Encore: This CEO Says Humanoid Robots Are The "Space Race" of Our Time

0:28Find your perfect home on Realtor.com.

0:33hey listeners christopher mims here from bold names we've got something a little different for you this week and last week's episode we talked about one of the big ideas we are paying attention to in the year ahead and that's humanoid robots so we thought it was the perfect time to revisit one of those prescient conversations we had last year yeah that's right this week we're re-upping our conversation with Jeff Cardenas, the CEO and co-founder of Aptronic, which is building the humanoid robot of the future. We hope you enjoy this episode and bold names. We'll be back with something new for you next week.

1:15You were supposed to be on earlier this year, but you canceled on us at the last minute because I think you had a special delivery that arrived. That is correct. Yeah, I had a new baby girl. that entered the world a few weeks early. Congratulations. I mean, so here's the guy who you're trying to replace human workers with robots, but you got a side gig making babies. Does that mean you're hedging your bets? No, I think both will coexist. That's the name of the game. Today on Bold Names, we have Jeff Cardenas. He is the CEO of Austin, Texas, humanoid robot company, Aptronic. And while he may have just added a human to his family, he imagines a future where families will be adding humanoids to their households.

2:04Except something that looks like C-3PO with a glow-up. Yeah, Tim, the thought warms my heart. Adding one of these robots to my own household, it's a long-term goal. They're going to be more like Rosie, though, from the Jetsons.

2:21And I realize not all of you are going to get that reference. She's the housekeeper in that show. Anyway, it doesn't matter. What matters is Aptronic is going to be how he realizes this vision of future home health robots. But it's more than that. He sees these things in industries. He sees them in health care, really driving the U.S. economy and helping the U.S. keep its place in the global order. But it's not so simple. It's a race with China, which he sees as a real rival to building, developing, and deploying this kind of technology. From the Wall Street Journal, I'm Christopher Mims. And I'm Tim Higgins.

3:01This is Bold Names, where you'll hear from the leaders of the bold name companies featured in the pages of the Wall Street Journal. Today we ask, why is Apptronic betting that it will create the home robot helper that everyone will want?

3:18Jeff, welcome. You're the CEO of a robot company. Does your robot have a name? Yes, a robot's name is Apollo. Yeah, and I've seen some videos on the internet of Apollo, and if it was a Tinder situation, I think the stats would be 5 '8", 160 pounds. Interests appear to be moving boxes around a factory. What exactly can Apollo do? Apollo has many interests. Moving boxes is just one of them. So the idea is long term, Apollo can do pretty much anything a person can do today, though we're still in the early days. So very simple things today. So largely we're focused in manufacturing and logistics, things where you're picking and placing, where you need a little more dexterity, but still relatively simple tasks compared to where you'll see these robots go in the future.

4:10So we love getting in the weeds in this show. Before we get too much deeper, though, I got to ask you, if we take a step back, what is your larger motivation? Why did you want to start a robot company in the first place? Well, I'll try to give the short answer for that. But I mean, my primary motivation is to be useful. What does it mean to improve the way we live and work? And where do you focus? So I was very interested in health care, the way we take care of each other, you know, and in particularly in the way we age. So I was very close to both of my granddads. I'd seen both of them age. And one of the sort of observations I made was that as people age, we lost our, they lose their dignity.

4:51So for both of my grandfathers, they, you know, they were war heroes. They were veterans. They never relied on anybody for anything. And in the late years of their lives, they had to rely on people for everything, including, you know, to take them to the bathroom, to do all sorts of things that they would rather not have relied on people to do. And so my dream was, well, what if you could build a robot to do all of these things that humans don't want to do? What would that free us up to do? And I think that's a really interesting question. And it turns out that's been a long journey. We're now over a decade into this.

5:25But that was the motivation is really to make a contribution to the world and to the species and try to push us forward. And so from where Apollo is now, working in factories, beginning to do dexterous manipulation to home health care. That is a long journey. There are many steps along the way. What do you see? What is your projected timeline for that? And what are the milestones along the way? Yeah. So the way I see this rolling out is in in three stages. You can kind of think of it in some ways as easy to hard, though I would say all of this is relatively hard. But stage one is the industrial base.

6:03This is where you can control the environment and where you have experts that can work around these machines. So these are like logistics and manufacturing. Stage two is where we start to introduce robots, let's say to the public. So now robots are working with and around humans and the public, around your kids, around your parents, your grandparents. These are places like retail and healthcare. So that's stage two. Still, this is a commercial setting in my view, where we have tasks that are more continuous, that are not as discreet as where you enter in phase three, where you have the home. And then the holy grail for me is assistive care.

6:41So I would say that the first phase is happening today. The thing I like to point out is that this is basically the personal computer for robotics. I think the humanoid is the general purpose platform that can scale, but it's early. This is the 80s maybe of if you use that analogy. And I think - Yeah, it could also be the 60s though. That's true. I think it depends on my view, spending a lot of time in this, and I sort of openly admit that we were very naive going into this. We frankly didn't know what we were getting ourselves into. I think it is more like the 80s. We do have the systems. We do have the capabilities.

7:21We have early commercial applications that can show tremendous value to our customers. So I think that by 2035, you will see these robots in homes. The question is, how quick is that uptake? And I think that's a point of debate. But I think we're now in the window, and I think you're going to see a very fast acceleration. Just make sure I understand things correctly. You all are essentially designing the robot body, right? One of the big things I've heard you talk about in the past is the importance of owning kind of the technology and the actuators, which are really, as I think about it, as kind of like the muscle of a robot.

7:59Why is that so critical for you to kind of own that? Well, I think, you know, this is a complex system that's very interdisciplinary. And you can think of what a robot is. It's basically a network of motors and sensors. So, and I don't know if this is too deep, but an explanation sort of how robotics... Go deep, go deep. Okay, so how robotics is shifting is traditional robots were called position-controlled robots. So basically I have motors and I'm just going to a place in space. What's happening now for robots - These are what these, you think of these arms and factories that move around. Like in an automotive factory.

8:35Right. That's right. Basically the way these robots work is you say, go to position one, go to position two, go to position three, close your hand. And then you write this sort of step-by-step process and you hit play and the robot just runs this script over and over. The way robots for the age of AI work is we're now having sensors in the feedback loop. So we have cameras, we have force sensing, touch sensing, we have temperature, we have an accelerometer that can sense position and orientation. We have all sorts of sensors that we're bringing together so the robot can be aware and can be sort of dynamic and respond to the environment that's out there.

9:12And so for us, this is the core building block. And this is what drives performance and cost in the robot is the actuators. Yeah. You know, it's interesting. One of the things I like to talk about is the business side. You've already raised something like$400 million this year. And I'm curious, who are some of those big name investors and what are you actually using that money for? Yeah. Google was a dream for us early on. We've been it's been really amazing to work with them and see all the innovations that are happening there. So Google was a big anchor investor. So you're creating this body.

9:45You're creating the muscles. You have kind of like the extended nervous system of the robot in terms of these sensors. But somebody's got to make the brain, right? And that's Google. Is that the idea here? Yes. Today, that's Google. So we've been working deeply with them. We're very well known for building amazing hardware. So we'll build the best platform or work with the best AI researchers in the world at Google DeepMind. And we'll come together and really try to make a contribution to the space. I'm curious, what is going on with AI right now in the developments or the advances that you see that is kind of making this a moment for AI to really be the brains of the body that you're making?

10:22So the simple idea is the robots are learning. So traditionally, what we've done is basically program the robot to do a particular function. This is how it was done in the past. Essentially, it was given the instructions on how to do a task. Correct. And you would build, even if it was AI enabled, you might build an application for a robot that, you know, was a box moving application. The challenge of that approach was every new thing you wanted the robot to do took the same amount of incremental time. So I'm down stacking boxes off of a pallet. Now I want to take those boxes. I want to move them to a conveyor belt.

10:59Each one of those discrete tasks took the same amount of effort, of engineering effort. What's changed, what we call this foundation model, so this is generative AI applied to robotics, is that the robots can now learn. And the interesting thing while you hear a lot of conversations about humanoids is where do you get that data? What are the robots learning from? And what's happening is the robots are learning directly from humans. We can see humans doing tasks in their environments. And if we have a robot that is shaped like a human, it can do the task in a very similar way, learning from a human.

11:34The promise of that is basically now we can very quickly add a whole host of new tasks where anything becomes fair game. And that kind of changes everything about what robots are capable of and where they can be applied in the future. Building AI-powered humanoid robots is no small feat. So how is Jeff dealing with the skeptics? And what makes Aptronic different from the other startups out there? One of the knocks we got when we first started was everyone said, we thought we would have humanoids in the 80s. We thought we would have them in the 90s. We thought we'd have them in the early 2000s.

12:10Like, is this the same thing all over again? Stay with us.

12:23How are the U.S. businesses of Philip Morris International invested in America? We're invested in advancing science, giving adults who smoke better options. We're invested in American manufacturing, helping local economies thrive. We're invested in community, supporting military veterans and their families, disaster relief, and economic empowerment. Because we're proud to be invested in America. See how at USPMI.com.

12:59So I have to say, as much as I would not bet against Google DeepMind, I'm skeptical that the field of AI as a whole is at an inflection point that will allow for widespread, flexible robots that can switch between tasks. What if Google DeepMind is wrong? I mean, is it possible that you have built a viable robot body, but that you are 20 years early? And if so, how are you going to pivot? Yeah, I mean, I think, of course, that's a possibility. And I think it's fair to have skepticism when you're making big claims about this. One of the knocks we got when we first started was everyone said, we thought we would have humanoids in the 80s.

13:45We thought we would have them in the 90s. We thought we'd have them in the early 2000s. Like, is this the same thing all over again? I would argue the answer to that is no, it's different, right? And the interesting thing about robotics is that seeing is believing, right? So show versus tell. We can actually demonstrate our results on a physical robot. And that's how you sort of solve the debate is by proving this stuff, not in the lab and not in a shiny video, well edited video, but actually out in the real world. And just as a quick aside, where are your pilot programs now where you're demonstrating this works in the real world?

14:19So we're still early in these pilots. We're working with folks like GXO and Mercedes, but a handful of others as well. GXO being a logistics company, so that's the warehouse stuff. GXO, logistics company, and then Mercedes. Those are our two deep partnerships that we've been very public about, but we have a handful of others that we're working with. To answer your question about how do we know that they're right, there's an interesting story about Google. So Google bought most of the humanoid companies on the market maybe 10 years ago. So they bought Boston Dynamics. Which they no longer own. They bought and flipped it.

14:52Yeah, they divested them. They bought a company called Mecca, Redwood Robotics. They bought a handful of companies, six or seven different companies. And they were early. The interesting thing about working with Google, we got started in 2016, and I always wanted to work with them. They're the ones that are, this is where all the action is happening. We have to work with Google. But by the time we got to Google, by the time we had enough technology and profile, humanoids was a dirty word at Google. It's like, hey, we just spent hundreds of millions of dollars on humanoids. It was too early. There was a lot of skepticism there.

15:24And it took the course of almost a decade to sort of get back to the point that Google said, OK, we were early last time, but now we're in the window. So when I listen to you and I kind of see that the parts that you're moving around, in a lot of ways, it kind of seems like you're trying to build the car company of robot companies. Really, you talk about the goal of these, the cost of these robots is something like a car, less than$50 ,000. And how far away from that are you? We know how to do sub$50 ,000 now. So, you know, Elon calls this the idiot test, right? Elon Musk, the guy who's got a lot of robot dreams of his own over at Tesla with Optimus, his humanoid robot.

16:05That's right. Yeah. So the cost of the raw materials relative to the cost of a final product. If you compare a humanoid robot to a car, there's 96 % less raw material by weight in a humanoid robot as compared to a car. So the base metal of a humanoid robot, at least our robots, are aluminum. There's only about$300 of raw aluminum in the entire robot. What that means is that from first principles, there's no inherent reason why these robots have to be expensive and why they won't be much cheaper than cars as you build up the supply chain and as you build up volume. But there's no supply chain that exists yet today.

16:44So we have to be we have to build a lot more than you might in the future. And we're not producing these in high enough volume yet where we can really crank that dial and sort of amortize all of the capital costs across a lot of robots. So I don't know if it's just about the cost of the raw materials going in. You're building an intelligent machine. Sure, sure. I mean, and we're building a lot of the components in these robots are things that are being used elsewhere. So, for example, the commercial drone industry is one of the things that drove down the cost of the types of motors that we need.

17:17Electric vehicles are one of the things that drove down the cost of batteries. So now batteries are good enough. So we're basically taking a lot of these component technologies that have been matured in other industries and you're kind of bringing them together into the ultimate form factor. Let's go a little bit deeper in that manufacturing strategy, because in a lot of ways, it sounds like the classic Silicon Valley asset like approach. You are turning to contract manufacturing for these robots, right, to a company called Jabil, which is similar to how Apple does those iPhones with Foxconn. Why is this part of your roadmap?

17:54So right now today, we are designing, assembling the systems here in Austin, Texas, and we're starting to transition more of this to Jabil as we ramp. So basically doing final assembly here in Austin, Texas, and we'll start to do more of that with Jabil over time. So a lot of people like to point out Tesla and they say, Tesla is entirely vertically integrated. Like, how are you guys not going to own manufacturing? How we compete against Tesla? But those folks don't do it from a point of view of actually understanding the history of Tesla. Tesla actually worked with contract manufacturers in the early days.

18:27People talk about the AI stack and they say, you have to own your AI stack or you're not competitive. Tesla didn't own Autopilot in the early days. They worked with Mobileye. So they worked with other experts at particular areas. And that's one of the things that allowed them to play for more and more vertical integration over time. And so the way that I think about building a company like this is you want to choose your battles. What we know how to do very well, what I would say better than anyone else in the world is build amazing robots, build amazing software. So we're really focused on that.

18:59And we want to work with the best manufacturers in the world. So we talk to a lot of different manufacturers and we have a great relationship with Jabil. And so this is where we're starting. It's not necessarily where we're ending over time. I mean, my dream coming out of graduate school was a big factory, robots building robots on the Colorado River right here in Austin, Texas. And, you know, my view is that's something that we have to earn. Right. So we have to get there. And we want to we want to do that in a smart way where we choose our battles. And part I have to imagine, because manufacturing is just a cash intensive business.

19:32Right. I mean, it just seems to eat cash in a way that I think a lot of people don't understand until they get into the deep of it. And you point out a very good, interesting point about Tesla. They began with contract manufacturing. They turned to Lotus to make the Roadster in their early days, but then very quickly realized that they didn't quite control everything in a way that they needed to, and they ran into some trouble with cash and that sort of thing. How confident are you that you can kind of get your fingers into the details and manage this stuff as it starts to become more complex?

20:03We tried to learn from all those lessons. I'm a student of other entrepreneurs and other failures, so we've tried to learn from all of those. Bring in folks who have been successful at doing things like this in the past. But could we be wrong? Sure. Could I come back on the show in a year and say, hey, we tried this strategy and it was painful and our yields were bad and I've decided that we need to be more vertically integrated? Sure. Well, I guess what did you learn? It sounds like you studied Tesla a lot. What did you learn from those early days of their exploration and asset life? Well, not from Tesla directly, but there's many stories of startups raising a bunch of money, building something like a factory that can be a cash furnace if you don't have predictable demand.

20:49right we're still iterating quite a bit on the different models of these robots in the early days and so all of that fixed assets that you have to build to all the personnel that you need early on that that's killed other startups and aerospace and you know the list goes on but you know the the most valuable lessons is the school of hard knocks the lessons you learn along the journey and so you know we're at the front end of this i'm sure we'll learn tough lessons along the way. But right now, this has worked very well. You know, we're in the early stages of this, so we'll see how it develops over the next year.

21:24We reached out to Google, which did not respond to a request for comment. And a quick note, earlier in the conversation, Jeff was referring to Elon Musk's so-called idiot index when he said idiot test. After the break, we find out why Jeff says the stakes are high for his company and the future of the American robotics industry. I mean, what is an economy, right? An economy is like productivity per person. If you change the number of productive units, you change an economy fundamentally. So my view is there's nothing bigger than this, right? Like, what's the key to solving healthcare, to cleaning up our environment, to building the types of goods and services that we need?

22:08That's next.

22:21so 400 million approximately your raise that seems like a lot of money uh given how capital intensive this is i think if you succeed uh at achieving your goals that's gonna feel like a drop in the bucket at some point in the future how much more are you gonna need and when i mean we're gonna need billions of dollars to do what we aim to do right you know we're already getting a lot of inbound. There's a lot of excitement and demand from investors across the world. There's only a handful of companies that my view is that are really competing for this space. In the early days, I believe there would be many different players in the space because the market is so large.

22:58Like, where do you want to start? And will there be niche companies that are focused on healthcare and focused on logistics? And potentially that's how it will play out in the early stages. But I think as you start to get to scale, the amount of capital that's required to do something like this at the scope that we aim to will require billions of dollars. Let's start talking about that competition that you just alluded to. At this point, who is your biggest competitor? I think Tesla and Figure are the folks that we think about consistently. We see Figure out in the market as well. They're also doing parts handling in a factory, I believe.