How AI Will Transform Roblox Games into Photorealistic Worlds | CEO David Baszucki
In a Nutshell
Roblox CEO David Baszucki envisions AI transforming Roblox into a photorealistic "holodeck" supporting 10,000-player multiplayer worlds with advanced physics, acoustic simulation, vector-stored history of 13 billion monthly hours, and privacy-compliant NPCs trained as virtual doppelgangers for gaming, social simulation, and coaching. AI will supercharge Roblox Studio via cloud-based generative tools, on-demand 3D upsampling, agentic testing, and procedural assets, enabling creators to build high-fidelity experiences faster while raising quality expectations. Long-term goals include 10% of global gaming (300M DAUs), hybrid AI stacks blending world models with native multiplayer sync, and compounding innovations that endure 20-40 years.
These notes were generated by AI and may contain inaccuracies.
What proportion of you think your children will have at least one serious relationship with an AI throughout their lifetime? I think zero right now. I don't think we're close to crossing the human AI barrier. But things are changing every day. We have to incorporate that. There's weird universal truths of things that stick around for 40 years and you kind of have to blend them together. If we build Roblox right, it might be the kind of thing that's kicking around in 40 years.
Hi listeners, welcome back to No Priors. Today Allad and I are here with Dave Baszucki, the founder and CEO of Roblox, the 3D immersive world where more than 150 million users come every day to hang out. We talked with Dave about the future of AI and gaming, their investments in AI, how NPCs are going to change and really transform Roblox's studio, one of the largest coding platforms out there, and their 20-year mission to build the holodeck. Welcome, Dave.
Two-step on this one. First, there is a Roblox business plan PowerPoint deck from almost 20 years ago that pretends a new category. One could call it the category of human coexperience. It's got the sizes of a bunch of companies from 20 years ago. So you can see social networking companies, you can see YouTube video, you can see toys, and you can see all of that. What it imagined is really a new category. Some have called it the metaverse. Some have called it the holodeck which is really the ultimate high-fidelity simulation where people can come together and do stuff. Feels like Ready Player One the movie. Exactly. And so we do have that business plan slide. What we had always imagined is if this high fidelity space was backed by physics simulation and reality simulation, you'd be able to do stuff. You'd be able to build a car, put wheels on it, drive it around. You'd be able to go to a birthday party and blow out the candles. You'd be able to chop trees down and make a house of it. That was literally the genesis of Roblox. And here we are 20 years later. There's so much more to do.
One vision of AI is how do you superpower the creation of that holodeck? And how do you get it photorealistic as soon as possible? How do you get 10,000 people in it instead of a hundred? How do you do acoustic simulation that sounds realistic with 10,000 people? AI is super interesting about just trying to get a vision that's been around for 20 years and has arguably been in sci-fi once again, Snow Crash, holodeck forever.
There's another interesting vision: what's the future of just a single person by themselves dreaming what they want to dream and we sometimes call that real time dreaming category. It's a category that we can see a bit of trappings of with short form video. Someone's 2 a.m. in their bed doom scrolling. It's reacting to dwell time. It's reacting to what you favorite. And you're kind of getting a little bit of a pre-built dream that I think we sometimes imagine could go all the way to the famous Tom Cruise movie Vanilla Sky where he literally real time dreamed for several years in an imaginary universe where everyone around him was primarily an NPC and he didn't even know he was doing it. That's more world building as content. That is exactly right and so if you put those at two extremes one is communication platform high-fidelity with people you know the other is real time dreaming where everyone's an NPC everything in between is possible and I think we see everything in between ultimately as being possible and we may see weird product categories we never imagined. Does the text prompt turn into a voice prompt and does it automatically turn into a video prompt if there's like I want to learn French video? How you're in a French cafe, start talking to an NPC? There's a wide range of products that start to get into this space we're interested in, which is kind of that 3D reality space.
Do you think that same space will exist for both consumer and business applications? Because people also talk about this in the context of you're instead of doing a zoom why don't you go into a shared space with someone and meet with that what degree of your life becomes virtual. On the communication side people just want more and more fidelity. We can go back to paintings on the wall of a cave. People want more fidelity. We have smoke signals. We have the mail system. We have the Pony Express. We have the telegraph system. We have the phone system. We have arguably the COVID initiated video. We also did have the COVID initiated 3D experience on Roblox actually and we found side-by-side video. A lot of young people used Roblox as an early test case of how to co-experience. As the technology for multiplayer 4D simulation gets better and more photorealistic, it's almost going to be like video is the downsampling. And it's like if we were having this 4D immersive communication, we'd say go to legacy analog mode and make it look like Zoom basically. And we could always do that, but we'd be able to do things we could never do on Zoom. You'd say, "Hey, let's pop up and go walk around my office and I want to show you something. Come and follow me."
The other thing for business is there's a few things that 4D and I use 4D because it's not just 3D shape, but its function in a way will provide that video does not provide is acoustical things. And there's some real tricky physics problems when you're having a company meeting of a thousand people. If we try to do it on Zoom, we have like a bunch of squares. What's who's live? Who's not live? How do we mix the sound? Which is very tricky. If we do it in a three-dimensional simulation, I can hear everyone with attenuation based on how far they are. And it can actually be a much more natural experience. It's like I walk closer to you and I hear you. And that may actually be a better human interface paradigm. There's one final technical thing that's going to be very very difficult is when people try to sing happy birthday to you together and because no matter whether you're on Zoom or whether you're in a 3D simulation, you're hearing them like 30 milliseconds later. What we're going to find is people are actually allowing each one of us to sing happy birthday with a time-forward extrapolation of each other and then remixing it. For the business side I do think ultimately we're going to see some interesting conferencing solutions.
Roblox has this amazing wild story with physics simulation and physics engine as part of the birth of Roblox. But then today or for the last decade I think of this as a immersive world where people have their first coding experiences. They build amazing games. They interact with each other. I don't think of that world as being super focused on realism. We're pushing really hard on the physics simulation capability and at the same time we have this incredible free market that we can't control on content generation. We are just started talking about full acoustic simulation. Approximate attenuation and all of this kind of things and echoes that's going to start happening on the physics simulator. It's behind the scenes gotten better and better and better. It'll keep getting better. The diversity of developers are beyond our expectation. Some of the things that have gone viral, Dress to Impress, for example, Grow a Garden are less physical. But they have leveraged cloud capabilities we've built side by side. Grow a Garden was essentially able to build an experience where things keep growing even when you're not there by leveraging a lot of the cloud persistence we provided. Dress to Impress was able to do some very interesting things with how they run contests and layer clothing on. The specification for our product could be we just want to simulate the real world with 10,000 people. And the more we go in that direction, the more we'll just see creators using it in different ways.
How much attention have you been paying to world models and what's been happening in the AI world in terms of how people are shifting how they do physics simulation? We're really watching it. The biggest thing we have focused on and we'll continue to focus on is how do we get that 10,000 player multiplayer type experience and how do we fall back to 100 or 10 or one? We feel it's super robust to build a communication platform and that's part of our goal of getting to 10% of gaming. The technology for this we're going to start unraveling it over the next 5 to 10 years is what's the most efficient way to synchronize the state of 10,000 people and when it's a deep infrastructure problem versus necessarily a physics. How do we synchronize the state and the memory over the last 5 hours of 10,000 people? Where have they been? Where have they been in 3D space? What have they been doing? The world model thing is very exciting because they're storing memory literally video latents. A minute of frames. It's showing super early promise. The technical revolution is going to be what's the ultimate synchronized state for 10,000 people. Is it in video latents? Is it in much more native 3D format? Is it a hybrid new discovery of some 3D video latent space? To be determined. The format we're using, part of the vision for us is to ultimately store the history of everything on Roblox and to store it not raster like video, but store it vector, which means the ability to play back anything that's ever happened on Roblox. That's 13 billion hours a month.
What do you think that does for you? The history of your world. There's a sci-fi book by Arthur C. Clarke called The Light of Other Days, which talks about the fantasy of what would happen if we had infinite playback in our world. It's like a real social problem. Because if everything we've ever done can be played back, actually in that sci-fi book, it's not just everything could be played back, but you could look at any playback. And so everything's completely transparent. There's no private conversations. It actually delves into that. For us, we would then use it very thoughtfully and privacy compliant and judiciously. For very simple things, you can imagine there was a safety incident. Let's go back and put five cameras and listen to the audio and see what happened and make a good judgment call. There's a really special thing that maybe I did with someone in my family that's really special given I have access to my history. Go play that back and possibly because it's 3D vector data, not raster data, reshoot it from make a cool video of the best moment. Where it ultimately ends up for us is there's a huge push now for people to find kind of hybrid video ASDW human interaction data. People are starting to train on video with keyboard stuff and like where do I find 200 million hours of this data? The data we have which is 13 billion hours a month can be reproduced from any camera angle and can interact with the 3D space. So it's very powerful data. It leads to a really interesting computer science program idea around how to create great NPCs that are more than just an LLM.
Go check out on X my feed. I think the last day or two we released a bunch of native video demos. What we are starting to train is NPCs based on a combination of all of the data we have in a privacy compliant way that are starting to navigate and starting to play games. Challenge level one, everyone has access to an NPC that can get pretty good at playing any Roblox game. Challenge level two is by watching if I so opt in in a privacy compliant way, my behaviors, the way I act, literally my gestures, the way I look at things, the way I talk. Would we allow you to have your own virtual doppelganger if you wanted? That's fascinating. Number three would be if I have a good virtual doppelganger just as agentic is all the buzz in other areas. Is there a simple user interface for agentic virtual doppelgangers? That starts to get interesting because one could imagine sending their virtual self out to go do something. It could be my son or my daughter wants to play with me. I'm having to do some work. Could my virtual doppelganger fill in for 15 minutes? Sure, that's great. That could be really interesting. There's a lot of other things just like we're going to see agentic space in the 2D and the work space and the productivity area. You're basically using these 13 billion hours a month to train a Roblox privacy compliant model. Will never share the data. And then over time, no ship date, make that customizable as your own virtual doppelganger or for creators with the addition of a prompt, upload Benjamin Franklin's history or it's already embedded in an LLM and just say, "Hey, I want Ben Franklin." And I want a happy Ben Franklin.
Are there experiences that you imagine creators making with these NPCs that are really different? There's a lot of standard game NPCs that have done such a good job. Like Grand Theft Auto, they've been working on this forever. That is a very rich world with a lot of very high performance NPC behavior imagining someday many people can create them. To make that happen these platforms have to be very cloud connected platforms rather than running local to spin up inference on all of these things. You could imagine a very young creator saying, "I'd like to build an American history thing where you go and talk to Benjamin Franklin." So, we would want to enable both of those. The transition is like in self-driving. where you went from mapping out heuristics in subbehaviors and sort of fine detail to capture edge cases to just doing end to end models for deep learning and it sounds like you're basically considering doing the same thing for NPCs. That's the exciting computer science capability is rather than all of these wonderful techniques people have used decision trees and this getting to a purely more general modeling solution would be really fascinating.
What do you think the role of like a Roblox game designer or any game designer is five years out from now? Do you think it's different? I'm optimistic that typically industrial revolution or whatever revolution there's a thought that there won't be any jobs after this. I'm a little bit more optimistic that a new class of things will show up and there's a huge opportunity for more creators. I'm not to the point where I believe what we call AI slop today is going to take over. The human touch has got a long runway on it. More diversity or quality of experiences supplemented by AI. The role will be more leveraged and the quality coming out of five people will be astounding but we'll come to expect that quality so quality will go up. Iteration on testing will go up. Where we see platforms like Roblox going is just as people are starting to spin up agents to develop code on Roblox platform against the Roblox cloud. You'll be able to go away for 24 hours and your agents are going to be like tweaking and testing, spinning up a Roblox experience, sending 20 NPCs into them on various client emulators. Here's an NPC on a phone. There's an NPC here tuning the game. You'll feel more power in what you're building.
There's a set of research around world models that is just trying to directly generate game experiences with like there's no physics engine underneath like you don't control the mechanisms it's just video. What's your view on that? It's like a super exciting interesting thing to think about. The higher the fidelity is memory stored in video latents or does there need to be another research breakthrough where memory is ultimately stored in some new latent space that's 3D or four dimension. But the fidelity is amazing. The way we would see the ultimate architectures of these platforms is probably for a while not an all-in-one thing. We would probably see some hyper efficient thousand person synchronization state engine. We might also see some of the photo realism not being done server side but being done in intermediate spaces or client side as well. And we may see multi-step pipelines where there's the raw framework, there's 3D upsampling, then there's local 2D upsampling. People are putting together a bunch of pieces, dedicated NPC capability, dedicated synchronization capability, 3D upsampling, 2D upsampling, and potentially world model stuff. You can imagine that being used even in its early state today rather than maybe a gameplay state where rather than making a video, you talk about the video while you walk around and you kind of say, "Put me in a western world. I want to go left. I want to go right. Add a few horses." And that video actually goes to a multiplayer experience. Does short form video does someone hit it just like they did with short form video into a highly retentive like local dreaming product that could just react to what you're doing. It's still early for both of those but it's a big possibility. We are doubled down on a hybrid multi-AI tech stack that supports multiplayer natively.
One of the ideas that's internationally very interesting has been the idea that you could generate serial micro dramas as a form that is not yet that popular in the west but seems like a good format for the technology. Micro dramas. I've watched a few. It's really interesting. Some entrepreneurs tried to do that five or 10 years ago then they naturally emerged kind of thing. It's interesting to think if those micro dramas are easier to produce. They don't need live actors, but what is the plotline that really resonates. We'll see with that. You can mass test that at scale if you're using AI, right? You can actually a very large number of iterations.
This kind of one of the arguments people make around the Fermi paradox of like why haven't we seen alien life? And the idea is there's alien life. They are just stuck in these virtual worlds now where it's fun the real world. What's the famous quote where is everyone. One option is they're in some virtual landscape that's more compelling than the real world and so therefore there's no reason to travel the universe or it could be we could really start conjecturing what's the future evolution is it human or machine life and all of that and once it gets to maybe human machine life what happens. The subsumption. That's the Vanilla Sky thing. Or the optimistic thing is more the holodeck thing like everyone's hanging out with other people rather than a simulation of themselves.
Are there experiences that you imagine from a communication or a social perspective that you think are going to be different or new or that you're going to have real insights into from these really great NPCs? You've seen what has happened with the rise of AI companions and how engaging they are given they're just text interfaces. There's some fun things that like there's a Black Mirror dating episode where they have a lot of virtual doppelganger power. Roblox is not doing dating. We're just talking about this right now, 18 plus ID verified, all of that. Let's think about a future in the Black Mirror episode where you literally can spawn the life of a thousand virtual doppelgangers. That thousand virtual doppelgangers, you're on your new dating app in three minutes. Someone else's virtual doppelgangers live a thousand virtual lives with your virtual doppelganger. And then they say, "What's the success rate?" And when the thousand virtual doppelgangers say 98% of the time we had a good life, then you say, "Oh, we should go on a date or something." Social simulation. There's a lot of things we think are going to be really hard to project. One of the things we would focus on is the more we can just build raw high performance infrastructure, have AI as a service, have all of these things capable in a cloud, have the same thing run on a phone as a computer, have it auto translate. Hopefully, we would discover some of these things. I'm really optimistic about that given the engagement that people see with these companions as just a text interface with the beginning of memory. Versus if you're Roblox, you have embodiment, you have cloud persistence. And it's just going to make the characters much more interesting very quickly.
Thinking through how we would embody memory for NPCs, which is more than just a prompt. Any NPC on Roblox because we have recording of the whole thing, they actually have access to everything they've ever said or done in a very lean format. So, they can go back through that, retrain on that. That hopefully would be a native part of the platform. There's a lot of interesting engagement with these types of companions and imagining fully 3D versions will be like pretty fascinating. A friend of mine basically asks at dinner parties, he runs a well-known AI company and his sort of controversial question for the group is often, "What proportion of you
Interviewer: Think your children will have at least one serious relationship with an AI like throughout their lifetime?
David Baszucki: Zero right now. We're not close to crossing the human AI barrier. Even if superficial, with text behavior and video virtual worlds, I can imagine a real-time coach therapist in your earbud hanging out all the time.
Interviewer: For less invasive over-your-shoulder coaching, that's already a major use case for ChatGPT and other platforms, like mental health advice or coaching.
David Baszucki: That will be very valuable.
Interviewer: Why hasn't cheaper asset creation with generative tools changed gaming industry dynamics yet?
David Baszucki: No matter how cheap assets are to create, consumer quality expectations rise at the same velocity. What was not possible before becomes expected.
Gaming industry technology must move to cloud vertically integrated asset management; can't ship on DVD or giant downloads. Roblox launched dynamic LOD for textures, meshes on the way. Every asset (sound, 3D object, image) can exist in various formats.
- Traditional: texture, mesh, audio file.
- Future: AI prompt generated procedurally on demand.
Both AI-enabled. Requires cloud connected for full LOD range: low LOD for low-end devices, 4K for high-end, on-demand AI generation. Demo on 3D upsampling: primitive Roblox experience augmented with prompt like "make it medieval and realistic," assets auto-upsample in 3D in cloud for free on demand.
Roblox Studio is one of the biggest development environments. Users glue together models in cloud code to control Studio.
Roblox Studio runs native assistant and native code engine, following industry best practices. Needs environmental generation component: prompt (text, image, video, world model) to primitive 3D skeleton, iterate to fully functional game. Run side-by-side with AI workflows out of the box.
Cloud enables spinning up jobs, agents, running test plans, grinding away.
Top Roblox creators earn tens and 30s plus million a year. Healthy sign: creator #1 average yearly revenue rises faster than creator #1, out to creator #1000 making a living.
More competitive, pattern of live ops: cloud enables daily/weekly updates like good websites. Users love updates, keeps games fresh.
Discovery approach: completely transparent algorithm, pressure to make good content. Transparent discovery keeps pressure on creators.
Top 20 experiences all really good, vying for #1 (not true 3-4 years ago). Content distribution richer, less spiky.
Interviewer: High conviction long-term oriented leader with 20-year Roblox plan, all-in on AI. How square long-term view with fast-changing environment?
David Baszucki: Pair take the long view (one of four big values) with get stuff done. Like Gartner magic quadrant: rapid iteration inside company.
Ship exciting product in 6 months by breaking into weekly ships toward target. AI team, facial age estimation, safety team work weekly; AI daily.
6-12 month vision aligned, weekly iterations move quickly. Roblox vision: build the holiday – 10,000 people, multi-player realtime environment modification, NPCs photorealistic. Stable spec for 20 years. AI moves you closer.
Two years ago less crisp (e.g., billion DAUs). Now: 10% of global gaming content (~300 million DAUs, 20 billion engagement hours, 3x current). Forensic: USA higher than 10%. Clarifying for product teams. Know 3x for planning, 10x to sleep at night.
Roblox rethought hiring for values-aligned problem solvers who are creative.
Acquired Embellis 5 years ago: 3D assessment tooling by scientists (SAT test involved) for problem solvers/creatives. Tuned for new college grad intern program.
Run 10-50-60k applicants through fair, non-biased assessment. High confidence hires mentor future leaders: building wave of 100, 200, 300, 400 people.
Fairness testing done. Correlation: traditional universities not necessarily top talent. Found talent in community college, small Midwestern engineering schools. Ignore university signal.
Tests public: roadblocks – 3D problems like programming a factory, geometric language for robot (turtle graphics).
Interviewer: Long-term predictor, something in AI you believe 5 years out that not everybody believes?
David Baszucki: Typically more people believe more than I do. Power of time/compounding over amazing invention. Amazed by graphical user interface or Microsoft Excel (played in 80s, 40 years later not different).
Blend daily AI changes with universal truths that stick 40 years. If build roadblocks right, could kick around 40 years.
Keep No Priors: AI, Machine Learning, Tech, & Startups in your library
Save the videos and channels worth coming back to, and find them again in one place.
