Recently, Brett Adcock, founder and CEO of humanoid robot company Figure AI , had a conversation with Amit Kukreja, a podcast host on technology and investment, at Figure AI's headquarters. Brett Adcock mentioned in the interview that after the first customer , Figure AI has received an order from the second customer, and will deliver tens of thousands of humanoid robots to these two customers in the next four years. In the first half of last year, Figure raised $675 million in its Series B financing, with a valuation of $2.6 billion, and investors include Microsoft, Nvidia, OpenAI and other institutions.
Not long ago, Brett Adcock said at X that Figure will bid farewell to its cooperation with OpenAI and will build Figure's robot AI end-to-end within the company. In the interview, he also emphasized why he wanted to do this: first, large language models actually lack robot data in physical space, and thus cannot truly guide every action of the robot; second, the training of robot AI needs to rely on excellent hardware and data generated in real physical scenarios.
Therefore, Adcock believes that the key to the success of humanoid robots is to build an integrated company from AI (software) to hardware, so as to truly create reliable, intelligent and easy-to-use products. On this basis, Figure will also vertically integrate the manufacturing plant. In a sense, Figure is like a robot version of Tesla (vertical integration, AI and manufacturing).
As for the industry structure, Adcock believes that the industry will eventually converge to about three dominant companies, but the potential market size of the industry can reach trillions of dollars.
Adcock is a typical "serial entrepreneur" who has successfully founded two companies, Vettery and Archer Aviation (ACHR.US). Vettery is a talent recruitment platform that was eventually acquired for $100 million; Archer is an eVTOL company that went public at the end of 2020. In addition, he also founded Cover, a campus security company in 2023.
01
Serial entrepreneurs are now solving the most difficult problems
Amit: Brett Adcock, founder and CEO of Figure, thanks for having me here and having a discussion.
Brett: So glad you’re here with us, Amit.
Amit: If people look at your career, it’s very interesting. Vettery, a talent recruitment marketplace, exited with a $100 million acquisition, Archer is a hard tech company that went public, and now you want to take over the world with robots. I think there’s one word that ties all three together, and that’s ambition. Where does that ambition come from?
Brett: In short, I’ve been building tech companies for 20 years. You know, it’s fun and addictive to build something that’s actually useful to the world. Now, I’ve reached a point in my career where I can focus on solving some of the world’s hardest problems. It’s like a dopamine hit for me, and it’s what makes me wake up every morning excited, and it’s what I plan to dedicate the rest of my life to.
Amit: Not everyone can wake up in the morning and say, they even have a clear goal to say we’re going to build robots, we’re going to build flying cars, we’re going to build a growing recruitment marketplace with network effects…Dopamine is one thing, but is there a bigger drive that drives you to make an impact that drives all of this?
Brett: I think we can do something and build something that can actually be successful. For me, what I really want is to finally be able to live in a world that is exciting and inspiring. I want to build something that is truly important to the world with companies like Vettery, Archer, and Figure.
Amit: When did you decide to become an entrepreneur?
Brett: I grew up in a three-generation farm family in the Midwest. My parents have been in business for almost their entire lives. From a very young age, my parents told me that if you wanted to control your own destiny, you had to start your own business . So I guess from a very young age, I was involved in different entrepreneurial projects. I loved it and enjoyed it, and I knew I would do it for the rest of my life.
Amit : Vettery, can we start with your first startup and give people a sense of your experience? It was a successful company, but it also took a lot out of you and I think it taught you a lot before you started Figure. So what does Vettery do?
Brett: I founded Vettery in 2012. In simple terms, Vettery is a recruitment marketplace. We have employers and job seekers, and we try to match them without any human intervention , using early versions of artificial intelligence technology to match employers with job seekers, thereby avoiding the need to hire third-party recruiters.
You know, most people aren’t happy with their jobs, except for the time they spend at work. If you go and look for a job, it’s a terrible experience. So our goal was to try to reinvent the whole process of finding great talent and applying for a job, and basically build an online marketplace to do that. So Vettery was not only a marketplace, it was a two-sided marketplace. By the time we sold the company in 2017, we were in about 18 cities.
It was a bumpy ride, and we had to pivot several times and raise several emergency convertible note rounds to keep the company afloat. At one point, I even borrowed $50,000 to pay for rent and living expenses in New York City. It was tough.
Amit: A $100 million (M&A) exit, is this a life-changing fortune?
Brett: Yeah, I went from being in debt to having tens of millions of dollars, and it had a huge impact on my life.
Amit: From Vettery to Archer, you moved from the talent market to eVTOL (electric vertical take-off and landing vehicle).
Brett: I ended up selling Vettery to a Swiss company, Deco Group, which is the world’s largest recruitment company. It took a full year. At the time, when I received the acquisition letter of intent, I was looking to leave the software industry and go into the hardware field. I have always been passionate about aviation, and we are in the wave of electrification.
Amit : Did this passion start from your first flight?
Brett : Yeah, I’ve been following aviation my whole life, and it was clear that electrification was going to happen in my generation. So, I started Archer in 2018. At the time, hardware was very difficult to raise money for entrepreneurs . I had to fund the company myself for the first few years. Basically, I invested everything I had earned in Vettery, and it was a tough experience for me. You know, I went from $50,000 in debt to tens of millions of dollars, and then I put almost all of it back into Archer.
Brett: It was hard. Now we are listed and the market value is about 4 billion to 5 billion US dollars (note: as of press time, the company’s market value is 4.82 billion US dollars), but at that time, the process of going public was like torture.
Amit: So, what was the mindset like then? Because you went through the network effects of Vettery, and network effects are not easy for anyone trying to build a platform. And then you moved to hard tech, which is not only hard in terms of financing, but it’s also very hard to actually build a product like that. Do you enjoy the pain?
Brett: I really enjoy working on companies and products, and I love challenges. Just coming to Figure every day is like entering a world of challenges. I get to work on the hardest, most challenging problems in the company every day, and that’s very interesting to me. I don’t spend time on things that are already running well , you know, I’m solving those problems every day. That’s the fate of entrepreneurs, you choose this… The whole process is not fun.
Vettery and Archer and Figure are no different in that respect, they were all painful, but also very rewarding. Archer was a fresh start for me, getting me into hardware. I basically went back to school, back to the University of Florida, and worked with the aerospace and mechanical engineering departments to learn how to build a flying vehicle from scratch. So, we built our own batteries, electric motors, control software, and embedded systems at Archer, and it was like a robot that could fly , in a sense .
Amit: From Vettery to Archer to Figure, we are here at the company headquarters with a robot walking in the background. First of all, the Series B investors are lucky enough to see what you have built in person, which is incredible. It is like science fiction, like The Jetsons, and beyond our imagination. So, what prompted you to leave Archer and move on to the world’s most difficult challenge – building a humanoid robot?
Source: Amit Kukreja Podcast
Brett : Yeah, I hope you can feel something after taking a walk around here today.
Amit : There is a lot of energy here.
Brett : Robotics is also in the works. I think a few years ago, even 10 years ago, it would have been too early for all of this, but 10 years later, it’s too late. We’re at the right stage now to get humanoid robots to market and to drive that. It’s very exciting. The reality is that we’re now vertically integrating all of the hardware design from the ground up, including the electronics, the joint actuators, the battery systems, the embedded system software, all of that. All of that is in progress. And the hardware is getting more powerful. You know, the hardware is getting better every week. We’ve never been in a situation where we walked in and found that the hardware was worse.
Secondly, I think you probably saw our demos today, the robots are working. The neural networks on these robots are doing really amazing things, being able to command robots with a high degree of freedom . It’s like magic when you see that happen. You know, you saw something today. Commanding robots with neural networks, it looks like magic, I mean, you can see it.
Amit: It’s magic, frankly, seeing them actually work and act autonomously tells us that in the future, people will no longer have to do heavy manual labor . For example, my father no longer has to do those heavy manual labors, or at least my children don’t have to worry about doing those jobs. If Figure is successful, so are other robotics companies, why do we need robots?
Brett: We need some way to bring the AI systems we have today into the physical world , to touch and move objects, to do useful work. In the future, everything that moves will be a robot, and most of them will be humanoid robots.
Amit: Why are you so sure about this? Why is this right?
Brett: It’s interesting. We feel like we can see what’s going to happen 12 months into the future. At least at Figure, we can see the robots, we know what’s on the product roadmap, and we can see how the latest technologies work here. We can see all of this 6-12 months ahead of everyone else. We have the confidence that we can basically build a “robot version of a human” now. This is a breakthrough solution in robotics, where you can have a humanoid robot do everything a human can do, which will eventually be in the billions of workers and eventually in billions of homes.
So basically, you don’t have to make any changes to our hardware to have the robot learn new things, you just update the software, and that’s an extremely powerful platform that’s coming to fruition and working over the long term. I think that’s very important. I’ve been in a Waymo self-driving car, and it’s an incredible experience. I’ve been in it and I can feel the engineering. They’ve been working on this for 16 years. You know, we’ve been around for two and a half years. If we do this for another 10 years, I think you’re basically going to get to a world where there are humanoid robots everywhere, and they’re all running neural networks, they’re all communicating through language, they’re able to reason about language, across every scenario, and be able to do anything you ask them to do.
Amit: So, when did you realize that—even though Archer was ambitious—it was time to move on to something more difficult?
Brett: I think there are a couple of things here. I’ve been fortunate enough to have started and funded a couple of companies that I think are really important. I started a company in 2023 called Cover that’s focused on technology for school shootings, helping detect concealed weapons. Archer has an incredible path to enable sustainable transportation in the sky and helping people get around.
But Figure has an opportunity here to build what could be one of the most important assets of our lifetime. Once we have humanoid robots, they will be doing useful labor every day in the economy , and I can’t think of a more important asset than that. To me, yes. This is a very hard technical problem, but it’s also going to be a business that’s going to have a big impact on humanity in the long term.
Amit: It’s like swiping a video, it suddenly appears in Instagram’s recommendations because you can’t suddenly wake up and think about robots.
Brett: I was always reading Isaac Asimov and dreaming about robots and flying cars and robots. The future was going to be incredible. I think I had a better understanding of how robots work at a fundamental level, and I thought it was achievable from a technology perspective. So, we pushed it harder than anybody else, and we tried to move as fast as we could and allocate capital carefully. We basically innovated as fast as we could in technology to improve it and push robot learning as far as we could . For us , AI in robotics was the future, so we pushed as hard as we could in that regard. Yeah, I believe this could become one of the largest industries in the world.
02
The core competitiveness of humanoid robots
Amit: So let’s get into this, labor is a multi-trillion dollar number in global GDP. Most of the world’s jobs are in factories and warehouses. Is Figure’s goal to eventually replace those jobs so that people don’t have to do them anymore and the company’s margins and free cash flow will go way up because they don’t have to pay these expensive wages? Or is Figure’s long-term goal different from an economic perspective ?
Brett: Yeah, there are two parts to our business. On one hand, we want to put robots into homes. That’s going to face very difficult problems of safety, common sense reasoning, and intelligence, and we’re working very hard to drive that in the company, and we’ll be talking more about that in the near future. The other side is how to deploy robots in the workforce , and every potential customer that we approach, the large enterprises that we’ve signed up now, they have a huge workforce problem. They can’t find enough people to do these jobs. Employee turnover is extremely high, and it’s very expensive to find replacements.
Amit: People don’t want to do these jobs anymore.
Brett: These jobs are not good. They are boring, repetitive, and nobody wants to do them. If you ask me to do the same thing 10 hours a day for 10 years or five years, how long can you keep doing it? A lot of the jobs we see have an annual turnover rate of over 100%. Recruiting the workforce is a very difficult problem. We have about 10 million unfilled jobs in the United States. I think it will take a while to fill that.
Amit: You are completely contrary to the view that robots will take people’s jobs if there is a labor shortage.
Brett: No, I think what we’re seeing right now is not a one-to-one attempt to replace human jobs. But I think over time, you know, humans have gotten pretty good at automation. I grew up on a farm, and there are hardly any farmers left now. I don’t think people are upset that a Kuka robot arm is lifting a chassis inside a car manufacturer, you know, no one is going to say, “I don’t like that I don’t have that job anymore.” It’s a bad job.
So I think that’s a choice that humans can choose to say, any physical work that we do, whether it’s manufacturing, labor work, or work in the home, like walking the dog, making coffee. That would be a choice. You can choose to do those things yourself in the traditional way, or you can have a robot do it. You just tell the Figure robot, “I need you to do this,” and it will do it.
Amit : Do you really think there will come a day when you can just tell a robot to do your laundry and it will do it?
Brett : Yes , we are already doing something similar in our office. Yes, in the long run, the UI of humanoid robots is speech . You don’t need to take out your phone or type to the robot. It is right next to you, it will understand the language, and it will output actions intelligently.
Amit: It’s interesting that you mentioned the phone. Apple has about $90 billion in revenue per quarter, $45 billion of which comes from the iPhone. Will there ever be a day when $45 billion becomes insignificant compared to the robotics business? Figure’s robotics business will be worth hundreds of billions or even trillions of dollars.
Brett: Robotics will be a multi-trillion dollar market.
Amit: Why?
Brett : Basically, if you had unlimited resources, you would be able to put very cheap robotic solutions into anything you wanted, whether it’s laundry at home, dishes, chores that no one wants to do, or in manufacturing or in the workforce. You would be able to choose to do those things yourself or have a robot do them. I think that would be a choice, and I think that’s very exciting for all of us. You know, our average lifespan is about 80 years, and we spend most of that time working, which is kind of bad, and in the long run I think this business is going to be very large , and you’re going to be able to deploy billions of robots in the workforce …
Amit: From an ambition perspective, there is no bigger market opportunity than humanoid robots.
Brett: Approximately 50 % of global GDP is human labor, which will be the largest industry in the world , far exceeding any other business , in the long run.
Amit : What is NVIDIA’s role in all of this? If we’re going to have hundreds of thousands, billions of robots, will they all need GPUs?
Brett: In the long term, at least internally, the code that we traditionally use for inside the robot is moving towards inference. You can see it moving towards the rules that we run on the robot . You know, we use NVIDIA (GPUs) in two ways , on the one hand, we use them for training, we do a lot of new model training now. On the other hand, we use them to run inference on board the robot . So, each robot is equipped with an NVIDIA GPU for inference.
Amit : And without reasoning it is impossible to build a (good) robot.
Brett: We also need some way to direct the robot’s movements. I don’t think it has to be NVIDIA – I think they’re great. We work well with them, they’re an investor. I know Jensen, he’s a great guy, and their team has been very supportive. I think there may be other options besides them, but at least for now we’re using NVIDIA.
Am it: We know that you recently announced that our cooperation agreement with OpenAI has ended. So, how do you train these bots? OpenAI has 300 million weekly active users and they have one of the best models. Why don’t your bots use their models?
Brett: That’s a great topic. You know, they participated in our Series B last year. We’ve been trying to develop new AI models for robots.
So, I would say two pieces, one is the VLM ( Vision- Language Model ), which is basically very important to us , the semantic foundation . You really need to understand what people are saying, you need basic language conditions, you need to understand what objects look like, and the VLM basically provides us with this information. The second piece is that what we need is real robot actions , such as the robot obtains data from the observation space that we can use for training. How does the robot understand how to move its arms and limbs? How does it actually command the trajectory to grab something? And these are not in the LLM (Large Language Model), the LLM knows nothing about these because there is no data of the robot in the LLM.
Right now, we basically use open source models, but we also use our own internal models and data that we collect internally on the robots, and we basically build the base models ourselves, and we’ve been doing this for about a year now, and we think we’re the best in the world at robotic AI learning .
Amit: Because you have the data you collect.
Brett: We are vertically integrated , custom built , and we can keep making it better , from the underlying hardware all the way to the AI neural network inference . When we interview people, they will say OpenAI is so and so, and we will say, no, we actually do this ourselves, you should join us, the team is completely crazy, everyone is basically the top talent in their field in terms of robot learning.
A few years ago, you could argue that AI is about unique data, and robotics hardware is a commodity product, and you don’t want to get into the robotics business. But now I think that argument has been completely overturned – you need robotics hardware to do the training in the robotics field, and you need to be able to get data from robots to train. There is no good hardware on the market. You can’t go out and buy a humanoid robot and say, “Oh, it’s good quality, and then take it and train a neural network.” So we were basically “forced” to build really reliable hardware from scratch.
Combined with the third generation of robots that we just finished , I think I can say that we are pretty good at it . What we have done on it is to build the best neural network in the world , and we are constantly iterating to make it better, and we will have huge improvements in the next cycle, and the robotic hardware will make the neural network faster, smarter, and better performing.
In this field, there are two different groups. One group is like the Chinese robotics companies, they are like “yeah, here is the hardware. We don’t care what AI is, this is hardware, you can use it however you want”; the other group is almost the exact opposite, they are more like a service model, they say “I don’t want to make robotic hardware, but I want to build the brain of the robot”, Google DeepMind has been working on this field before, NVIDIA also has a team doing this. There are some startups doing this in this field.
But what is really important is to do both hardware and AI at the same time, and to integrate them well .
Amit: Is that a secret that other startups don’t understand yet?
Brett: I don’t know if they understand it or not. But what we really care about is giving AI body . In order for the body to work, it needs to have certain conditions, and obviously the AI needs to be good. Maybe other teams are in business model reasons, maybe they don’t fully understand what is important. Figure is in the middle .
Amit: But logically it is not difficult to understand – if neural networks want to become better, they must be trained on corresponding hardware. Why don’t those startups realize this?
Brett: In this world, humans can basically operate any robot. So you could argue that the ultimate AGI might not need specific hardware, they can run on any hardware. Because the AI policies are good enough to actually command any robotic hardware.
Amit: Do you believe this?
Brett: I haven’t seen any evidence of that. There’s some truth to that. But the really fundamental question is, where is your AI system going to be deployed? For example, if you don’t have any hardware today – your AI is great, you can put it on any hardware, what you really want is to get as much data from robots as possible, especially for specific application scenarios. If you don’t have the hardware, then your AI system can’t get that data. You have to make sure your AI can be deployed in the real world. To do that, you basically have two options: either make the hardware yourself, so that your AI has a carrier to run on; or sign an exclusive agreement with an AI company and let them rely on your hardware.
But neither of these are particularly attractive options. If you choose the former, it’s very difficult to make hardware; and if you choose the latter, who wants to be just an outsourcer providing hardware to other AI brains? These are all very tricky problems. You can see similar situations in the field of autonomous driving – you really need robot data for specific scenarios, such as car data. You can’t simply transplant Waymo’s neural network to Cruise because their hardware and data are different. There is currently no evidence that this “hardware doesn’t matter” AI strategy is feasible . Of course, there are also some teams working on this problem.
We are lucky because we have our own hardware and we can control the quantity and customization of the hardware. So, we don’t have to worry about solving this problem because we already have our own hardware to carry AI. I think some people will say that hardware is difficult and don’t want to get involved in this field. Others will say, here is a hardware, you go do something smart with it. But these approaches don’t seem to have achieved very good results.
Amit: Did you foresee that “hardware and software integration” would be an advantage two years ago, or did you think of it by chance?
Brett: No, our long-term vision was clear from the beginning: we needed to be an AI – first company , and our product line would span the entire market from the workforce to the home. If we look back, my idea was that we needed to make a really powerful, safe, and high-performance robot in the workforce. That would pave the way for us to enter the home market. I think our plans to enter the home market are accelerating almost every month, which I’m very excited about. The other point is that we need to make really great hardware. You can’t make a great robot with bad hardware , it has to be unparalleled and it has to work consistently, just like your phone.
Amit: Yes, just like the iPhone, it has to be beautiful and easy to use.
Brett: Yes, and it has to work consistently. Right now, humanoid robots are still in their developmental stages, and like early “explosive rockets,” they still make mistakes and they’re not perfect. But they’re going to eventually become very capable and even very reliable within this century. So we need to build really great hardware to do that, and then have great robotic learning on that hardware. In a sense, for Figure, this is a sequential problem: we can’t have great robotic learning without great hardware.
So we spent the first year to year and a half focused on making great hardware. We’ve spent the last year working on great robotic learning. If that hasn’t been apparent to the public, in the next 30 days we’re going to show something that’s never been shown in the history of humanoid robots.
03
The Tesla Model: Vertical Integration of Humanoid Robots
Amit: Are you always trying to prove something?
Brett: I don’t.
Amit: You’ve had a lot of success. So is your drive just to do hard things and make hard products, or do you want to prove something to the world? Is there a need for the Brett Adcock name to live on after you’re gone?
Brett: No, I think it’s important that what we do here and what we’re doing lasts long enough to really help humanity. For me, it’s never been about money. I’ve made a lot of money in my career. It’s about the opportunity to build what may be the most important thing of our lifetime. We have an opportunity to do it now, and this year is going to be an extremely challenging year for us, more challenging than any year we’ve ever had. If we took two years off, we would have definitely lost our opportunity. So, we have to go all in now to build a truly transformative company. We have 200 engineers who have all joined us with this attitude: We have an opportunity to do something really great, and we’re going to work hard, we’re going to work weekends if we have to, and we’re going to do it. We have such a truly amazing group of people, and I don’t think we talk about that too much. Everything we do here is because of these amazing people. They’re here, working hard, trying to launch products and make this happen.
Amit : What is the typical talent profile at Figure?
Brett : We have a unique company culture. I learned a lot when I started Vettery, and I learned even more when I started Archer, grew the team, and went public. When I started Figure, I had the opportunity to stop and think about how to do better. I did something different at Figure than at Archer. I created a 10-year vision plan, which I also did at Archer. But I also did something unique. I created a culture document, which is a document about how I expect employees to act and what we believe in. We start with mission, vision, and values, and then go deeper into what we care about. We advocate a very flat organizational structure, almost no bureaucracy, very fast, and do the right thing.
Amit: There is no middle management.
Brett: Yeah, there’s no middle management. We don’t have management that doesn’t actually do the work. Everyone here either writes code or draws the design, even those at the director level, there’s no one here that doesn’t actually do the work, including myself.
I was in the late stages of Archer, and one day I realized that I was sitting in a conference room full of executives, participating in board meetings and analyst calls, and I asked myself, “What am I doing here?” I need to get back to the front line, working with my engineering and product design teams every day, and understanding the realities of product development. So I now sit next to the mechanical, AI, and controls teams, participate in every stand-up meeting and design review, and spend all my time on product development. I think this is a continuous learning process in my career, understanding what I should do as a CEO and what is really important. And that is to focus on product and engineering, because at the end of the day, what really matters is launching the product.
Amit: Yesterday I met the co-founder of Robinhood, who is working on a new startup. He mentioned that when they completed the first commission-free trade on Robinhood in 2015, that moment changed the world of retail investing. So, for you, is there a similar “Robinhood moment” when you realized “this is going to be a big thing”?
Brett: I think it can be divided into two phases. The first is in AI. Our first generation of hardware, Figure 1, was not ideal. The hardware was made very quickly. We did this intentionally because we wanted to have something to give us AI and control as soon as possible. But it was not reliable and often failed. However, about a year ago, we started doing a lot of AI work internally, and you can see that based on very little robot data , the robot’s neural network performs very well , or learns and adapts very quickly. We released some videos showing these results, which are really impressive. I think this is the first phase.
The second phase has been in the last four months with the introduction of Figure 2, which is a very reliable robot that almost never falls over and just works all the time. Now, all the issues we have are software-focused, and we can fix them with software updates . It’s not a hardware issue. You know, we’re working on the next generation of robots right now, which will be another order of magnitude leap in design.
So I think in the last 90 to 120 days, we’ve seen this all come together, and it’s started working faster than we expected, which has been two and a half years. Now, the bottleneck becomes (the need for) more robots, more computing power , more data , more training data. How do you get more? We’re going to build more robots, and over the next year, everything will be brought to a new level, an order of magnitude higher than what we did last year.
Amit: Why do we need more computing power, more infrastructure, and more scale? Because the demand is endless?
Brett: Yes, if robots can continue to create value in the world and do work, and if these robots are relatively affordable, then they will continue to drive down the price of goods and services. They make everything more abundant and cheaper. Of course, this is not infinite, but for us, at least it looks like that at the moment. We have two customers right now.
Amit: You just signed your second commercial client.
Brett: Yeah, you see the work we’re doing. Between those two customers, we expect to need to deploy a hundred thousand robots over the next four years. That’s a lot just from those two customers, and they’re large companies. We’ve been in touch with over fifty other companies, and we went through the screening process. We ended up choosing those two customers because… you know, this is just the beginning. But in the last six months, we’ve seen so many requests for proposals from large companies asking us to submit proposals on how to use robots in their facilities.
Two years ago, we were going to C-level executives and trying to explain to them what a humanoid robot was. You’d hear them say, “What is that? A robot that walks on two legs? What are you talking about?” You’d try to explain it to them, “Well, it’s a robot that walks on two legs…” They’d say, “Can you put that in my company?” You’d hear some of the biggest CEOs say, “That’s the craziest thing I’ve ever heard.”
Now, these companies are coming to us and asking when we can deploy robots in their facilities and test use cases. The entire industry landscape has completely changed in 24 months …
Amit: I want to talk about Tesla. So, what does Elon Musk mean to you? What do you think of him as a CEO and entrepreneur?
Brett: I don’t know what I think of him today, but certainly when I was a kid, Elon was a generation older than me, and I watched him become one of the best people in my lifetime. I think he did a lot of things that might not be unique to a lot of founders and entrepreneurs, but when you look back at the greatest entrepreneurs of the last hundred years, they are very unique, and it’s very rare to see someone who cares so deeply about doing great work. So, he’s absolutely extraordinary.
Amit: Tesla is obviously working on the same problem with their Optimus robot. Is the market big enough for both Figure and Optimus to succeed? Is this a winner-take-all market? How do you understand the relationship between demand and the companies chasing it?
Brett : My view is that it’s going to be a situation where a few winners take most of the market , and I don’t believe there’s going to be one team that’s going to be 20 years ahead of the rest of the teams. I don’t believe that everybody is going to figure it out. So, take electric vehicles and autonomous vehicles, and even in the eVTOL space, when I started Archer, there was an eVTOL directory. I remember when we went public, people said there were 800 eVTOL teams and it was very competitive, and I said probably half of them were just pie in the sky. You know, a lot of them have gone bankrupt. Just like Archer and Joby, globally, at least outside of China, they’re trying to win this race. It’s true for electric vehicles, it’s true for aircraft, and it’s going to be true for humanoid robots. There’s going to be a few teams in the world that are really good, that attract most of the funding, that have the best talent, and that make the fastest progress in product development.
A lot of teams have been working on this for the last 10 to 20 years with little progress and eventually fade away. You’re already seeing this with teams that have been in the humanoid field for a decade but haven’t made as much progress as Figure. Of course, we’re also seeing some companies making significant progress.
Of course, China is also developing humanoid robots as a whole like crazy. So, I think there will be a few teams, maybe one to three teams, that will achieve great success in this field.
Amit: Just three?
Brett: The problem is that this market requires huge unit volumes. You need to build a lot of robots for two reasons.
First, the only way to get costs down is by making products in large quantities, driving costs down the manufacturing experience curve. If you want to get the cost of a robot down to $10,000, $20,000, or $30,000, we need to make millions of robots a year and we have to make them very well.
Secondly, while you build a lot of robots to drive down costs , you also need a lot of robots to collect data and learn together. One of the big advantages of humanoid fleet knows it. This is very different from how humans learn. You see my children learning to walk or do other things, they don’t learn from each other, they learn by failing, just like the little robots. They fall down, fall down again, and then they learn. A huge advantage of humanoid robots is that the whole fleet can share the learning results.
To achieve this, you need a super-excellent team that can move fast and design products that can really provide value to customers; you also need billions or even hundreds of billions of dollars in funding. So, this is not an opportunity for everyone. There are many teams that have been formed for a long time , they have been raising funds , but have made little progress. This market is not like B2B SaaS, where everyone has a chance.
This is going to be a very hard technical problem, one of the hardest technical problems we face in our modern era, or at least in the top five. It is very hard to commercialize a humanoid robot at scale that can interact with the world and learn through reasoning via neural networks.
Amit: Some say Tesla is vertically integrated and therefore they have a competitive advantage in manufacturing.
Brett : We make almost all of our hardware in-house. We design all of our software, we design and deploy our own neural networks. And then we do our own manufacturing. So you see the Figure 2 line , and this year’s Figure 3 will be here . We plan to do this ourselves and do it well. I think one of the things that I’m realizing more and more is that car manufacturing is very different from robotic manufacturing. The biggest difference is , you can’t hold a car by hand , right? You have to have machines to move these thousands of pounds of chassis around the facility from A to Z. So you need to develop a complete system. You need a lot of space, and you have to do it because you can’t hold it by hand.
But you can take any part of a phone or a robot, right? This gives us tremendous flexibility in manufacturing . We don’t need to build a complete facility to do all the end-to-end work like a car manufacturer . This makes the manufacturing process very complex. We are designing automated production lines here (headquarters). So, we are basically robots plus manufacturing automation to be able to complete end-to-end production.
But it’s a completely different experience than what Tesla or the car manufacturers are doing. Car manufacturing is a very complex process, and robotic manufacturing is completely different. I think for car manufacturers, if they can do it with high efficiency, then they will have a certain advantage. But I don’t think that’s the long-term goal of robotic manufacturing. I think the hardest part of making robots is not the manufacturing process itself. If you can make a million robots, you will have the largest factory in the world, and no one can make so many robots. It’s not like if you can make two million or 200 million mobile phones, you can make them by hand. So, I don’t think the manufacturing process is the hardest part. You can’t make 200 million mobile phones by hand. So, I don’t think the manufacturing process is the hardest part.
I grew up in this type of environment, so I was excited to be involved in the manufacturing process. I was excited to walk the production line, we had built the prototype line, I was personally involved in manufacturing parts and seeing how they would fit. We had weekly reviews with the manufacturing engineering team to design the production line layout for the various components of the robot. I assumed we would encounter a lot of resistance trying to make this the best it could be, but we were determined to give it our all and complete it at the highest level possible.
04
Self-analysis, work and life
Amit: How was the process of raising the Series A for Figure?
Brett : It was pretty smooth (laughs). You know, compared to Archer’s Series A, which was the hardest round I’ve ever had to raise. In 2019, 2020, almost no one was willing to invest in deep tech companies. Yeah, in 2020, no one was willing to write checks. I had to put up my own money and then raise money from some people who were willing to give me money. Yeah, it was like, you come from a software background, can you do hardware? And there were no funds that were really focused on deep tech at the time. Now there are a lot of good funds starting to look at deep tech, but at the time… I was almost the largest investor in the Series A, and I personally invested most of the money. I was almost prepared to pay all of it myself. So, it was like a “don’t care” attitude that won me over a lot of people. They thought, maybe this guy really succeeded at Archer and now he’s going to try Figure. I don’t know. But fundraising is always difficult, but this time is not the most difficult.
Amit : Do you enjoy the process of persuading people? Do you enjoy the process of selling?
Brett : Not really. Right now, I prefer to have candidates come to the company for a tour or during the fundraising process to come and see what we’re doing. I want the work to speak for itself. I don’t need to convince someone that humanoid robots are going to be big or anything like that. I just want them to come here and get a feel for it and see what we’re doing. It’s completely different than just browsing the website or hearing about it. We have a saying, “We don’t make an offer until they come to the company for a tour.” So you can’t really get a feel for what we’re doing until you’re here. That’s hard to convey any other way. One of my goals this year is to figure out how to convey that “feeling of being here” to everyone on the outside, which obviously helps with everything we do, whether it’s commercial sales, recruiting, or raising money. You don’t really get a sense of the atmosphere until you actually see 50 robots walking around. You know, that’s hard to get until you’re here. So we’re working on getting better at conveying that feeling to the outside world.
Amit: …What is your favorite movie?
Brett: I’m a science fiction fan, so I love science fiction movies.
Amit: On future topics.
Brett: The Terminator is great, I, Robot is great, Contact is great. I watch almost any sci-fi movie.
Amit: What is your favorite book?
Brett : I read a lot of Isaac Asimov as a kid , and especially now that I’m working in robotics, it’s a real privilege to actually try to implement these ideas . I also read a lot of nonfiction books, such as a lot of books on manufacturing and motors on my desk. I think that knowing these disciplines and understanding how they work together can help me make better system engineering decisions.
Amit : Are you able to sit down and read without your ADHD distracting you from doing other things? Are you able to maintain deep concentration?
Brett: Sometimes.
Amit: How do you define leadership?
Brett: I think the most important thing is that we need to be united and moving in the same direction. There are two things here , direction and speed. We need to make sure that the company is always moving in the right direction, and this requires the CEO and the company leadership to set a common vision. If we don’t communicate frequently, the team will be scattered and everyone will move in different directions, right? So, you need to constantly reiterate where our “North Star” is and which direction we are moving in. The second part is speed. Speed only helps if you know you are moving in the right direction. If you move quickly in the wrong direction, it will only make you deviate from your goal faster. But these two need to be combined very carefully.
I think the best leaders I’ve ever seen are the ones who can really do both of those things well. They’re the ones who are able to attract the best talent in the world and ultimately build the best products. Here’s the formula: We need a really great product that users love to use, that they keep using, that makes them happy. And to do that, you really need the best team in the world. It’s really simple: You need to have a shared vision and you need to move fast.
And then you need to iterate on the product as much as you can, incorporating customer feedback and everything else into the product. Getting that right is fundamentally the most important part of the game.
Amit: The most successful investments I’ve made have been those where the founders really cared about the product , and the worst investments I’ve made have been when the board brought in a guy in a suit and asked him to run the company. Is founder-led vision and founders who really care about the product a competitive advantage that Figure has over other robotics companies ?
Brett: Yeah, there’s a biological connection, it’s something you built from scratch. Yeah, and then you put a personal effort into it. You sacrifice time with your family, with your friends, for it. There’s something great about it, and when you see the founders really invested in it, it makes the sacrifice worth it. You know, if I want to do something, I want to go all in and be super competitive. So, I want to work at a place where the founders are really invested and willing to give it everything they have. They’re trying to build the best product.
Amit: Do you follow sports?
Brett : Sometimes I do.
Amit : Does Jordan inspire you?
Brett : The Last Dance is such a great documentary. All he did, all he cared about, was trying to be the best player in the world.
Amit : Because Michael Jordan didn’t care about money or contracts, he cared about winning every game. I think you are the same.
Brett: We want to do everything we can to win. If we put so much energy into this and come out with mediocre results 10, 15, or 20 years later, I’d be very disappointed. I have enough money, I don’t need to do this. If the end result is just mediocre, I might as well go to the beach and sunbathe.
Brett: But we’re here to win, and we want to build something truly transformative. We need to go all in now to make it happen. We need to really get millions of robots to market, as fast as possible. We need to put our heads down and really make this happen. We need to figure out how to get robots out to the world quickly.
Brett: This is a real competition. Aviation is fun, robotics is fun, but now we need to actually make it happen.
Amit: This journey must have been crazy for your parents and your wife? You do all this on a daily basis and it must have had a big impact on them.
Brett: My wife and I have been together for 20 years, and she’s been there for me since college. She’s been very supportive, and so have my parents. Yeah, I think in the first few years when I started my business, everyone was like, “What are you doing?” And then when things started to take off, they were like, “Oh, this looks interesting, keep it up.” Hardware has gone through that phase in the last decade or so. Yeah, I think, everyone is like, “Oh, this looks interesting, keep it up.” Yeah, everyone has been very supportive. I got to this point in my life where, you know, you have three things: work, family, and friends … I made a decision about six years ago to not be involved in those social activities. I can’t do all three at the same time, so I chose to focus on family and work and try to make things work.
Amit: My pleasure.
Brett: Thank you.
Author：明亮公司
Source：https://mp.weixin.qq.com/s/nLSGIh-Gw7IFSihZmXVDSQ
The copyright belongs to the author. For commercial reprints, please contact the author for authorization. For non-commercial reprints, please indicate the source.
