David Baszucki, Founder and CEO of Roblox: Trust Your Gut

[music] Dave, welcome back to the farm.

>> Thank you for having me.

>> Thanks for being here.

>> Really appreciate it >> and all of your prep and all of this.

>> Thank you.

>> Well, it's an it's an honor to have you here.

>> Yeah.

>> Uh we're really excited for it. I wanted

to take you back to where it all started. It's 1985 and this is the James

started. It's 1985 and this is the James Team look. Yeah, [laughter]

Team look. Yeah, [laughter] >> that's I think that's the Stanford yearbook in 1985.

>> Yeah, you're majoring in electrical engineering.

>> That's correct.

>> Yeah. Future to be determined.

>> That had no idea.

>> No idea.

>> That's correct. With a lot of mechanical engineering and design stuff, too.

>> Wow.

>> That was the foundation of the eyab back then, you know, David Kelly, all that.

So, I took a lot of that, too. So, I

just want to put the audience in this moment for a second. It's 1985, right?

First ever internet domain name was registered. Microsoft launched Windows

registered. Microsoft launched Windows 1.0. Tetris was invented.

1.0. Tetris was invented.

>> That's right.

>> And Back to the Future was a movie a year. Marty McFly. So, the look makes

year. Marty McFly. So, the look makes sense.

>> Yeah. No. Uh, the Mac had just come out.

I had a Mac Plus. Um, you know, you would switch floppies, uh, dot matrix printing, uh, just right on the edge of a lot of interesting consumer software stuff.

>> Yeah, it was an exciting time obviously to be building a company.

>> Did you know at that moment that you were born to be an entrepreneur? Is that

where it all started?

>> Um, not really. Although I would say um when I went home in my after my freshman or sophomore year to um figure out what to do for this summer, I I had no idea

what I was going to do that summer. And

some other student in my freshman dorm.

It's almost like in that um in the graduate where they say plastics, they said window cleaning. And and so someone said window cleaning to me. I went home

for that summer. My brother and I spun up this window cleaning business, right?

We just made buck like we we went knocked went knocking on doors in the evening. We just say, "Hey, you know, we

evening. We just say, "Hey, you know, we can come wash all your windows." And

just worked all summer. So there was a little bit of that. But I would say following the graduation, I actually I I think encountered a

difficult time where unrelated to the window cleaning, you know, it's like school's fun and you just follow the path and you do this and you do that.

And I think when I went to find a job, I was a little lost. Like it just didn't land in my lap in a way.

>> Yeah.

So that that is really difficult and a lot of students here are kind of grappling with what they're going to do after their two years here. So we'll get into that a bit.

>> Yeah. Yeah. Yeah.

>> So you start with the window cleaning company and then you actually naturally progressed to knowledge revolution the physics software simulation platform.

>> So that's [laughter] a natural progression. Um, it was I would say along the way, uh, I I tell a lot of young people this, there were two or three years of just the absolute

worst jobs in the world. So, two years of massive disappointment. Um, arguably

myself not figuring out how to go to companies I loved and kind of mosing all around. And, you can remember calling my

around. And, you can remember calling my dad and just not believing how terrible my predicament was. And um so after a

couple of these jobs I I took some time off and I would say flipped to a different mode which was intuitive mode.

And at that time it was the Apple 2 and then the Macintosh and educational software and I made a much more intuitive mood just like I'm going to get into educational software.

>> Wow.

>> So it's obviously an exciting time to be doing that. You co-founded the company

doing that. You co-founded the company with your brother Greg and later Eric Castle joined. he would go on to

Castle joined. he would go on to co-found Roblox with you.

>> That's right.

>> For all the folks here who are thinking about co-founders and starting a company right now. How did you go about that

right now. How did you go about that decision to start a company with those two people?

>> So, um I would say um a lot of luck actually like I think that the ecosystem today is so much more sophisticated about finding adjacent people and

amazing co-founders. on the on the

amazing co-founders. on the on the knowledge revolution side. I wrote the first vision of this thing called interactive physics as like a old physics laboratory simulator type thing

and it got some traction. Um and the way we found Eric is he saw a review in Mac user magazine literally and he was a physics major at Cornell University.

Absolutely brilliant. Um he called me up. He flew out from Cornell and he

up. He flew out from Cornell and he joined us. So like literally

joined us. So like literally serendipity.

>> Wow. a little bit of luck but also some openness to to what the world may bring you.

>> Yeah, absolutely. And and then he joined us in that company that the vision of that company was um educational software. We built this really amazing

software. We built this really amazing physics simulation stuff and for a while it was you know used with every textbook and all of that. And then we made I I would say typically made a terrible

decision um because a lot of people like to play with it and we're talking to all kinds of gaming companies and it was really fun but we made more of like a you know oh the technology is really

good let's get into mechanical engineering um which which was really um I would say non-intuitive not where our strengths were where more of our

technology was but that was successful as well. Yeah. Suffice it to say,

as well. Yeah. Suffice it to say, Knowledge Revolution was a big success.

Yeah.

>> Uh without raising any capital, you sold the business for $20 million, and this is in 1998.

>> That's right.

>> What did you learn from that first company building experience that you took forward with you?

Um well I learned to um you know I think one really big picture thing that um we learned we could really intuit it the customers for the educational software

business and and we were in touch with them. We could understand who they were.

them. We could understand who they were.

We could match the software exactly to these customers and we arguably could have intuited a continuation of those customers into what Roblox became. But

instead of using that intuition, we moved into this more once again mechanical engineering thing. So I think we we went back to that learning when we you know after we sold it and that kind

of led to the genesis of Roblox. So I

think I I think maybe a Stanford Biz School thing would be like if you have deep empathy connection with a certain customer like that's extremely valuable.

>> Yeah. That ends up being a huge part of the Roblox story and the connectivity you have with your community.

>> That's right.

>> Between the two companies, you, your wife Jan, and your three kids, Jan's here.

>> Yay.

>> Uh your three kids, I think, aged five, two, and maybe three months at the time.

>> That's right.

>> You picked up and you did a road trip across the country in a mobile home.

>> That's right.

>> Uh why? [laughter]

>> I felt I feel like this is one area where we were ahead of our time, right?

like now um in my neighborhood um depending on where you live I think there are nine um sprinter vans in my neighborhood like little RVs like it's

getting more and more hip but I I had um we had a lot of relatives in Canada up through Minnesota all of that it was just kind of a maybe a pallet cleanser

after all of that like just what's the most radical thing to do so got this big you know 40 foot RV towed a car went all over the place. Um, I've always been

into that kind of thing. I did a um for a while I I did a stint as a talk radio show host in Santa Cruz for a while. A

couple of those like things you get out of your system kind of thing. It was it was like an amazing time >> before you build the multi-billion dollar companies.

>> Yeah. Yeah.

>> Yeah. [laughter] Makes sense.

>> So, a lot of the students here today, we're we're on this kind of two-year metaphorical road trip ourselves. not

exactly going to Minnesota or hosting a radio show, but we're trying to figure out what's next in our careers. Yeah.

>> So, what did you learn from that break?

And how would you advise folks here on how they think about their >> I think I've learned it twice actually because I I can remember um in this terrible time right out of college trying to figure out what I was going to

do and rather than trusting my intuition or whatever. I can remember having a

or whatever. I can remember having a spreadsheet of nine potential careers and then all these metrics for each like well if I if if this is it really good

for this but it's not so good for this.

It was like a really weird way to try to figure out your career. And then I would say after the selling of knowledge revolution, which what was really

interesting is I got a little bit into um this mode of like, yeah, I just deserve to be a CEO of some new hot company kind of thing. Like I I just

sold this knowledge revolution, but like well it's interesting not everyone's calling me to be this CEO of this company. like what's going on? And then

company. like what's going on? And then

um and then I I I can remember um poking around on a few interesting areas that I I don't think were completely vision aligned for what I was interested in.

>> Like I I was just really interested in 3D simulation, consumer software, all of that kind of stuff. And here I was poking all the way on the periphery of

it. And I literally remember I I had

it. And I literally remember I I had what I felt like was some vision where what I was doing was just never going to work. And I had to go back to what I was

work. And I had to go back to what I was really fascinated with, which was I would say an extension of knowledge revolution. You know, bigger, better,

revolution. You know, bigger, better, multiplayer, consumer fun, all of that.

And that that was a much riskier direction, but it seemed like the direction I had to go.

>> Yeah. You were ahead of your time on the road trip. And then you're ahead of your

road trip. And then you're ahead of your time in building basically a metaverse, but this is in 2004. So after the sbatical, you guys

2004. So after the sbatical, you guys launched what was a initially a puzzle game called Dino Blocks.

>> That's right.

>> And then you took it down and relaunched it 9 months later in what eventually became Roblox.

>> So for the folks here that are, you know, launching their first version of their products, what advice do you have on when to to kill it and rebuild?

>> So a couple good things happened. We we

had had the knowledge revolution experience and and we realized even with this very simple 2D physics simulator it was right there in front of us. It's

like we got to be cloud, we got to be multiplayer, we got to be 3D, we got to be simulated. Like we think that's going

be simulated. Like we think that's going to be interesting. And and from that we were feeling a little bit like there could be this new category, call it the

metaverse, the holiday deck, snow crash, you name it. And and we were also thinking, you know, if we can build this thing out completely horizontally, get all the aspects of it, it's going to be

pretty interesting. It was a time when

pretty interesting. It was a time when we were not alone in these ideas. Second

Life was popping out. Their.com, if you can there there were a lot of these um companies trying to figure out what this thing was going to be. And um what we

did initially is um we had this whole version. I said, "Let's do a like a

version. I said, "Let's do a like a really subset of it. Let's just do a 3D puzzle thing along that thing because we want to get this live and it's going to take us a long time to get this puzzle

thing live and all of that." And so we do the puzzle thing and you know, we invited maybe a hundred people and two weeks later, okay, like no one's using

it. It's like what do we do? This is

it. It's like what do we do? This is

very depressing. you know, maybe maybe someone um you know, says, "Oh, I used it for 15 minutes. That was great." kind

of thing.

>> Yeah.

>> Um and so then um then it was a big decision. But I think intuitively we

decision. But I think intuitively we knew that was not the complete product.

The the complete complete product had to be multiplayer and people making stuff and people playing in it and all of that. So that was a that was a bit of a

that. So that was a that was a bit of a time like yeah honestly like we knew it, right? like we've got another nine

right? like we've got another nine months here or a year to build this thing. And so we just like shut it down,

thing. And so we just like shut it down, keep going, and relaunch a year later.

>> So if you're not getting the usage, you restart, you rebuild.

>> I I think in this case, it's a subtle nuance. If the big vision is right and

nuance. If the big vision is right and you can just see it th that we had enough fortitude to say this this big vision is absolutely spot-on, like we're

going to suck it up for another 9 to 12 months. And that big vision, just so

months. And that big vision, just so everybody here can understand what you were inv imagining in 2003, 2004, it's 3D, it's multiplayer, it's cloud-based, what made it fundamentally different

from gaming platforms and social media of that time?

>> Yeah, I think the um what's really fundamental about this is we felt that I think even at that time we felt there's going to be some natural technology

evolutions that are just inexurable.

this is going to just happen and and we've seen that before with the telegraph system. We've seen it before

telegraph system. We've seen it before with the telephone system at that time.

You know, we're starting to get into text. Video was still, you know, world

text. Video was still, you know, world fair kind of in everyone's imagination.

And we were, you know, gaming was starting to get a little bit multiplayer. So, we we knew, I think,

multiplayer. So, we we knew, I think, fundamentally that 3D simulation could be a communication platform. this could

be a way that people ultimately go to school or work together or those kind of things. And the gaming market was

things. And the gaming market was showing some of that technology, but it really wasn't um fully usergenerated. It

wasn't just you build whatever you want.

Um it was very pre-anned and and usergenerated is really it means self-service or or whatever you build can go live. So we were we were kind of seeing the combination of that you know

self-service whatever you build can go live and um I think that was a big um arguably Second Life was doing that there we were doing it I think in a much

more scalable kind of sustainable way.

>> So I'm trying to put myself in your shoes in 2004 2005 and you're telling people around you I'm building an online platform for people to build games for other users.

>> That's right. And I got to imagine people are like, "You're Dave, this is crazy. What are you doing here?" Uh, so

crazy. What are you doing here?" Uh, so how did you continue to forge on despite the doubters?

>> This was very different than the puzzle game. Um, there were a lot of doubters

game. Um, there were a lot of doubters because we we we were much more into let's build the whole horizontal product out. Um, which is cloud. We're going to

out. Um, which is cloud. We're going to use S3 that just came online. We're

going to have a development environment.

We're going to have all these clients.

We're going to have all of that stuff.

and and we had a little bit of the philosophy everything's going to be very rough but we're going to get the complete product to feel it out. So you

you are right everyone we would talk to or show it to this looks like crap or like this is like no or this has got to be prettier or all of that. Um but

something really magical happened and that is the day we launched the ability for people to create stuff. Um um and at that time we maybe had one or two

hundred users we we just knew we had hit virality. So after all of those other

virality. So after all of those other experiences we knew probably within four hours just like okay >> in four hours you could just see it.

>> That's right. We I would say we had had an earlier slightly viral period where we would buy users and we had maybe 500 users on the platform and we would talk to them. That was going pretty good. But

to them. That was going pretty good. But

then when we launched uh user created content and self-publishing, we saw within four hours just creation, creation, creation, creation. There were

four of us in an office and we kind of like that night knew, okay, >> wow.

>> Yeah, it's pretty good.

>> So it's four of you in an office in Menllo Park.

>> That's right. Let's maybe actually let's fast forward a few years. It's now maybe 2007 and and Roblox usage is up. You've

seen that morality, but the economy is starting to slow show some signs of weakness and you guys are starting to think about monetization in a more thoughtful way.

>> That's right.

>> So, that was an I think I've heard you say that's a near-death moment for the company.

>> That was one of um several. Right. So,

that would be a moment. Um

>> do you want to go through all of them or [laughter] >> Well, no. that I mean that that's one of the times as an entrepreneur where I can remember saying relative to everything

else I've done in my life this is the hardest thing ever like this is so hard and um but at the same time I can

remember us um we had a problem where users was growing like that but dollars per user was going like that so dollars was flat so users is going up dollars is

flat that means you're spending more and infrastructure, all of that. So, that's

a bad situation. And so, we I I think we would we did what the traditional response would be. The traditional

response is, "What did we break?" or um, "Oh my gosh, there's there's so there's so many ideas we can do to improve monetization. We have a list of 50 of

monetization. We have a list of 50 of them." And and so we did that initial

them." And and so we did that initial what's the bang for the buck of all the 50? Let's figure out everything that's

50? Let's figure out everything that's broken. um behind the scenes I would say

broken. um behind the scenes I would say just like the starting when we finally hit the complete product we actually somewhat knew that the big thing is

virtual economy and the big thing is let's build a perpetual motion machine where creators can earn money and sell stuff and all of that but that was

seemed like a little too hard and so we actually didn't consider it we spent three or four months churning on all this stuff but sure enough four months later just like it's not getting better.

We've done like 15 or 20 of these things and and then it's like, okay, you're right. We got to go do this big thing.

right. We got to go do this big thing.

But then that clarity, it kind of put a skip in our step. It's just like we're we're giving up on all of the things we broke. We're giving up on all the

broke. We're giving up on all the tactical stuff. We're just going to get

tactical stuff. We're just going to get go hardcore and build this full virtual economy as fast as we can. Um it had to be a complete product, right? digital

currency, developers having the flexibility to sell things for currency, users having the ability to buy currency, developers having the ability

to cash out on the currency, and some search and discovery acceleration. So

that um whoa uh some of the devs who are making money get more visibility. So we

we kind of felt it was like some closed loop operational amplifier. if we had all of these pieces, it might just go viral again. And sure enough, just like

viral again. And sure enough, just like when we had introduced Roblox Studio for the content, we put all these five components, like everyone's talking about them. We're going to launch. And

about them. We're going to launch. And

then in probably four hours, we said, "Yeah, it's going to work." Because

because every dev just says, "Oh my gosh, I could make a living. I'm going

to put some I'm going to sell a flashlight in my game or I'm going to sell a motorcycle motor scooter or something."

something." um they started bubbling to our discovery system. Users started buying

discovery system. Users started buying the virtual currency. It just like >> wow.

>> So we kind of relived that lesson all over again.

>> So it's it's one of the toughest moments clearly, but it's also this moment of tremendous resilience and it builds this muscle at Roblox to do the hard thing first.

>> That's right. and you guys start generating revenue in a big way, but still over the coming decade or so that

you guys face wave after wave of wouldbe Roblox killers. So that's Minecraft,

Roblox killers. So that's Minecraft, that's Fortnite.

>> That's right.

>> So what did you decide to do and critically what not to do as each competitor emerged?

>> Yeah, I can remember. Um, so the good news is when Minecraft came out, for example, we had in the pipeline and

behind the scenes arguably five to eight really big technical innovations. Like

we were thinking out of the box on this.

We were thinking autosharting in the cloud. We were thinking multi-device. We

cloud. We were thinking multi-device. We

were thinking super stable APIs. We were

thinking build it. We're thinking Lua language. We were thinking abstraction.

language. We were thinking abstraction.

We were thinking coming mobile. They

they were all kind of bubbling under the surface. But then yes, sure enough when

surface. But then yes, sure enough when Minecraft came out of the the woods just right and we kind of lived through I would say a two to five year Minecraft

craze just like that was there. But

behind the scenes, you know, their their pipeline of innovation was nothing like ours. Like we we had an experience four

ours. Like we we had an experience four weeks ago that 25 million people at the same time were playing and that that was all based on a sharded you know cloud

super pipeline really um infrastructure abstraction in a way and augmentation with our infra so we could burst into the cloud as well. Um but um they

weren't doing any of that charting kind of stuff. Um we we had um they split

of stuff. Um we we had um they split their platform. confirmed they had like

their platform. confirmed they had like a Java edition and a Windows thing. We

thought it was all developers, all stable APIs, like the DOSs prompt kind of thing. So, we knew we had that

of thing. So, we knew we had that coming. We knew we were going to try to

coming. We knew we were going to try to run the same thing on mobile as PC. So,

sure enough, like, you know, behind the scenes, all of that innovation came to play. And then the same story actually

play. And then the same story actually happened with Fortnite probably four or five years ago. Um, Fortnite came out of the blue, Epic Software just to be this

giant um gaming platform. the same thing like whoa this isn't fair like how come that thing's so big all of a sudden like we're working really hard but um also

arguably um once again a beautiful product um in the spirit of doing less more arguably really hard to manage both

um a C++ engine a store a product called Fortnite like trying to do all of these at the one time whereas we were just 100% all in on this kind of single

platform. So I think over time that that

platform. So I think over time that that focus for us and a lot of innovation once again brought us to where we are.

>> Yeah, you guys really doubled down on infrastructure and I know concurrency and your ability to serve concurrent users only continues to grow.

>> That's right.

>> You outlasted each of the fads and I want to just contextualize what a global behemoth Roblox is today. almost 400

million monthly users. Billions of hours spent on the platform each quarter.

Three out of every four American kids between the ages 9 and 12 plays Roblox regularly.

So at this scale, you're in a different league. You're competing for attention

league. You're competing for attention now with Meta, Tik Tok, YouTube. So, how

do you balance competing for attention with this societal concern around screen time in particular for young kids?

>> Yeah. So, that's really interesting and and one way to put it in context um from all of our investors out there is the global gaming market's about $190

billion market. We're probably running

billion market. We're probably running 3% of that through Roblox. So it it gives a sense of the potential scale

here um and what the opportunity is as well. Um a couple thoughts on this. Um

well. Um a couple thoughts on this. Um

in the midst of COVID, something really interesting happened. You know, we we

interesting happened. You know, we we had had this vision that we're not really a solo consumption platform.

We're not really an image sharing platform. We really are arguably the

platform. We really are arguably the future of the phone system or the future of how you stay in touch with someone else. And you know, just like there's

else. And you know, just like there's once the phone system, someday there'll be the hollow deck. We saw a little of that in the midst of COVID when people were not able to play outside or go see

their friends. A lot of parents

their friends. A lot of parents discovered Roblox is like the hollow deck to stay connected in COVID. That

was really a playing out of the thesis, I would say. And it it's a little bit I think a differentiator between what you might call a communication platform or a

solo consumption platform or an image sharing platform. There's some pretty

sharing platform. There's some pretty cool advantages of the communication platform. I think when I was um in high

platform. I think when I was um in high school, we had the phone, right? And on

a rainy day, we could call people up.

So, there's a little bit more of a social engagement aspect to that. I

think there's a little bit of a um STEM, art, design, production, business kind of aspect to that. But I would say at the same time, I mean, I've gotten um

clipped in certain news things like by the BBC where I said every parent should decide on their own screen time and like how could you say that, Dave Bazooki?

But we really we really do believe that and I I think we think um we'll work with parents to be this great communication platform which is I think a little different.

>> Yeah, I think it's a it's a very fair point. Not all screen time is created

point. Not all screen time is created equal and the types of activities kids are doing on Roblox is maybe a bit different.

>> Yeah.

>> I want to keep going on this uh this child user and and particularly safety >> challenge that you guys are addressing right now. So 11 days ago, the Texas

right now. So 11 days ago, the Texas Attorney General filed a lawsuit against Roblox. And in that lawsuit, he claims,

Roblox. And in that lawsuit, he claims, quote, "Robblocks markets itself as a safe digital space of creativity, but in reality, it has become a breeding ground

for predators.

This is one of a few lawsuits the company is facing.

So when engagement engagement and safety pull in different directions, which way do you go?" We always go safety and and arguably, you know, Ken Paxton in Texas,

the attorney generals around the country, senators we talked to, you know, whatever, they're really ultimately all trying to go for the same thing, which is child safety. And it's a

fairly sophisticated situation. A lot of parents right now um are not super aware that you give an 11year-old a phone and

there's a lot of apps that are unfiltered. images can be shared on

unfiltered. images can be shared on those apps. Um they're not monitoring

those apps. Um they're not monitoring for critical harms, all of that kind of thing. We're actually trying to educate

thing. We're actually trying to educate the industry about this. And so we are we actually think it's interesting that um device manufacturers don't have ages

on their devices. We um small world, right? Today I was on TV three three

right? Today I was on TV three three times talking about how we're going to innovate with age estimation for every user on our platform. So I do I'm very optimistic. ultimately about the

optimistic. ultimately about the industry learning about how to keep kids safe. And I think we have to design our

safe. And I think we have to design our product um assuming an 11-year-old is just handed a phone right now. And I

think we're innovating in a super good direction there.

>> So I want to I want to keep going on this because it's one thing to prioritize safety and then it's another to scale a company to multi,000 employees and have each of them prioritize safety.

>> That's right. So when each of your employees might be incentivized on engagement or revenue or other incentive structures in your company, how do you create a safety first culture

throughout?

>> Yeah, I think so. One of the things we've done and and I think it's an interesting way to architect the company is we're actually pretty um vertically architected rather than functionally

architected. And so when in a company

architected. And so when in a company that's functionally architected where there's just like so many engineers and so many product managers sometimes I

think it can be a little hard to say like how many people are working on one thing. We we we have much more moved to

thing. We we we have much more moved to we need to have completely dedicated safety stack. We need to know 100%

safety stack. We need to know 100% product 100% engineering 100% live ops.

who are all of those people and where are they and at the executive level make the decision on the the headcount like that is pure safety people I I would say

there's adjacent groups as well in our company both the user group and others that touch on it so I think it goes straight to the architecture of the company the size of the that safety

group and how we architect it >> so it's actually organizing specific teams so that they have this >> it's literally like we can measure the size of the company called Roblox Safety

within Roblox.

>> Wow. So now as a public company CEO, you deal with challenges like this one on safety and I'm sure many others. I would

say the safety one I I want to highlight I think it's actually an opportunity not a challenge because because we've started you know 6 through 60 and we're so focused on this including all we're

doing age estimation and other things long term we are going to have this amazing audience of 6 through 13 year olds that's very difficult for other

companies to kind of get to and a lot of our 15y olds and 20 year olds have been on Roblox since they were 12 so I actually think it's a long-term growth

growth opportunity as well as maybe a ethical moral responsibility.

>> I I like let's stick with it because I like that a lot. This this notion of responsibility and I know it's one of the core values at the company. You guys

talk about we are empowered and responsible for both the intended and unintended consequences of our actions.

So you talked about parents having a choice.

>> That's right.

Where does your responsibility as a platform begin and end?

>> Yeah, I think in this case our responsibility goes beyond financial or legal. It's like for me obviously

legal. It's like for me obviously financial gains in making Roblox awesome. Um legal we typically on the

awesome. Um legal we typically on the legal side we want to be way ahead of that. Like we want to be building stuff

that. Like we want to be building stuff that might get wrapped into law 5 years later. We don't want to be waiting. Um,

later. We don't want to be waiting. Um,

once again, what we're doing with age estimation, there's no law that says we have to do that. We're just doing that right now. But I do think there's

right now. But I do think there's another element in running the company and just um what's a fun type of company to run and Roblox is a pretty fun

company to run and it's fun to sleep well at night and have a good time. So,

so running the company on that third component which is just what feels values aligned is also very important.

Well, it's a fun company to run. You've

been running it for more than 20 years now, and you went from three employees in a room in Menllo to now thousands of employees in global impact. What's the

biggest change you had to make in yourself personally to scale from being a founder to a public company CEO?

>> That's a really big question, right?

Like my job is like a constant rearchitecture of myself really. Like

it's it's a lot. And I feel um it hasn't just been a business journey. It is like a personal journey. I think um learning

um where to put your strengths, learning not what you're not good at, adjusting um I've made a lot of adjustments. I

think in in the early time of a company, you know, you can be yelling and screaming and going crazy and upset and all that. as you get to bigger scale, I

all that. as you get to bigger scale, I think you got to be a little more zened out and kind of more jiu-jitsu in your

application of um kind of your making things happen. I think um one of the

things happen. I think um one of the biggest things for me is actually trying to be more intuitive on people and what we do as the company and less analytical

because my intuition is if anything I think my intuition tends to be pretty good and I've kind of clamped it in the past. So focus on that too.

past. So focus on that too.

>> What does it look like to be zen at a company that has thousands of employees?

I think I I can I think um if I'm in a meeting with 15 people and within four seconds we're reviewing some product thing and I'm just saying that is wrong,

that is not going to scale. Um

thoughtfully thinking through how am I going to optimize the delivery of this message? [laughter]

message? [laughter] >> So the intuition is there but you don't let it speak freely at all.

>> That's correct. like like actually I'm going to look at everyone in this room and rather than say that sucks. Um

[laughter] I'm going to in five words try to optimize the change all all of that.

>> Well, it it wouldn't be 2025 if I didn't transition us to talking about AI.

>> Yeah, >> we got to do it.

>> Um >> let's do it.

>> Let's do it. every company is racing to adopt AI and they're trying to convince investors that their businesses will improve as a result of AI. Um, so as you

think about Roblox's unique advantages, you talked about some of them earlier.

Um, what makes you guys uniquely positioned to take advantage of AI across your business?

>> It's really exciting. Um I think we're there's obviously a lot of froth in the true frontier three or four companies that are going for whatever super

intelligence whatever but I think there's a whole other dimension out there which is just radically accelerating your business. Um and for

our business we're running we have a huge AI team. We we're running over 400 different models that we've built. um

everything text safety, voice safety, search and discovery, 3D generation, you name it. Super broad. What's interesting

name it. Super broad. What's interesting

for us in AI is I would say there's there's been maybe three or four waves of AI for our company. Wave one would be behind the scenes, everything's getting better without the customer really

knowing it. Safety filtering, all of

knowing it. Safety filtering, all of that. Wave two is starting to see early

that. Wave two is starting to see early signs of generative AI. um the dream not just for a creator, but anyone walking around in a world in Roblox. Talk about

the clothing you'd like to magically appear or talking about the game you'd like to magically build, which I think is going to be transformative. We're

going to see new types of games. I think

the the next one though beyond that is going to start to be, you know, virtual avatars doppelgangers assistants all of those kind of things. And I I do

think the architecture of these types of platforms will be, you know, if if I so choose, there will be a a values aligned version of me that I could say, "Go play

with your kids for an hour because I can't do it." And they'd kind of be pretty close. Or there would be someone

pretty close. Or there would be someone who who I could give some commands to and they would go do stuff. So I think the future for us will will in addition

to 3D generation be like literally you know virtual humans in Roblox creation type thing. And I think that area is

type thing. And I think that area is super hot. We have people building

super hot. We have people building robots. We have people trying to do

robots. We have people trying to do world models. We have all of that kind

world models. We have all of that kind of stuff. Roblox has one very unique

of stuff. Roblox has one very unique thing and that's about 11 billion hours of human interaction data every month.

It's not just video. It's um what their faces are doing, where their arms are moving, what they're doing in a 3D environment, and that's a huge um repository of training data that we

would never sell or leak, but use to make the platform better. So, I think there's a big opportunity for us there.

>> Yeah, it's an incredible amount of engagement and creates this flywheel for you guys.

>> That's right. Um, on that second wave, you mentioned developers, and I'm curious if do you have do you have any concerns about what AI will mean for developers who are really the lifeblood

of Roblox's thriving economy. They build

the games that people play.

>> I'm a little more on the optimistic side, aka Industrial Revolution. It's

not like all of the people who were building horse buggies lost their jobs, that there's like some new types of industry for them to do and accelerate and do better. So, I I think what we

will initially see with our creators is um higher quality, more amazing type creations. Um maybe a shifting in the

creations. Um maybe a shifting in the landscape, but I'm overall optimistic that we're not going to have an unemployment crisis. I think we're going

unemployment crisis. I think we're going to see new types of creative work.

Well, that optimism is exciting. It

means that you're probably going to continue to run the company for a long time. Um, I have one more question for

time. Um, I have one more question for you before I turn it to some audience Q&A.

Stepping back as you reflect on the journey thus far. Any regrets?

>> No.

>> None.

>> I would say mistakes left and right.

>> No regrets.

>> A regret would be like if I had to replay my life, I would have changed it.

I wouldn't.

Lots of mistakes, lots of learning.

>> Tell us about one.

>> Oh, just I I would say along the way all of the places where we've taken shortcuts, not trusted our intuition, you know, could have done things better.

I would say if anything, not um not there were times when we could have leaned into our vision even harder, even been more aggressive kind of things. Um

I think there have been um but I I think our values are representative of that learning like take the long view do the hard stuff

first build systems um try to educate people you know short-term hacks aren't going to work um that kind of stuff and I I think um now I think we're in a

great opportunity really to take the long view on the AI stuff as well.

Well, we'll now turn it over to the audience where three students have prepared questions for you. Uh, I'll

turn it over now.

>> Hello. Uh, I'm Mensuk. I'm from the class of 2020, uh, 2027. I'm an MBA 1.

Uh, thank you so much for sharing your stories. I think it's incredibly

stories. I think it's incredibly inspiring. Um I I do have a question for

inspiring. Um I I do have a question for you regarding as we scale as you scale your service as we implement AI as more users come on board and as simulation

fidelity and as we increase to use more compute what are your thoughts on scaling infrastructure and how do you see yourself reconciling with the demand for compute and more service and the quality of service as we go forward

thank you >> yeah we have this term we use inside the company called infrastructure adjacency and it's really interesting thing to think about in that um we build we have

a lot of data centers. We have like 40 data centers. Um at the same time on a

data centers. Um at the same time on a Saturday at sometimes what we typically would cover will go up by a factor of three or four. So we have our own bare

metal um data center but we have cloud partners like AWS and Azure and GCP where we can turn on massive extra

capacity. And when we run the math, it's

capacity. And when we run the math, it's actually pretty optimal. Like we want the reliability of all of our own stuff most of the time, but we'd like to be

two or three times bigger for two or three hours. The um this is going to

three hours. The um this is going to continue. So I would say um we we we

continue. So I would say um we we we feel very lucky that we are not dependent on the cloud providers.

They're typically three times more expensive than our own stuff both on CPUs, on GPUs, all of that. We're

continuing to build this out, but we'll always have this adjacent um cloud bursting capability. And I I think I

bursting capability. And I I think I could see us um you know, if we imagine 10% of the gaming space, 3x bigger, I think we can handle that. But but I I think the other thing is a little secret

of Roblox is we are kind of an infrastructure company.

>> Thank you for being here. My name is Lexi uh NBA class of 2026, NBA 2 also from gaming industry. Uh my question is previous in interview you mentioned

gaming is still in the prehistory era and so just curious about for Roblox like what's the plan for the next five years what should we expect given on the statements

>> like that question right because I I do feel um get there's probably at least three generations of gaming technology and most of the market is arguably could

be in the first generation and and there is um most games today um that are beautiful in high fidelity on a PC don't run on a phone, which is kind of weird.

Most games um don't stream in 3D in a low latency way. They either stream video, which hasn't worked so well, or they're downloaded, some as big as 200

gigabytes. So this the magic combination

gigabytes. So this the magic combination of exact same experience low-end 2 gigabyte RAM Android phone and gaming PC

instantly joining around the world um with low latency is is where I think the next gen is. So that in our developer conference that's kind of a huge architectural thing we've been talking

about but there's potential another even architecture beyond that which is what comes to bear with essentially massive AI acceleration and upsampling and all

of that. So it still is kind of in the I

of that. So it still is kind of in the I would say stone age generally.

>> Hi my name is Marshall. I'm an NBA one here class of 27. Um I was first introduced to Roblox back in 2007. I was

in about the fourth grade at the time.

So, I'm curious how what was your original vision for what kids would kind of take away from the platform and how has that evolved over time?

>> We um so a couple things that are really fun about this product is number one, we were not heavily gamers. We were I would think more thinking if we built this

physically simulated cloud stuff that people could build on there'd be a lot of infinite possibility and we'd see a lot of creativity and that that I think actually has come to be true and not

being you know constrained as like we're game developers but we're more building this 3D cloud platform I think has been one of the most satisfying aspects of

the whole thing.

Thank you to the students for those questions. Uh Dave, before we let you

questions. Uh Dave, before we let you go, we have to dive into a view from the top tradition. It's our rapid fire

top tradition. It's our rapid fire segment.

>> Yes.

>> All right, let's do it.

>> Never seen somebody so fired up.

>> Like a roar sack. Chess, right?

>> Yeah. Exactly. [laughter]

>> Okay.

>> Okay. You ready?

>> Yeah.

>> Favorite childhood game.

>> Wow. I'm going to say chess. Me and my brother >> on chess.com now at GSP. No, just like we used to like a physical chess board.

>> Okay, fair enough.

>> And then I would say outside um interesting inside chess outside like pretending we were in the army kind of thing.

>> Oh yeah, [laughter] kind of similar. Yeah. Um if you weren't running Roblox, what would you be doing?

>> Trying to build a Roblox competitor.

[laughter] >> Is that a real answer?

>> I think that's real. Yeah. Yeah. Uh,

favorite destination on that iconic family road trip.

>> Whoa. Uh, we went everywhere. I would

say Mount Rushmore.

>> Mount Rushmore.

>> Mount Rushmore. Just because it was a beautiful night and uh the magnitude of it and they had a cheesy mo movie. Um,

but the whole thing just came together for me. It was a beautiful night.

for me. It was a beautiful night.

>> And favorite memory as a Stanford undergrad?

Uh, we were just talking about what the football game like I'm riffing back to we were talking about what it was like to go to a Stanford football game in 1980s.

You could bring a keg into the stadium.

[laughter] >> You said the band was pretty runchy back then.

>> The band was arguably at that time even considered politically incorrect. So

>> sounds fun.

>> Yeah, it was. [snorts] Well, final reflective question before we let you go properly.

>> It's not a rapid fire question.

>> Okay, I'm gonna think about this one.

>> Yeah.

>> What is the best piece of leadership advice you have received?

>> That that makes me think there's many pieces of leadership advice that were bad advice. And so I actually two-part

bad advice. And so I actually two-part answer. A lot a lot of my development

answer. A lot a lot of my development has been trying to over time ignore advice I've been given. Um and then I

would say it uh the best advice probably 80 people you know give it to you when you're just like having a rough time and they just say trust your gut

and many people have given me that.

>> Dave, thank you for being here. Ladies

and gentlemen, Dave Bazooki. [applause]

Appreciate it.

[music]