World Models: a trillion dollar opportunity

0:00Hey, folks, Jeff Berman here, co-host of Masters of Scale. I am thrilled to share some of the new names who will be joining us at this year's Masters of Scale Summit. They are the leaders driving the most pressing conversations in AI. Replit founder and CEO Amjad Massad, Cloudflare's Matthew Prince, Signal president Meredith Whitaker, and many, many more who will take the stage this October 20th through 22nd in San Francisco. We want you there with us, too. Join us at mastersofscale.com slash pioneers. That's mastersofscale.com slash pioneers.

1:00and the bottom line, how they navigate uncertainty, pressure, and high-stakes moments. I'm Bob Safian, former editor-in-chief of Fast Company, and I'll be your host as each episode breaks down what you need to know right now. You can find Rapid Response wherever you get your podcasts.

1:31Gaming data, in particular for world models, really includes so much interesting interactions that these models can learn effectively from. Companies are paying people to play games. Don't tell my son that. If you think about driving, you can have lots of actual dash cam videos of people driving real cars, but you can also have video game examples. And we don't want the model to learn from people driving a GTA, right? In World Models, I think the company that will win is the company that's most willing to explore all the applications. Driverless cars, as much as robotics, as much as gaming, as defense, as healthcare, all these other industries.

2:13Even energy. In the first half of 2026, investors have poured over$3 billion into the next frontier of AI. World Models. Companies like Google DeepMind, NVIDIA, Fei-Fei Li's World Labs, Yanma Kun's Ami Labs, and Odyssey are all building in this space. Today, I'm speaking with Oliver Cameron, a co-founder and CEO of Odyssey, on why world models are so hot right now. Oliver and I are diving into the trillion-dollar opportunity behind world models, why it's the next frontier of AI, and some of the applications that world models unlock. Everything from robotics to education and healthcare. I'm Rana El-Khalyubi, and this is Pioneers of AI, a podcast taking you behind the scenes of the AI revolution.

3:11Oliver, welcome to Pioneers of AI. I'm so excited to have you on the show. Thank you for having me. It's great to be here. I have to ask you, the name of the company is in the news right now. because of the movie, Odyssey. Did you take the whole team out to see the movie? We did. We rented a theater, yes. Oh my God, that's awesome. We actually named the company originally after 2001, A Space Odyssey. And this sort of pioneering movie of sci-fi, the future, new technologies. And so it's cool, I guess, to see sort of another version of that. Yeah, yeah. So before we dive in, I want to disclose that I'm an investor in your company, Odyssey, through my fund, Blue Tulip Ventures.

3:56And in June, you raised$310 million for your Series B round at a$1.45 billion valuation, which is amazing. Congratulations. Thank you so much. Yeah, amazing. So I want to set the stage. So a few years ago, the chat GPT moment, right? We saw large language models become mainstream and ubiquitous. And it's this idea that you can synthesize large amounts of text, images, maybe voice, even video to generate stuff. Right. And that is kind of what is really magical about large language models. But there's been an evolution in this frontier. And the next frontier of AI is world models. I would love your definition of world models and also like why now and why should we care?

4:40Absolutely. I think you nailed it with language models. They really ingest so much language that they deeply understand language and then can simulate language. But importantly, language is a representation of the world that is lossy and is biased. Right. We've written down things for thousands of years. It's our opinions, our thoughts. and there is so much information that is lost when we are transcribing our thoughts or our observations into text. And so although language models are these incredible intelligences, we think we can go further. And to go further, really what we want to train a model to do is to understand and simulate the world.

5:28And how do you do that? Well, instead of training it on lots of text. We instead train it on lots of visual observations of the world. Everything that you could possibly ever imagine happening in the world is represented as a video that the model is learning from. And that model, after it watches enough videos, deeply understands physics, dynamics, cause and effect, humans, our behaviors, how we work together, and all of these things that you just don't really necessarily synthesize in text. And so that model to me is effectively learning about reality. And so I feel that model will just be more powerful and will be this leap of intelligence that is difficult to comprehend.

6:16Yeah. You know, I'm actually more excited about the idea of world models and all of the applications that it can unlock. I'm more passionate about that than the vision or the mission of getting to artificial general intelligence. I'm curious, how do you think about AGI versus world models? And are they the same in your brain or are they different? I think it's likely that world models play a key role in enabling a superintelligence. And here's how I think this plays out. One of the applications who are very excited about world models is as learning environments for other AIs. Because, I mean, just to use our own analogy, we live in this world, we live in this universe, and we have all these agents, all these humans existing within it.

7:05And the world is constantly trying to kill us, right? Or has done since we became human. And we adapt, right? We learn not to die, and we learn to improve our world, to make it safer, to make it better, to make it more plentiful. I think that the analogy for a world model is that you have these worlds that a world model can provide, these infinite simulations of all these different worlds. Agents can exist and learn within those worlds. And they learn things by being exposed to that world. And interestingly, also, they also push the world itself to be better. And so I think to answer your question about superintelligence, it's very likely that a superintelligence is learning inside a world model.

7:54The world model itself may or may not be a superintelligence, but I think it will play a key role in enabling an intelligence to exist. Yeah. So world models are very hot right now. Well, just in the first half of 2026 alone, venture investors have invested over$3 billion into world model startups. And I'll just give a few examples. Obviously, Odyssey raised$310 million. Ami Labs, Suyan LeCun's company raised a billion dollars. Fei Fei Li, World Labs raised a billion dollars. And even Runway, which is a generative AI video startup, they've kind of just pivoted to build a world model. Why is there an urgency to build and invest in world models right now?

8:38I think it's two parts. So the first part is that it just presents something that feels distinct and really encouraging beyond language models. We could, I guess, just, or in the venture community decide language models are it, funnel every dollar into that industry. I don't think the venture industry works that way, right? Thankfully, it basically says there's other possible intelligences, other approaches. Let's also fund them and world models, I believe, are there. So that's one aspect. I think the second aspect is you just have to look at where the talent is headed. I posted a chart to our Slack today, actually, of the frequency of mention of the term world model in published papers.

9:28And you see this incredibly steep rise in 2025 and 2026 of just the number of mentions of world models across publications. And that shows you that researchers are just increasingly spending their time on this thing, world models. And so I think the money is following the talent. I also think it dovetails nicely with the excitement in robotics. We're seeing, of course, just so much maturation in robotics hardware and... More generalized robots, I guess. Exactly. Humanoids of all different types and different new types of robots. And it then becomes quite clear that those robots need to become intelligent.

10:13And thus, how do you do that will model presenter a very interesting path. Can you clarify for us, like why does physical AI, for example, a humanoid robot need a world model as opposed to just operate on a large language model? Yes. Let's imagine we're a robot here. So I've got a coffee cup here and we need to grasp the coffee cup, pick up the coffee cup, place the coffee cup somewhere else. And all of those things can be described in text, right? You can type in, pick up the coffee cup and all sorts of different things. But there is just a translation that occurs between the language model and the physical actuation that goes on.

10:57And that translation in some cases is fine. Like it just works and it works fine. You'd never notice. But our belief is that in lots of cases, it doesn't And it won't because there is just a loss of information that happens between the language model thinking in text and then actually interacting with the physical world. And that can result in task success not being great, meaning it drops the cup. It doesn't understand from looking at the cup that there is no top on the cup and that it should be more careful about picking up the cup. or it doesn't understand that with the lighting that's going on here, that this is plastic, not glass or a mug.

11:41And so a world model you can think of as effectively learning the language of the world. It doesn't speak English. It speaks physics. It speaks dynamics. It speaks lighting and all of these different things. And so it will just theoretically still today, we're having to prove this, but still we believe will be a better performant version of this. Other evidence that points to this is that we hear this all the time that we need data. Like robotics is bottlenecked on data. So you want your robot to pick up coffee cups all the time. You need to show it lots of examples of coffee cups being picked up.

12:22And if you think about trying to make a general purpose robot where you need to show tons of examples of a thing doing a thing, that really doesn't feel that scalable, right? You're having to collect lots of examples, do it here, here, here. You hear from this user, okay, now more here. Then all of a sudden you're just scurrying around collecting all these tasks. And there's been evidence that shows a world model can adapt to a task with dramatically fewer examples necessary. Meaning you don't need to show it a thousand coffee cup pickups. You just need to show it 10. Because it understands like the physics involved in doing this action.

12:57Another example of this that is something we've been working on at Odyssey, which is near and dear to our hearts, is driverless cars. So that was my beginning. I spent eight years building driverless cars. And the paradigm of driverless cars still today was that you train models on thousands, maybe even tens of thousands of hours of driving. You show lots of examples of cars moving, and it thus learns to drive. and this is very different than how humans learn right when we go to learn at maybe the age of 17 we get behind the wheel for the first time we don't sit in front of a television and watch thousands of hours of people driving what we do is instead say i've watched my parents drive for a long time i know which side of the road to drive on i know that crashing a car is a really bad thing.

13:50I know I should be cautious. And so we get to learn to drive very quickly. It takes us less than 30 hours of training to actually be behind the wheel of a two-ton tank, effectively. And we believe that world models can accomplish something similar, where instead of self-driving models, we throw hundreds of thousands of hours. You can instead just show 10 hours of driving demonstrations to a general world model. And it's already learned all of these concepts of physics and not to bump into things and what side of the road to drive on. And it just applies that knowledge. And so, long story short, world models should be more adept at interacting with the world.

14:27Amazing. So, actually, I do want to go to your origin story a bit. So, you and I overlapped a little bit in the automotive industry. When I was running Affectiva, we were building machine learning models that understand human behavior. And one of the applications, it turns out, was in the automotive industry to understand, initially, driver distraction and drowsiness. But then eventually we expanded that to in-cabin monitoring, and the application was semi-autonomous and fully autonomous vehicles where you want to understand what's happening inside the vehicle. Before Odyssey, you started a company called Voyage that was building self-driving cars, and you sold that to Cruise.

15:02I would love to hear a little bit about your experience doing that. Like what got you into self-driving in the first place? Especially that before that you were at Udacity and you were in the education space. So what's the arc here? Exactly. So really, it's one guy that I got to work with, a guy called Sebastian Thrun. He was the founder of Udacity, and Udacity taught online concepts like machine learning, robotics, self-driving cars, all these crazy things that were locked up in academia could now be suddenly taught to everyone on the planet. You know, I did a computer vision class for Udacity, which was awesome.

15:39I do remember that, yes. Yeah. And so one day we decided, hey, wouldn't it be cool if we taught people how to build a self-driving car? That seems fun. That seems futuristic. Let's build a class. And this was early 2015. And so we built this class and we actually built a open source self-driving car at the time because we wanted people to have this space to build on. And in fact, we had wanted students to be able to run code on a car in the real world, which was crazy to think about. Ambitious. It was the safety concerns, let's say, but we solved those and it was great. And that to me felt like after working in machine learning for lots of different things, it felt like the most magical form of machine learning I'd seen.

16:23that we could have machines, save lives, interact in the most complex places on the planet. And that felt incredible. And so in 2016, started Voyage, my self-driving car startup. And our goal really was to build state-of-the-art driverless vehicles and to find places where the state-of-the-art worked already and to have an impact. And so we focused on deployments in places like retirement communities, military bases, These sorts of slower walks of life where you can go from A to B without the crazy traffic of San Francisco. We built that company up over five years. It's an incredible experience.

17:04And then ultimately sold that company to Cruise. Spent two and a half years there scaling driverless systems. I'll be back with more of my conversation with Oliver after this break.

17:53We'll see you next time. who've changed the game. It's anything but business as usual. Find Masters of Scale on Apple Podcasts, Spotify, YouTube, or wherever else you get podcasts.

18:11One of my key learnings from starting Affectiva was the importance of getting the timing right. We were so early, like when we were building like emotion recognition technology and emotion AI, it was like, before smartphones existed. And now there's a lot more companies building in this space. When you started Voyage, did you expect that we would now be like all in driverless cars? Did you get the timing right? I think both yes and no. So on the first question, I did think it would proliferate faster. but I can go outside my house today, call a Waymo, travel 50 miles north fully autonomously, get back home.

18:59It's amazing. And so I think the question of Voyage being too early or not, I think my reflection was that the self-driving industry isn't one where you can start a company very easily. And so that just hasn't even definitely today, that just isn't many, if any, self-driving car startups started, right? So it's kind of hard to say, like, did we pick the right time or not? I think my lesson learned from Voyage was that in artificial intelligence, I think it's counter to most intuition for other companies, which is you need to simply do the hardest thing first. I think mostly in company building, you try and find a relatively simple thing to do first, and then you build upon that and build upon that all the way to the most complex thing.

19:48And I think my lesson learned was that the companies that were solving San Francisco and self-driving like Cruise, they were in fact going to get to driverless anywhere faster because they focused initially on San Francisco, not slower. And that was really our whole thesis. And so that took, you know, a lot of courage to admit, but it really felt like that at the tail end. And so I do carry that over into Odyssey. I love the advice, do the hardest thing first. How did your experience at Voyage and Cruise kind of inform your decision to start Odyssey? And I just like as a kind of a former entrepreneur, I guess, I'm always curious, did the idea come about while you were at Cruise?

20:37Or did you take a sabbatical and then the idea came about? Like, how did that all transpire? Absolutely. So the technological insight in self-driving, the paradigm was that you had these different models to do different tasks. So you had a perception model to perceive the world. You had a planning model to plan the path through the world. And then you had what we call prediction models. Sounds very generic, I know. But really what a prediction model did was say, I have a visual observation of the world, the camera. I'm going to predict what the world looks like in five, maybe seven seconds. And I was always in.

21:18Like if it saw like somebody walking across the street, it's going to predict that this person's going to get in front of the car in the next few seconds. Something like that. Exactly that. Yes. And that goes for cars turning, merging into your lane. basically what behaviors are they exhibiting that you can use to predict what they might do next and I always felt that this was magic the idea that we're time traveling into the future albeit five seconds and that these models tend to be right that they tend to predict correctly what is going to happen and I felt that was magic and so really the origin of Odyssey was was that it was that if you can predict a possible future of the world, that seems really powerful.

22:06That should apply to lots of different industries, enable new applications. Let's go build general, what we now call world models. So that was the origin. And how it came about really was I was at Cruise and met someone that was the first researcher at Wave, another self-driving car company. and we met to go for a ride in a cruise vehicle around San Francisco, fully driverless at the time. And we just threw around different ideas and that was the one that we both came to and a few months went by, we started throwing around more details of the idea and then we decided to start a company. Yeah, that's amazing.

22:46World models have tons of applications. What is the total addressable market? What's a TAM for world models? Yeah, so I think it's, easily in the trillions. And I think what you have to do to believe that is to basically say that they are going to be integral to robotics. And at some point in the next 10 years, we're going to see upwards of hundreds of millions, maybe a billion robots that start to be in our daily lives in many different ways. And that world models will play a key role in that. I also believe that you can see world models playing more fundamental roles in science and will discover things that will lead to whether it's drug discovery or material science and that world models should take a piece of those discoveries.

23:41And then industries like education, gaming, healthcare. Can you kind of give us examples? Like what could the application of a world model look like in the education space? So in education, my belief is that there's no reason every kid and adult shouldn't have Einstein in their pocket that is a natural teacher of whatever that person wants to learn. And you might ask, well, can't language models do this today? And the truth is they can produce text that sort of resembles Einstein. But a great teacher, from my perspective, is one that's looking you in the eye that is able to understand if you're getting the concept that they're teaching.

24:27If you're not getting it based on your body language, they change what they're saying and then your body language changes and then hopefully you get the concept, right? That is so near and dear to my heart, as you would expect, because I spent years, right? Like looking at these emotion signals. Yeah. And a world model could do that, right? Precisely. Like a world model can have an input of video and that video is you as you currently are. and it's continuously generating simulations that are responding to your body language. And it just simply has the knowledge of body language and pre-training to better react to it and to become a better teacher.

25:07Because world models can unlock so many applications, how are you prioritizing what industries to go after? Yeah, we have a somewhat unique take on this, I think. So So in world models, I think the company that will win is the company that's most willing to explore all the applications. And I think, again, that kind of goes counter to what lots of early stage companies do, which is to focus, focus, focus on one thing, two things maybe. And we're quite intentional about saying, no, we're going to go very uncomfortably broad. We're going to explore driverless cars as much as robotics, as much as gaming, as defense, as healthcare, as all these other industries.

25:44even energy. And so that necessitates, I think, building both a culture, but also a team that is capable of doing that breadth of exploration. And so that's quite intentional. And I think it all leads back to this is a foundation model. Ultimately, it's a model that can help all these different industries, do more. And we need to be the company that has explored the best, most interesting applications. I've heard you say that we are in the equivalent of the GPT-2 moment in world models. When do you think we're going to kind of see the chat GPT moment, right, of world models? And where are we on the spectrum of doing basic research to actually getting closer to commercialization?

26:36You know, my mind has changed on this. So I think this chat GBT moment will look like one world model that can drive a car, fly a drone, control a robot, generate a game, teach your kids the ABCs. If one world model can do all of those tasks, then that is a hugely monumental moment i feel and that is a gbt3-esque moment and so that's what we're running towards and it may not feel like the chap gbt moment i think also that's difficult it's like trying to you know sort of replicate the iphone moment uh in some ways right it's just a relatively unique time but i think in terms of its uh importance i think what I described that one model to do all those things, that would be more than equivalent at that moment.

27:31And do you think we're close? I'm not going to hold you to a time. I do. I really think it's very close. And I think in some part, it is even closer than I thought. I'll be right back. But first, a quick break.

28:12This is the part of the interview that I'm actually most excited about. I would love for you to take us behind the scenes and help us unpack what does it take to build a world model. And in my mind, the way I think about it, if you're building any type of AI, there's three components. There's the data, there's the algorithm, and there's the compute. So let's start with data. What type of data is needed to train a world model? So there are a few sources of data. We can think of this as a big pyramid. At the very base of the pyramid is internet video. On the internet, there is boundless volumes of video that describe everything you could possibly imagine about the world.

28:55And it's growing faster than text on the internet, for example, and it's evergreen. It's just continuously updating every day. And so that to us is the bitter lesson, this term that's thrown around a lot in artificial intelligence, where simply go to the best, most fast-growing data source you can for the task that you want. And internet video is the base of the pyramid. Can I ask you a question on that? Sure. Because you've got YouTube videos of people driving cars and whatnot, but you've also got TikTok videos of people on vacation saying, hey, this is me and like whatever. Are these all useful videos or are you looking for specific types of videos?