you will see these new technologies.

One of the areas where we've been missing is in terms of grasping and manipulation. It's been very hard, if you think about it, at your fingertips, you have about 300 mechanoreceptors that allow you to have very agile control. Until recently, that was difficult to do. With new materials, we can now actually go in and build very integrated gripper technology that will allow us to do manipulation as we've never done before. That implies as I mentioned early on, we can do the picking of strawberries. We can handle grapes in vineyards. So it's all of these that's coming out, and at the same time, we come at it at a cost that is accessible to even relatively small-sized companies. In terms of sensing and perception, we've seen the evolution of AI and machine learning that has enabled us to get much, much further than we've done before.

We can use this, as you see in the top image, to go in and recognize all of the things in a living room or in whatever area. So, whether you're building a smart vacuum cleaner that knows, oh, wait a minute, I'm going under the couch, this is where all the dust is, we can build in that kind of knowledge, but we can also use it to do autonomous driving cars so that we can drive coast to coast without driver intervention. The same thing, we can do monitoring of people, so we can do tracking of them. We can do biometrics on this. So it's interesting that we've seen this very cheap camera technology, very cheap computing power with a combination of AI that's going to revolutionize how we build intelligent systems for the future. We've also seen sort of it used to be that it was very hard to do programming of these systems in a reasonable way.

We are now getting to codeless programming where you don't have to write code in a traditional fashion, instead, you can go in and move around a few blocks. And by doing this, you can completely reprogram a manufacturing line or a supply chain line to do this. That's very interesting because it enables us to build a paradigm where you need very limited training and you can still get to a very good place of doing this. So, it used to be that our rule of thumb was that for a system being deployed, 25% of the cost was the robot, 25% of the cost was for the hardware, and then half of the cost was actually software. If we can take that 50% of the cost and reduce it to something like 10%, the cost of the overall system goes down by 40%, and that opens up a number of markets we've never seen before. But it also implies that we can have people that are largely unskilled labor that can design and program these systems.

So this is really changing how we enter the market and why we have seen such a tremendous amount of growth that we've not seen before. Of course, I would be amiss if I also didn't mention that AI, machine learning, and deep learning is revolutionizing anywhere from automating electrical processes, but all the way down to doing logistics, to doing manufacturing. We can do much better analytics. We can individualize it. We can set up these systems. We can program them to have a very high degree of agility, which we haven't really seen before. So this is an area where we will continue to see various significant growth. It used to be that it was mainly some of the big, the FAANG companies that were doing this, but now we are seeing this getting commoditized down to a level where everybody is using it.

Same time we're getting to a point where we are using standardized or sort of pre-trained systems to do this, so you don't have to spend days doing training. We can talk for about minutes or seconds and still get very high performance. So in summary, the world is really adopting robots. We're doing this to increase the quality of life. So we've seen these people want to remain autonomous. Throughout this, we want to be able to help with healthcare technology to make sure we can address whatever the challenges are. We increase the flexibility so that we can get to the e-commerce, we can get to millions of different configurations of various products, and at the same time, enables us then to use this to do food, addressing climate challenges, addressing economic growth, and democratizing it, so it becomes available for everybody.

And that's why we continue to see overall growth for the entire sector of 15 to 18% in terms of adoption, in terms of industry statistics, and that then, in turn, we see very significant growth for a number of different companies. And with that, I will hand it back to Jeremie.

Jeremie Capron:

Thank you very much, Henrik. Let me ask you a few questions while we wait for the participants' questions. I'll remind everybody, you can use the Q and A box at the bottom. I see we have a few already coming in, but let me kick it off with... Well, you're working with a lot of students, entrepreneurs, governments, big corporations around the world. I know you're just back from Europe. Let me ask you this, what's the most exciting project that you're involved with today or in the recent past?

Henrik Christensen:

So for me, the most exciting is that we built vehicles for doing micro-mobility, where we can do package deliveries in very crowded environments. So the fact that I could drive one of my autonomous driving vehicles on fifth avenue in New York, and able to actually manage to go through all of the pedestrians and do this in a way that is appropriate. The fact that we've gotten to a point where I can make this cheap enough, I can make it accessible enough, and I can do it in a way where people can drive through it, I think, is credibly interesting. So that's sort of one. So doing the micro-mobility. I think the other one was the project that I mentioned at our research institute we now have so that we can do teleoperation, teleoperated surgery on an aircraft carrier in the middle of the Pacific. This opens up very interesting opportunities to do this. So these are two examples where I feel we're really changing the world.

Jeremie Capron:

Right. Yeah. Henrik, we're very excited about the deployment of autonomous systems. We've seen them in manufacturing. Now we're seeing them in logistics with autonomous mobile robots, moving goods and parts in warehouses, and now you're talking about micro-mobility and actual vehicles on the road in public areas. We've seen robo taxis going commercial with values degrees of success, but it's here. In fact, we're also looking at a company that makes autonomous indoor drones that can fly in swarms. For logistics applications, what do you think is the current state of the art for autonomous systems, and where do you see the most commercial appeal?

Henrik Christensen:

I think I'll break it into two because there's the indoor and there's the outdoor. And for the outdoor, we are now starting to see coast to coast autonomous driving with these vehicles, where you get on the highway. So it’s sort of a ramp to ramp autonomy. A lot of this is actually not driven by, can I get the driver out of the cabin. It's driven by that we see five to 10% fuel savings. So cars are actually better at driving, or computers are better at driving than people are. I can do coasting. I can predict what the train looks like. I can predict the traffic. And if you look at this from, if you are some of the big carriers, 5-10% fuel savings is a tremendous amount of money. So there, I think we're seeing this, and I think we'll see a lot of them, the technology being matured on the interstate for ramp to ramp autonomy before we really get into the cities. It's incredibly hard to do driving in the middle of this.

So I'm worried about when certain companies promise we will have level five autonomous cars and we will have robo taxis. We're not quite there yet. It'll take a little bit of time. So that's why I think the interstate will be where we're doing the testing. Indoor for autonomy, we will see fully autonomous warehouses that are doing this, but also we will see as you mentioned, places where you use drones, for instance, to do inventory management, where you can go in and you can monitor all of the things that are in your warehouse using drones because it's much easier than going and using a staircase to climb up and down the shelves to figure out what's on the shelves. So I think we will see indoor touchless less monitoring and management of warehouses, and outdoor, it's going to be ramp to ramp autonomy that's going to drive the economy.

Jeremie Capron:

Okay. Well, I see we have a couple of questions around the impact of the pandemic, and I think clearly the pandemic has turbocharged the digitization of the economy. It became very apparent last year. Most companies we research and talk to the members of the ROBO index. They tell us that automation demand is increased significantly, and now the financial markets certainly reflect that to some extent, right? The ROBO index portfolio was up 45% last year. The AI portfolio was up in the sixties, and they continue strong into 2021. So to me, this is somewhat similar to 2017. The last time we were hearing from automation companies that demand was extremely strong. What do you think is happening here and how sustainable is that new trajectory?

Henrik Christensen:

Well, so I think the two different questions in there. So in terms of the post-COVID logistics automation, I think people have realized, why do I need to go to the grocery store for a lot of these everyday purchases. I can go in, I can get it delivered. If you are a parent with two kids that you otherwise would have to pick up in daycare, go through the grocery store, where you are trying to do crowd management of your family while you're also trying to do this is incredibly stressful. You realize I don't have to do this anymore. I can go home, I can order it. And somebody will show up at my doorstep, and that fundamentally changes it. So I think we've certainly if I look at my own family, we are in a point where we go to the grocery store less than half of what we did before because you just go in and you put it on the list of what do we need.

And then eventually at the end of the day, you press please deliver tomorrow before noon, and it just happens. This gives us a lot of flexibility that we didn't have before. I don't have to worry about which store do I have to go to. It gets delivered. So I think this is here to stay. It gives us a level of flexibility we've never imagined before. I don't see people going back anytime soon. There are still people that want to go in and say, no, I want the latest, I want the freshest vegetables and stuff like that. That is 10 to 15% of the customer base. It's not the majority.

So the majority will do this. And the fact that you can get it to your front door makes a tremendous difference. In terms of the digitization of the economy, which was the other part of the question, is that we are now at a point where, first of all, because of sensing, we can track every product. All the way, I can tell where it is from the time the strawberries were picked in San Luis Obispo until they get delivered to my front door.

This enables me to do much better analytics. I understand where's the loss. Where is it? I can track it. Is it fresh? How fresh is it? So all of this, and the same thing for the supply chain management. So the example I mentioned before, in a standard automotive factory today, storage capacity is 12 minutes. Stuff arrives at the factory, 12 minutes later, it's mounted in the car and it's on its way out. So we've gotten to this where we used to have big warehouses, a lot of storage capacity built into the system. Those days are gone. We can use AI, we can use analytics, we can do predictive control to predict, what do we need for Thanksgiving? What do we need for Valentines?

And all of this allows us to do much better management of our capacity. Of course, then the challenge is we've made these systems incredibly lean. So if you see a challenge like the Suez Channel, suddenly our system is, uh-oh, we didn't build in storage capacity to handle this because... So our system is still not as robust as you would like, but overall, I think we made these systems incredibly lean through AI, through automation. It will stay with us, but we still have to figure out how do we robustify these systems for big challenges like the Suez Channel.

Jeremie Capron:

Henrik, we have some questions around the labor impact and there's a labor shortage right now in the U S certainly coming out of the pandemic. Where do you see robots helping solve that problem, and longer-term, how do you see the impact on the workforce?

Henrik Christensen:

So, we did sort of an analysis for the roadmap where we looked at this, and there's almost one correlation between investments in robotics and growth in jobs, and it's not causation. It's not like if you invest in robotics, you also employ more people, but there's a strong correlation which has to do with stronger economies. So I think it will be the dirty dull and dangerous jobs that are getting replaced my labor. If you look at the standard warehouse, the turnover is somewhere around three people a year. People don't want to work in these warehouses. There's no way you see, it's a hundred degrees, the humidity is 90. This is not where people want to be. So we will see a number of these jobs will get replaced. And at the same time, it will give us much higher agility in terms of managing these processes.

I can build warehouses in my local neighborhood. So it's a lot of these process that we will see, where we'll get in. And the goodness is that allows us to displace people to the areas where you really need the human touch. I'm still surprised that today... So the other thing I do, I'm a food connoisseur. I love to go to restaurants and stuff like that. Many restaurants are still only at 50% capacity because they can get people. So, we need to go away from the places where we really don't need people and put them in places where we really need people.

And so we will see a shift in workforce around this. I'm not seeing at least in the short term, a significant reduction in the need for workforce, it's going to continue to grow, but it is going to be the very traditional, highly repetitive jobs in tough work environments that will go away. This is where we do the replacement. That's true for India. It's true for China. It's true for the US. So this is a global effort that we will see. But we've done many studies and so far, we've not been able to show that at a macro scale, robots are replacing jobs. It's shifting jobs to new domains, but it's not eliminating jobs.

Jeremie Capron:

Understood. I see some questions around AI coming in. As you know, Henrik, we think AI is really one of those key enabling technologies that are expanding the scope of potential applications for robotics. And it's really an important part of our investment strategy, both for the ROBO and bank portfolios. So, a lot has been done around computer vision and the recognition around that in digital pictures and then the natural language processing we've made tremendous progress. Now, AI is pushing into other broader ranges of applications. Where do you see AI being deployed successfully? Where is it going? I see, we have a question about the five, to 10-year horizon. Maybe we don't need to go that far out, but where is it going?

Henrik Christensen:

So for me, it's, we've seen a tremendous amount of progress on… most of the progress we've seen on AI has been offline. The fact that I can do face recognition in my image library, the fact that I can do a lot of… but it's largely been where I could do a lot of training ahead of time. We're now moving to a place where I can do a lot of training on the fly, and doing this so we can really get to real-time AI. And that would imply that it'll help me take better pictures on my phone. So if you take… I was very impressed a few years ago, I went to Huawei’s factory in Shenzhen, and I said, that's a lot of processing power. And then it's like, oh yeah, whenever you take a portrait with your camera, we will take 60 images.

And we will only return the one where you smile. So, as an example, this is a camera that takes amazing pictures. Yeah. Because we use AI to throw away the 59 pictures that we don't think you would like. So this is an example where we're getting from image processing, from doing this… I flew back to, as you said, from Europe recently, and Atlanta airport, they do face recognition of you. You don't have to show your ticket anymore. This is where we're getting all of this. The user interface will fundamentally change how we do this. So that's one area of AI, we’ll do much better analytics than we've done before. And then, of course, we're getting to where we can do natural language processing. So where we can do translation, we can enable it.

Now, I used to turn on grammar checks, blah, blah, blah, in words, to make sure that I actually wrote a reasonable set of texts today. I have all of these AI tools that said, oh, you are aware, you should have a comment here, and you should really change this to this other word. So in terms of NLP, we will all become prolific, amazing writers. Even though naturally I might not be a prolific, amazing writer. So we're seeing this, it's enabled us in our daily lives to compensate for our limitations and bringing them forward. So I think that's nice.  The third area is in healthcare. We can do much better diagnostics. We can monitor you. The scary part is also, I have all of these technologies that monitor me on a daily basis that will tell me, are you working out hard enough?

Are you exerting yourself too much? Are you aware your temperature just went up 0.5 degrees? Maybe you should take it easy today. So we will get this massive amount of data, and we've already had that massive amount of data, but now we can process it and provide decision support to people. Whether it's for your daily exercise routine, it's for food it's for your business processes. I would say we've only touched the first 5% of AI. So this is an area where you're going to see massive growth going forward. The same for TV recommendations, nobody sits down and does the schedule and says, what's on at eight o'clock? You go and say, hey I want to watch a comedy right now or...

So we are getting to where we are much more personalized in anything we do, and it's being enabled by AI. And it's because it's becoming so cheap. As I said with my camera, I can now get to where I can do this for 10 bucks. And that implies that you can do it everywhere for your burglary system, for your TV entertainment, for your food prep, that changes it. So I think that's why we're very excited about the THNQ portfolio because it opens up for technologies that will impact all aspects of your daily life.

Jeremie Capron:

And following up on that, Henrik, I see some questions around the software side, the hardware side of the robotics industry. What are the enabling technologies that matter the most to the continued expansion of robotics? We have a participant asking if it's more about the software and the algorithms or the machine itself and the parts. I'm curious to hear, where do you see the opportunities in terms of enabling technologies?

Henrik Christen...: Sure. I've been quoted before in terms of saying that I feel that the last 50 years of robotics have been dominated by mechanical engineers. It's been about building mechanisms, building them so that you have high precision, you have a high lift, you have high efficiency. All of these things are now in the software. And I think the enabling part has been, we've got two sensors that are sort of robust enough and cheap enough that we can instrument everything. At the same time, we'd gotten to the amount of processing power, and there it's both the CPU, but also having access to the graphical processing units, the GPUs that can do massive amounts of data processing in very limited time. So these tools have now been commoditized and made small enough that you can put them everywhere. And now it's the software. At the same time with the software, it used to be incredibly complicated to program these systems.

Now we're getting to where through machine learning, we can do data-driven extraction of key parameters. So if you look at some of the AI-based companies, they will do pre-training so that you're 90% at a solution before you show up at the customer. And then you show up at the customer and you do the last adjustments, and you could do this, it used to be days or weeks. Now you can do this in seconds and minutes. Come and adjust to the particular line. So we're seeing this, so I'm sure some of you saw that Google just spun out intrinsic, which is trying to go after a way of actually doing this. We've seen co-variant, we've seen companies like Berkshire Grey that are trying to do these things. So a number of these companies that are coming out that allow you to very rapidly stand up a system and get it going.

And so we have now the sensors, we have the computing power, but also we have software tools that make it accessible to non-PhD people who actually do programming of these systems and doing it a very limited amount of time. So before I said 25, 25, 50, in terms of cost, that 50% software cost has come down to something that is much lower and with a much higher degree of agility. So, that fundamentally changes it. Still, we need better tools. But the other thing is that we're getting very large-scale data sets that allow us to do a lot of the training before. Before it was hard. Somebody came to me and said, we're going to use a petabyte of data to train your system. I would have been, this is really scary. Today, put them on video, sure, not a big deal. But it will take us a few hours to process it, but 10 years ago that would have been intractable.

So the question is, how are we... the next thing is, I think where we will see the next revolution is in hardware-software co-design. Where we will get to a point where you show me the overall needs of your application. And I optimize it for what should I do in hardware? What should it do in software? And so five years out, hardware, software co-design, and that's where some of the companies we've seen less growth from. So AMD, Intel, all of these companies will come back because this will open up entirely new ways. We've seen growth that we haven't seen recently.

Jeremie Capron:

I think there's an important question here around the environmental impacts and how those technologies can perhaps help. And here at ROBO Global, this is a very important topic to us. Not only environmental but social and governance aspects. We pay a lot of attention to those factors when we design those portfolios. I think in general, there's a natural tendency for robotics and automation to have a positive impact in terms of the environment, just given that they're focused on productivity and efficiency and using less energy and natural resource. But on the other hand, you also have this dramatic increase in compute power and we're using more and more chips and perhaps some different types of materials that could be harder to recycle. So we have a question from a participant about AI helping with climate change. What are your thoughts around that, Henrik?

Henrik Christensen:

So I think, there's quite a number of different aspects to it. So I mentioned before, driving coast to coast in trucks, if you do autonomous driving, we can reduce the fuel consumption by 5% percent. That's a pretty big deal that we can actually go in and do this. Computers are much better at anticipating traffic, anticipating when can I do coasting? When can I not do coasting? So that's an example where we see it, but also we are seeing it in terms of being able to do manufacturing on demand. So it's not like if you're harvesting fruits or whatever, that you're worried about, am I actually building stuff or producing stuff that is not needed, which would impact in terms of, I live in California, so I care deeply about water. So being able to do this.

So I think we will get to a lean manufacturing system where we save water. We only apply pesticides where we really need pesticides. So all of this will allow us to build a much more sustainable economy. The downside has been, for instance for the deep learning, that some of the very large scale, for instance, natural language processors takes more CPU power to train than 100 flights across the Atlantic. So that's pretty scary that some of these training… you take a super-computing center at one of the big companies, and you run it for three days. That generates as much CO2 output as flying across the Atlantic 100 times. So there we are now getting to where we can be much smarter about how we do training, but also we can stage it so we can do pre-training, which can be used again and again.

So we used to do group force applications of computing power, but we have now recognized that unless we're very careful about how we use these systems, this could have a negative environmental impact. So if you go and ask the big companies, the first question is, what is this going to do for the climate? The same thing we will see for all the things I think we can optimize processes to be much more environmentally friendly than what we've seen before because we can really make them lean. We can really make them optimize whether it's driving or manufacturing. We can optimize it for materials to do much better recycling than we've done before. We can sort recycling much better than we've done before.

And then finally, we're getting to the place where we apply AI in a stage fashion to make sure that it does not have a negative impact on the environment. So I think you should expect to see this all over the place, but also you should expect that this is something we see much more emphasis on. 10 years ago people were like, yeah, in a moment. Now, this is question number one or number two, when we talk about how do we evaluate processes?

Jeremie Capron: 

So, we're running a little bit out of time, but I want to address a couple more of the questions coming in Henrik, especially around China. Here at ROBO, we've chosen to include in our portfolios, some technology and market leaders in China. And China has been the fastest-growing market for automation for some time now and recently received some of the local champions displace foreign leaders. What do you think of the US-China technology race, and perhaps some of the implications we should consider as investors?

Henrik Christensen:

Well, so I'll give you multiple answers to it. So the first is, of course, as ROBO global, we have not only a US investment vehicle, we have, the index is Hong Kong, Singapore, Australia, London, you name it, we're all over the place. So for us, we can’t afford to only think about this from a US perspective. We have to think about this from a global perspective. The fastest-growing market for AI or for robotics today is China. So I mentioned before 30% of all cars manufactured worldwide are made in China. And most people might not realize this. The US is not even in the top five for manufacturing cars. It's China, it's Japan, it's India, it's Germany and it's Korea. And then number six is the US. So we need to be careful. And in cars, in particular, there's a tremendous adoption of robotics technology to make this possible.

So we see year over year continued 50% plus growth in China. And we're seeing, Midea acquired KUKA. They're the third-largest robotics company in the world. That is still going to be a very important part. We will see some of that move out of China because of the US… China's other trade complex, or we'll see some of it move out. I'm not convinced it's going to move back to the US. I think it will move to Vietnam. It will move to India. India just announced this big initiative of, make it in India. The domestic market in India is as large as the Chinese market. It's not the US. So we will see a diffusion of technology going over this. But at the same time, we have to be aware that with these very large domestic markets, we will see technology.

Also in AI, we're seeing Baidu, we're seeing Alibaba. We're seeing a number of these companies that are very strong global players in this market. And I think at the end of the day, our responsibility to our investors is to make sure that we return maximum capital to them. We want to have this and we can’t afford to be, at least from my point of view, caught up too much in the political climate between the US and China. We have to look at this and we will continue to see this. So in China, the salaries have gone up 350% over the last 10 years, and that has pushed for a much larger adoption of technologies. It is now more expensive to do manufacturing in China than it is to do in Mexico. Just to give you a data point, but still 30% of the automotive market.

We see other things, textiles, all of these areas are in China. The supply chain is in China. For that reason, they are automating. They are doing the data science that we've talked about to a degree that we've not seen anywhere else. This is an opportunity. That's why you see ABB robotics headquarters in China, KUKA Midea is there. We're seeing CSR, we're seeing a number of these companies that are doing investments there because this is a very interesting market. But I think you will slowly see a diffusion. We will see it pick up in India, we've seen it already in Vietnam and in other areas. So we have to think about this as an international landscape and see, where are the growth opportunities? But you should expect to see precision gears, motors, CPU power, all of that will have a tremendous amount of growth in China for the next five to 10 years, at least. And we need to at least keep an eye on this to make sure we're making the right investment decisions.

Jeremie Capron:

Right. Well, as I said, we do include the Chinese technology and market leaders in our portfolios. And so far it's been a good place to be despite the recent concerns around the internet giants in China getting under a lot more scrutiny from the Chinese government that, thankfully that hasn't really affected the robotics and automation companies over there.

Henrik Christensen:

At the same time, the best autonomous car I've driven, I drove in China. So Huawei took me on a tour showing their autonomous driving cars outside of Shenzhen, where they were using a combination of sensors, CPU, and 5G. And it was just unbelievable. I haven't seen a demo like this anywhere else in the world. So, we need to keep an eye on this to make sure that we really understand who's going to win this race. And I think ignoring China would be naive. We need to keep an eye on where are those technologies and who do we think are going to be the future technology leaders.

Jeremie Capron:

Okay. Well, I think it's time to wrap up here. I want to remind everyone that if you want to learn more about investing in robotics, automation, and AI, please visit our website, roboglobal.com, where we share some of our research on companies in the ROBO, THNQ, and HTEC portfolios. You can sign up for our newsletter. We'll give you a biweekly email with some of the research highlights. Thank you very much, Henrik, for sharing with us today. I know you're on a pretty tight schedule. I want to thank also all of you who joined us on this call today, and we look forward to speaking with you again very soon. Thank you all goodbye.

Henrik Christensen:

Thanks Jeremie.