Mark Zuckerberg
At the end of last year, we're like, all right, we are quite excited about the progress. We haven't seen the models yet. You know, we're going to write this check for like, I don't know, like $100 billion, $200 billion. I'm like, Alex, are you sure this is going to work? It's like—
John Collison
Cheers.
Mark Zuckerberg
Cheers.
John Collison
Yeah.
Mark Zuckerberg
Cheers.
Alexandr Wang
Cheers.
Mark Zuckerberg
Cheers.
John Collison
So, give me the update on Muse. If you were looking at the third-party reports, you would see that this might be the fastest-growing agent or AI app that I think has existed yet.
John Collison
What are your guys' personal favorite Muse use cases, either yours or other people's?
Alexandr Wang
I was actually talking about this at dinner yesterday. My favorite thing is I have my Muse, whose name is Oiler, constantly source increasingly obscure teas from increasingly obscure vendors of tea. So at this point he's, on a weekly basis, my Muse is buying some random tea from some farm in Taiwan or some farm in Japan and sourcing it over and bringing it back.
John Collison
What's your favorite Muse use case?
Mark Zuckerberg
I don't know. I mean, I use it for a ton of stuff. I still like... My favorite both use case and probably eval for the product is having it coach me in MMA. So I put cameras in my gym, and I basically... It watches the cameras when I'm in there, and it gives me feedback in between the rounds, which is a very hard problem, it turns out, because not only does it have to... It basically has to process all the video quickly enough because you only get 1 minute in between the rounds.
John Collison
Oh, it coaches you between the rounds?
Mark Zuckerberg
Yeah. I want feedback.
John Collison
I want feedback on how to win. You know what I mean?
Mark Zuckerberg
It's like, come on.
John Collison
Did you have to instruct it... I've wondered about this too, because looking at videos is a really good AI use case. It feels like you can't just give it a multi-gigabyte video file. So do you tell it how to slice stills out of it?
Mark Zuckerberg
No, it basically... Yeah, it pulls stills. I think it kind of figured that out by itself. Um, it turns out that Muse has pretty good native understanding of MMA, so that's good. So like the multi-modal—
Alexandr Wang
The pre-training.
Mark Zuckerberg
Yeah, someone on the team knew. Now... Just kind of out there in the corpus. I think it's a good test of the kind of coding ability that... Because really basically what it's doing, it's like, you know, within the agent, I'm not seeing any code, but it basically is writing this whole video processing, image processing pipeline.
John Collison
Yes.
Mark Zuckerberg
And I'm constantly telling it like, all right... Like, look you took 30 seconds to give me that feedback. I only have a 60-second break. Like, you gotta go faster. Just optimize yourself, optimize yourself. And it's, I think, a good example of how stuff is going to go in the future where basically everyone is actually going to be writing tons of code. They're just not going to know that they're writing code.
John Collison
What kind of MMA steers does a model give you?
Mark Zuckerberg
Um, you want to see?
John Collison
Yeah.
Mark Zuckerberg
Basically, let me see if I can pull some stuff out. But basically it gives me a lot of stuff about balance and, like, hip position. Or it's like when you're kind of in side control here. It's just like you didn't get this underhook in this place so you're going to... You're vulnerable to this getting rolled over here. I'll pull up some stuff. You guys should try this.
Alexandr Wang
It was definitely one of his first use cases actually of even the initial prototypes.
John Collison
Yeah. Yeah.
Mark Zuckerberg
I mean, I did this with OpenClaw before we even had Muse working. It was just like... Then that's why I was saying it's kind of like become an eval for this too.
Alexandr Wang
It's good. I actually didn't know that it was an eval for the product. So noted.
Mark Zuckerberg
Yeah. There you go.
John Collison
How is all this... In the model's training data, like the fine-grained, this is what an MMA fighter looks like when they're appropriately balanced, and this is when they're like too much on the back foot. Just, I'm wondering, how has it consumed so many images of appropriate versus inappropriate MMA posture?
Mark Zuckerberg
Here you can see it. It just pulls out frames from it and it's like too much on my back leg there.
John Collison
Yeah, but is this the emergent intelligence? Like it does actually... Again, does it know from seeing lots of images of MMA or does it know because it understands the concept of where you're meant to be balancing? It's read books about MMA, and then it sees that you're too much on the back foot, or—
Mark Zuckerberg
Well, we did train the models with a lot of visual perception data.
John Collison
Yeah.
Mark Zuckerberg
So, I mean, I think that that's like kind of making good multimodal models. I mean, this gets into like, I think, a core approach thing that we have, which is we're training the models end-to-end to be good personal agents, right? So like the question is why would Meta want to build these models?
Mark Zuckerberg
First of all, I think just superintelligence overall is going to allow us to run the company better and build all kinds of different things. But for the products that we're building, there's certain capabilities and certain mixes of skills that basically are just much more applicable to these cases. So I mean, this is one of the things that Alex focused on from the beginning was like, all right, we really want to make sure that our multimodal program is good because that's disproportionately important for the types of things that people use all these services for.
John Collison
How do you train a good multimodal model?
John Collison
Is this where Meta just has an advantage because you guys have so much image and video data?
Mark Zuckerberg
It's very helpful.
Alexandr Wang
Yeah. And I think ultimately, a lot of this, it comes down to how are you focused on it, right? I think because of the fact that we wanted to build personal superintelligence. And fundamentally, I think most people in the world really experience the world in a highly multimodal way. They like talking to other people and listening to things and seeing videos and filming things of their own. I think most consumers have this highly multimodal experience. And especially earlier this year, there was actually a point of tension almost in the industry because I think so much of the consensus view within the AI industry was that it was just all about coding and coding was all that mattered. And I think that for the success of a product like Muse and the success of a lot of these consumer use cases, it really depends on the ability to be strong and multimodal.
Mark Zuckerberg
Yeah. I think we're just kind of scratching the surface of that. I mean, a lot of the initial theory that we had was that, I mean, most people don't actually like reading that much, right? So even the text-based interaction model, I think it's something that people do. I mean, everyone texts. But when you think about stuff like the glasses that we're working on, or even our other products that are very visual, the glasses basically will eventually just be a pure multimodal product, right? It's like you can let it hear what you are listening to, and it can talk to you, and it can see what you see if you let it. And then eventually, I don't even know if we're going to have an app store on the glasses. It'll just be like a multimodal generation that kind of creates whatever you need right in your vision. And I imagine something like that will probably end up powering a lot of the apps that we build too. But yeah, I mean, we're still in the early days of that. But for now, the perception work has been really impressive.
John Collison
I want to ask you guys about the product space here in AI and try something out on you. If I was to describe to you, say, with Stripe, what I think a winning product looks like, it would take me pages and pages to talk about because we have the developer experience and we have to talk about all the international heterogeneity of payment methods. Then we have to talk about stablecoins. Like, it would take me a long time. Whereas I notice a phenomenon in AI where it feels like the winning products collapse very quickly onto an agreed-upon product space that everyone is pursuing, and so transformers worked, and then like the chat agent UI really worked. And now we are seeing... You mentioned OpenClaw earlier, the kind of OpenClaw persistent agent, has a computer, able to do things. You know, computer use out there in the world view. We're collapsing on that product space. But there's something about AI where the winning products tend to be very simple, pursued by all the companies that are taking it seriously. You could describe on the back of a postcard... Like you could describe very concisely. And obviously, implementing them is massively complicated.
John Collison
Like, we'll get into the new model. There's, you know, huge amounts of frontier work that goes into making a good version. But the winning brand—now you're talking about the glasses or the future versions. But somehow there's a very small number of clearly correct products in AI as opposed to it being this, like, very broad and complicated product space.
Mark Zuckerberg
I kind of disagree.
John Collison
Sure.
Mark Zuckerberg
Well, first of all, I think part of the theory from the beginning was that this kind of agent or whatever form it took was going to be quite a different thing from this text-based assistant, right? Every company has one of those. I mean, ours is Meta AI. It reaches a ton of people, but I mean, it's like the same thing that all these companies have. And we, I think, specifically think that that is used for very informational purposes, but people don't think about it as like a thing that gets stuff done for them.
John Collison
It's retrieval, not delegation or something.
Mark Zuckerberg
Yeah, and even like when you say it has a computer... Yeah, but there's lots of different ways to do that, right? You know, I think that it's only kind of this most recent wave that gives the agent a dedicated VM, which I think is actually incredibly important for making you feel confident that the security and privacy of your information is—it's not just like, okay, it's not connecting to some elastic compute that just everyone uses.
John Collison
But my claim isn't that it's easy to implement, nor that it's obvious—
Mark Zuckerberg
I think there are going to be a lot of things.
John Collison
Yeah. Okay. You think there's going to be a lot of different things? Because again, my claim is that now that people see Muse, everyone looks at that and says, "Oh yeah, that's clearly how you want the AI agent product to work, and that the future AI agent products that we'll see will be fairly Muse-shaped."
Alexandr Wang
So I think... My take on this is, I think we're just so early in the realm of what AI products will look like that I think a lot of the interesting differentiation about the products kind of gets collapsed in how we view the products today. I think the traditional way in which we wrote down and think about product specs, that is... Those aren't the relevant dimensions and levers and axes to consider the important design decisions on AI-native products. And so, Muse, for example, I think so much attention was put in the design of the model for the product. And there were a lot of very intentional decisions around, in all these situations, how exactly should the model behave? And what's the exact personality of the model in these situations? And how should it handle these sorts of tool-use scenarios? And these are very intentional decisions. And probably in the same way that the full spec of how we wanted the model to behave in all these scenarios would be an extremely long document. But because of just how we think about products today, it's just AI.
Alexandr Wang
It's kind of like how when mobile first started happening, it was like, "Oh, it's just an app and everything is just an app." Well, it turns out there's a lot of nuance in what exactly an app is. And we're kind of in this phase where, oh, it's just AI, or it's just an agent, or it's just an agent with a VM or something like that. But the interesting differentiations we have to tease out.
John Collison
Okay. So you guys think I'm making the classic mistake, which is like, my domain is complex and this domain is, again, from a product point-of-view, it's simple.
Mark Zuckerberg
All payment processors seem the same. Yeah, no, I really agree with what you're saying. One thing that we talked about from early on, is that basically there is this pattern where I think the products are getting thinner and the model is getting to be more of the product, right? So one of the things that we say internally is like, the model is the product.
John Collison
Yes.
Mark Zuckerberg
Right? So, and when you're talking about an agent, a lot of the complex product work that went into Muse is basically around the privacy and security model for handling your data and the things that get connected to it. And that, I think, is something that we're just very proud of that we really focused on. But that's a lot of the product surface that we need to innovate on. But the capabilities, the personality, a lot of the stuff is like you're training into the model. And whether they're the people in the lab who are working on that or the people on Nat's team who are working on the product, I think when Nat is thinking about what is the key stuff that we need to make the product feel good, he's talking about post-training the model, right? Not like... I mean, of course the UI of the product matters too, but the most important part of the UI is the product.
Alexandr Wang
Yeah, exactly. I think there's like, there's still very viscerally, I think from a user, there's a feeling of using Muse over time. And maybe some meaningful portion of that is just the reliability of Muse to handle certain tasks. But then another meaningful portion is how is it proactive? And in what context is it proactive? And when it talks to you, how do you feel about how it talks to you? And so there's clearly a lot of... And I'm starting to see more and more of this actually show up on the X feed, which are people are starting to tease out these, like... They're all proactive agents, but what are the actual meaningful differences between them? And these, I think, are the interesting product decisions, which, you know, the funny thing is how do you specify these product decisions? Well, it's ultimately in model design decisions, which ultimately boil down to training methods or data or evals or, you know, all these things.
John Collison
Yes. As we talk about product space, one notable Muse detail is that there is no model selector. and no effort selector. Do you guys think that's just obviously correct in a consumer AI product, or—
Mark Zuckerberg
I just want to make it simple.
Alexandr Wang
Yeah. I think that we've been... I think somewhat infamously, a lot of the AI products have been designed by a highly technical group that do not have these sort of randomly sampled American in their mental model. And so the products... I mean, the naming of some of the models is like so egregious.
John Collison
The numbers go backwards?
Alexandr Wang
Some of the times the numbers go backwards and sometimes it's just like a totally different word and you have another set of words. And like, I just like, how is somebody who's not carefully reading API design specs supposed to know what they're supposed to pick? Like, it's just far too confusing.
John Collison
Okay, we're talking about models a lot. Tell me about the new model. It sounds awesome.
Mark Zuckerberg
Yeah, I mean, I think we're quite proud of this. Um, yeah, the new model—Muse Wonder—it's a new class of model that is at the frontier. So we've basically been working on this since the beginning. Basically, when you're doing these pre-trains, you don't just go to whatever the biggest cluster is that you have. You want to work your way up the ladder training models that are maybe 4 or 5 times bigger, and then you just keep on repeating that process. So the first set of models that we've had Muse Spark was all based on this initial pre-training that we did in, what was it, December?
Alexandr Wang
Yeah, the end of last year.
Mark Zuckerberg
Yeah. And, and it's actually... I think we've all just been quite surprised and impressed with how much intelligence and capability the team has been able to build into a model which is like frankly not that big, right? I mean, and then after that we were working on this next model, and the pre-training has been done for a bit, and we've been working on the post-training, and now it is here. So it's the next class of model. It is significantly bigger. So I think you you will feel that in terms of the nuance in which it can think. And it really. A lot of the areas that we focused on were making it the best personal and professional agentic model. Some of the different things that you can optimize for that's it for us, and then yeah, a lot of the basic capabilities that we focused on building in. Very good memory, very good ability to persist through complicated tasks. But a ton of work went into this across the lab and the product teams and all of this.
Mark Zuckerberg
And I think it's also just sort of a testament to the the team being able to methodically figure out the process to scale these models. So I think this is also just, you know, a milestone on the way towards the time between each subsequently bigger class of models, I think will now start to compress. And that's very exciting for the type of stuff that we can build.
John Collison
At Meta or in the industry generally?
Mark Zuckerberg
At Meta. I mean, I think you're seeing something like this everywhere else, but—
John Collison
Is this an RSI effect or something else?
Mark Zuckerberg
Well, I think at Meta, part of this is just... We needed to... Well, I think quite famously, we were not happy with the trajectory that we were on with Llama 4. We kind of like rebuilt the whole team. That's been this kind of crazy thing over the last year and a half that we should probably spend a bunch of time on.
John Collison
Oh, we'll get to that. Yeah, yeah.
Mark Zuckerberg
Some war stories. But the team basically had to go build up like different tiers of models to get to this one. And now that we are kind of at the frontier, now everything that we do going forward will just be pushing the frontier, as opposed to there's a lot of background work that we needed to do before that. I don't know, what would you say?
Alexandr Wang
Yeah. Well, I think we're really excited about it. We will have 2 models in this series that are going to be available. So there's Muse Wonder and Muse Wonder Full. Laugh? Yeah, all right, there you go.
John Collison
Yeah, very good.
Alexandr Wang
Uh, which is basically a much higher—
John Collison
Studio band version.
Alexandr Wang
Yeah, no, a much higher-end model.
Mark Zuckerberg
It's the full version of Muse 1.
John Collison
I know, I got it.
Alexandr Wang
Yeah, yeah, it's not—
Mark Zuckerberg
It's 2 words. Wonder full version.
John Collison
You're clearly a father.
Mark Zuckerberg
What's up?
John Collison
This is an excellent dad joke.
Alexandr Wang
Yeah, there you go. Um, and, and the performance... I mean, we've seen like incredible performance across a slate of evaluations on personal, professional agent tasks. It's state-of-the-art based on our measurement. And then there's a number of, I think, specific highlights which we're really excited about. Which we've designed into the model over time. So active memory, like Mark mentioned, is one area where we think that to build personal agents, personal professional agents, like memory is just one of the clear requirements to build reliable agents and is one of the areas where we've invested, I think, disproportionately actually compared to many of the other efforts. And we have our own internal evals of Memory Bench which demonstrate a lot of the capabilities there.
John Collison
What does memory mean in a model to a layperson? Is it just how much stuff it can... Like the working memory of if you're working on a task, how much stuff the model can kind of keep in its—
Alexandr Wang
Yeah.
John Collison
How's it different from context?
Alexandr Wang
I think it's in the experience of using the model. How well does it remember things that you mentioned to it over days, weeks, or months, and then leverage those in the process of doing future tasks?
John Collison
I see. This is the stuff that's buggy AF in OpenClaw. Like, I think we're all extremely fond of OpenClaw. Very happy OpenClaw user, but it ultimately does this compaction where it has like the long-term memory and the working short-term memory or whatever. And I think once people use OpenClaw for a while, they find the memory pretty buggy.
Alexandr Wang
Yeah. So we've designed basically both the model and a system for the model to use to develop a much stronger form of memory. We call it active memory.
John Collison
Yeah.
Alexandr Wang
In addition, as Mark mentioned on the persistence—
John Collison
But so this is outside the model, like the model exists and it's—
Mark Zuckerberg
No, no, but some of the capabilities that you're talking about, like the compaction, are things that the model does, right? So basically being able to do this well, both with—
John Collison
It's not a harness, it's in the model itself.
Alexandr Wang
Yes, it's in the model itself. Exactly, yeah.
Alexandr Wang
So it's aware of its memory as a core system. And it's about how it's able to leverage various sources to be able to conduct tasks in the future, as well as understand its ability to do compaction, how it should do that compaction most effectively, et cetera.
John Collison
Okay. But this is a big deal because I think a lot of people's early experience of AI agents has been buggy memory. And so, yeah, improving that is—
Alexandr Wang
And to be clear, there's still plenty of room to improve. Our evaluations show this, but this is something that we're quite proud of, and we think this is a meaningful contribution.
John Collison
Okay. So active memory is one thing.
Mark Zuckerberg
Yeah.
Alexandr Wang
Active memory. Active focus is another, which is effectively just the ability for the model to, when conducting tasks, be able to carry them through, especially as we've seen Muse users as well as Muse Code users, give higher and higher complexity and more and more ambiguous and more and more complex tasks, we want the model to be able to conduct those and do those very effectively. And this is one of the things actually in early dogfooding and testing that a lot of people have noticed about the model that is sort of special and stands out with respect to other models that they've tried or used. The third one is domain expertise.
John Collison
What is focus in AI terms? Is it like putting a lot of compute on the task? Is it not getting distracted with other—
Mark Zuckerberg
Kind of, persistence?
Alexandr Wang
Yeah, like persistence. It's basically like the ability to conduct complete tasks over very long horizons.
John Collison
Okay.
Alexandr Wang
And then when faced with various setbacks along the way, the ability to identify other solutions to those problems in an aligned way.
John Collison
Okay.
Alexandr Wang
Then there's... It's actually a big step up in coding capabilities, so frontier coding capabilities. So it is actually state-of-the-art on DeepSWE and, SWE-Marathon and does very well on other benchmarks like Terminal-Bench 4 and many others. So we think that users who use it for coding, which will be a minority of all of our users, but the ones who do will see a big improvement and a big step up.
John Collison
Okay, so we should also just be using Muse Wonder Full for coding at Stripe?
Alexandr Wang
Potentially, yeah. I mean, let's go out there and—
John Collison
But like if it's state of the art on SWE-bench, yeah. We'd be missing out.
Alexandr Wang
Yeah, no, it's a big improvement on coding. I mean, obviously, we want people to try it and see how they experience it, but we've been very pleased in our own internal testing.
John Collison
Okay, so this is—
Mark Zuckerberg
Yeah, I think people will like it for a lot of things.
John Collison
Okay, so this is both just your latest frontier model, like people can come, you know, developers can come directly to you, they can use it on OpenRouter, like they can just consume it as a model, and it's also a big upgrade inside of Muse that people will just feel as active users of Muse.
Mark Zuckerberg
Yeah. Well, what we're going to do in Muse, because it's a lot bigger and it's more expensive to serve, we're going to start off primarily in paid subscriptions. So what we want to do with Muse as a product, we want it to be free for everyone to use. So we have this huge amount of free tokens that people can use with with Muse Spark as the model. But if you want to use Wonder for most of the hard tasks, then at least for now, you'll use the paid version. But, you know, I mean, eventually we'll, you know how quickly the field moves.
John Collison
Yes.
Mark Zuckerberg
And we hope to, you know, at some point not too far from now, have a model that is as good as Wonder, but in the Spark form factor. And that will be a great thing to be able to serve hundreds of millions of people around the world.
John Collison
Okay, so you go to Muse, you subscribe to the—what's it called—like the premium or pro plan, or... Pay money.
Alexandr Wang
Yeah.
John Collison
And just don't stop until you've taken out your credit card.
Alexandr Wang
Yeah.
John Collison
And then you'll be able to... You'll select it, or it'll just magically—
Mark Zuckerberg
Oh, it just gets better.
John Collison
It just gets better for you.
Mark Zuckerberg
Yeah.
Alexandr Wang
And then it'll also be available via API, on the Meta Model API and OpenRouter. And then it'll also be available for Muse Code users. And then Muse Wonder Full will be available to subscribers of Muse Code. So it's pretty exciting. I think we want to, as Mark mentioned, it is a much larger model. And so we're working through how to serve that most effectively to as many people as possible. But we're excited to get it in the hands of as many people as possible.
Mark Zuckerberg
It's a big step forward. I mean, this is like, and we really with Muse Spark, we were very happy with the performance that we were able to deliver as we were getting up to this point. But like, this is, the beginning of, like, MSL kind of being at the frontier. We feel like the models are quite competitive with the best models that are out there. And I think that this is—
John Collison
Okay, so you think it's like Astra, Fable, and Wonder altogether at the frontier? I mean, you can look at it empirically by the benchmarks.
Alexandr Wang
Empirically by the benchmarks, one would conclude that.
Mark Zuckerberg
Yes.
John Collison
I see.
Mark Zuckerberg
Yes.
Alexandr Wang
And I think actually one of the things that we're especially proud of is we have taken the time, as we've mentioned, to optimize models for the product form factor of Muse. So this is the state-of-the-art model. We think it really is the best model in the world for the use case of personal professional agents.
John Collison
Yeah, people inside companies like complaining about Goodhart's Law. You know, a metric that becomes a target ceases to become a good metric. You hear complaints that models are benchmaxed and that at a certain point, kind of overfitting to the benchmarks actually makes the models slightly worse. As people who actually have to train AI models, is that true? Are the benchmarks pretty good at grading models? Do you start to see weird edge failure cases from the—
Mark Zuckerberg
The benchmarks have to keep moving, right? Because it happens very quickly, but they just get saturated so quickly. So, I mean, I do think that—
John Collison
But you think non-saturated benchmarks are fine?
Mark Zuckerberg
I mean, as we've gone on, we steer the models more towards our internal benchmarks for the use cases that we're focused on. So, I mean, early on when we were kind of just doing the science of trying to climb the ladder, we cared more about what the external benchmarks were just to make sure we were on the right track. And then I'd say, you know, each release that goes by, the larger share of what we're looking at are internal ones.
Alexandr Wang
And one of the things that we actually quantified in our research blog post as part of Muse Wonder is this metric for big model smell, which is kind of, you know, big model smell and feeling of benchmaxing are kind of like inverses of one another.
John Collison
This idea that you could just like tell when you're using a really big model, it just... You can tell.
Mark Zuckerberg
It just has nuance.
John Collison
It's got that je ne sais quoi.
Alexandr Wang
Yeah. And so we, the way we quantify that is actually almost explicitly what you're mentioning, which is the gap between in-distribution performance and out-of-distribution performance. So what is the... We kind of measure how well does the model perform on tasks that are in distribution of our RL and post-training versus tasks that are out of distribution. And what does that gap look like for... What does that gap look like between the Muse Spark line and the Muse Wonder line? And we actually see this huge step up from the Muse Spark to the Muse Wonder line because of, as Mark mentioned, like, it's just a much better base model powering this new line. And there's an entire je ne sais quoi to that, but I think we've been able to actually quantify this effect. And it's an obviously dimension that we're excited to continue improving on.
John Collison
But if you're saying that external benchmarks have ceased to become as important, and you care much more about your internal benchmarks. Is that kind of an argument against benchmaxing, that you think that just trying to get the top score on the external benchmarks that everyone talks about is maybe not going to lead you to develop a great model?
Mark Zuckerberg
Well, yeah, but also we're focused on specific things. I mean, part of the whole strategy is personal superintelligence and just trying to build different types of products than other people are trying to build. So even if we did have the exact same evals as everyone else. It just—
John Collison
Yes. And though I would say a personal superintelligence is not like a super niche interest, like it's probably something that other models want to be able to do.
Mark Zuckerberg
I think that there are things that are very different about it. I mean, I think like the personality probably matters more. I mean, one of the things that we talk about is this idea of discretion, right? So if you want to use the model to get something done, it needs to be able to know a lot about you and get things done in the world while disclosing as little as possible. So, you know, if you have an allergy or you're pregnant or something, you're trying to get a table at a restaurant and you don't want to have to tell them your medical condition.
John Collison
Yes, yeah.
Mark Zuckerberg
You know, to make sure that it's going to be a good place for you. Like, okay, how do you handle that? And I mean, humans have common sense and the models so far... If you're building a coding agent that works in an environment where everyone has access to the same context, discretion isn't particularly important. So there are all these things that I think are just kind of more important for us.
John Collison
Yes.
Mark Zuckerberg
Aspects of coding where the code needs to work, but you don't... The consumers who are using it may not even ever see the code. I mean, that probably matters more in a personal superintelligence context rather than serving software developers. So I think these things are just a bit different.
Alexandr Wang
Yeah. Ultimately, we care about making useful models. So we want models when you put them into people's hands, they have a great experience using. And I think that is the most important metric for any model that we or other people produce. I think that really focusing on the external benchmarks, if you really dig into the supply chain a little bit, it's some PhD student who made a benchmark as part of their... As the course of them doing their PhD. And so you're kind of outsourcing all of your product thinking to that PhD student who happened to make a viral benchmark.
John Collison
Right.
Alexandr Wang
So clearly that's wrong.
John Collison
Yeah.
Alexandr Wang
And so that's why we care a lot... A huge part of what a lot of people at Meta and MSL are focused on is crafting amazing evals.
Mark Zuckerberg
And now that we have them, we can work on our own. But if you're building a... You're ramping up your lab, right? It's like those public ones are not a crazy place to start, but it's just not where you want to end up.
Alexandr Wang
At that point, you're just looking for signs of life. But then when you're like really in this mode, as we are now, where it's like we want to be very intentional about how we're improving our model over time to build better and better products. A lot of the cases that Mark just mentioned, these are actually detailed very rigorously in the spreadsheet of all the little capabilities and nuances of the product that we think really matter to the end user. And maybe each of them matters like 2% to the user, but in aggregate, that's the obsession that will ultimately result in these models that feel magical.
John Collison
Can you guys describe for me just what goes into training an RL model? Or sorry, a leading-edge frontier model? I'm just curious in that I think a lot of people listening to this have a rough idea where there's a pre-training and a post-training step. So pre-training is you ingest this huge amount of data. You look at a lot of pictures and these are cats and these are dogs. And after a while you start to develop a sense for what is catness and what is dogness and what is fur and everything like that. And so that's the pre-training step where you just hoover up data and maybe you can talk about the data acquisition and just what goes into that. And then there's this RL stage, which is the newish advance over the past few years of AI. And again, it sounds like that is where a lot of the specific nuance of this model has come in, where you guys have chosen better domains, better hills to climb for the model. But just at a high level, describe how many people, how long it takes, what the steps are, how you think about those in—
Alexandr Wang
How much time do you have?
John Collison
At a high level.
Alexandr Wang
This happens to be what we spend all of our time on.
John Collison
Exactly. And so summarizing—
Mark Zuckerberg
YouTube courses.
Alexandr Wang
Yeah, I think at a very high level, these are the stages. It's pre-training and post-training. Ironically, training is the superset of all of it. But that's—
John Collison
How many people are on this whole effort, like when you develop these models?
Mark Zuckerberg
So this, I think you're getting into what is interesting about MSL and the approach that we've taken, which is, from the very beginning, I think when I first reached out to you, I think we were just looking at this because we were just like, "Wow, okay, how far we've come." And I was just like, I pulled up the screenshot of like when I first reached out to Alex. My first thing was like, "Hey, do you have time to catch up and talk?" And then it's like, all right, we talked for—
John Collison
That's how they get you.
Mark Zuckerberg
44 minutes or whatever. And then after that call, 2 days later, Alex sends me this long thing, which is like, "Well, here's what I think I would do if I were in your position." And the very first thing is that you don't need a lot of people, and talent density is the thing that matters. So I'd say from the beginning of the plan, this was the core of your strategy, right? It's basically you actually don't need a massive number of people, but what you do need is for each of the people to be as good as possible, which informed this kind of crazy recruiting push and very intense filtering that we had to take some of the best people from inside Meta and a very intense process to get people from outside Meta. But I actually think like a lot of what has made MSL special and why a lot of people have wanted to join is because I think it's in some ways it's just like more fun to work on a talent-dense, smaller effort. It feels like the startup that's operating. In some ways it has all the benefits of a startup with the stability and resources of a larger company. And this was this environment that I think was point one in your plan for what Alex was like, this is what we need to go do. So I don't know the answer of like how big is a pre-training team... It doesn't need to be that big. But I think if you look at some of the other labs, it's probably quite big in some other places. So, I think part of the reason why these models are doing well is because you have people who we're betting on and they have a lot of resources because we're taking our compute. We're not peanut-buttering it across a lot of competing efforts. We're basically just investing in the key people, a lot of compute per researcher, and then they can do their work.
John Collison
Was Wonder the one big training effort? That was where most of the training was?
Mark Zuckerberg
It was the next pre-train. It was climbing up this ladder. So, we have this tiering system and then Spark. I mean, internally, like, Avocado was the code name for the Muse Spark line, and then Watermelon was the code name for this one. And we're out of fruits now. We're out of fruits. Yeah, so we're going to have to switch to a new convention.
Alexandr Wang
But, um, I think the only fruit that is technically larger is the durian, and you don't want to go to the durian.
John Collison
No, no, no, no. Yeah, yeah.
Alexandr Wang
No, you want to stay far away.
John Collison
A durian is bigger than a watermelon?
Alexandr Wang
I think it's definitely heavier.
John Collison
That's horrifying.
Mark Zuckerberg
Definitely heavier.
John Collison
Spikier.
Mark Zuckerberg
Spikier, yeah.
Alexandr Wang
But no, yeah, I think as Mark mentioned, each of the... I mean, each of the stages I'll describe, it's tens of people working on it at Meta versus hundreds at—
John Collison
So Meta is a huge company, but this is like a really tiny effort inside of Meta.
Mark Zuckerberg
Yeah. Oh yeah. The research team is less than 200 people.
Alexandr Wang
Yeah, it's very small. It's intentionally very, very tight. I think that the—
Mark Zuckerberg
It's closer to 100, right? And then there's a bunch of info folks. Yeah.
Alexandr Wang
It's a lot of infra work. So I think at a very high level, to go back to the 'what is it' question, you know, how does it happen? How does model happen?
John Collison
That's basically what I'm asking.
Alexandr Wang
Where do models come from? You take a huge amount of data for pre-training, and here you're—
John Collison
And it's here, like, roughly... It's like all the public internet and all the books you can get your hands on and just like as much data as possible.
Mark Zuckerberg
But different types.
John Collison
Plus the proprietary Meta data.
Mark Zuckerberg
But also different modalities, right? So if you want it to be good at perception, then you want to make sure it has a lot of images and audio and video and this kind of stuff. If you want it to be good at certain domains, you don't just want to RL that in at the end. You want to have it have exposure to that from the beginning.
John Collison
Yep.
Alexandr Wang
Yeah. And in this stage, you're much less... You care a lot less exactly what the data looks like or how it's formatted or all of everything about it. But it's just lots and lots of data.
John Collison
When you talk about this being the best personal agent model, was there data you had to get for the pre-train where it's like huge amounts of restaurant reservation, call booking... I don't know what it is, but like—
Mark Zuckerberg
You just kind of want to dump in large volumes of the modalities that people are going to care about. I mean, there were a bunch of things. I don't actually think we talked about this when we were talking about where Muse Wonder excels. But some of the areas that it's like, "All right, what do people care about," right? There are things like shopping that people care about a lot, personal advice that people care about a lot. People have personal legal questions, personal health questions, personal tax questions. Muse Wonder, I mean, is like quite good at those. I don't know if there's any quantification that we want to give, but it's, let's say it's—
John Collison
It's very good on those personal advice questions.
Mark Zuckerberg
Yeah, and so those types of things, I mean, you can RL them to some degree, but you just want to put in a large amount of data. You can imagine health is a very... There's a lot of data out there about health. And in pre-training, you just want it to see all these different things. And so you kind of just put these different corpora in.
Alexandr Wang
Yeah. So as Mark was mentioning, like, one of the highlights of Muse Wonder actually is its ability on various domain expertises.
John Collison
Yes.
Alexandr Wang
And I think our philosophy here is that the best personal agent should feel like a team of experts in your corner. Like, it should feel like it is your legal team, and your finance team, and your medical team, and, and, and, and, and. And so on a bunch of these domains, Muse Wonder is actually state-of-the-art on a lot of the finance evals, as well as a lot of the legal evals. It is the state-of-the-art model. And these domain areas, that starts in pre-training. You know, we look for a lot of domain data to put into pre-training.
John Collison
Yes.
Alexandr Wang
But a lot of the, a lot of the art of pre-training... Pre-training is... There's this one Roon tweet where it's like pre-training is full of the researchers who like chew glass and stare out into the distance and into the abyss or whatnot. But I think there's a truth to that because it is both the highest stakes of the research areas because you... We invested a huge percentage of our compute into training Muse Wonder, into pre-training Muse Wonder. And it was very high stakes. I think probably both of us at various points in that process, you know—
Mark Zuckerberg
You went through troughs of sadness.
Alexandr Wang
Yeah, troughs of uncertainties. All the 7 stages of grief happen while you're pre-training a model. And it is the most fundamentally scientific in an interesting way, because you just don't have very many samples. So, your ability to infer exactly what will happen scientifically really depends on your ability to sort of big-brain reason about everything that's happening in this process.
John Collison
Okay. So in the pre-training stage, everyone who's training frontier models basically is trying to get as much data as possible for the pre-train with maybe a bent towards if you want Muse Wonder to excel at legal and finance and tax questions and things like this, then you want to have as much of that specific corpus as possible in the pre-train.
Alexandr Wang
Yeah, you still want as much intentionality about where ultimately you want your model to be, but again, it just benefits from a lot more data.
Mark Zuckerberg
Yeah.
Alexandr Wang
And this is sort of the massive-data, massive-compute stage.
John Collison
Do you hit some saturation in the data you put in the pre-train, or ultimately is it just more data is better?
Mark Zuckerberg
Well, there's a relationship between if you want the model to be bigger, you need more data. Otherwise, it overfits.
John Collison
Yes, but my point is is there data where it's like, "We can't distract the model with that, we can't clutter up its brain attic with knowing that stuff," or is it just more data is better?
Alexandr Wang
There are some of these effects, I think.
Mark Zuckerberg
Yeah, because this is part of the science.
John Collison
Okay, we're getting into the secret sauce, sorry.
Mark Zuckerberg
No, it's not even that, it's just like... I think that there are all these questions about, like, "Well, we put a lot of these tokens in, is there any zero-sumness in a bunch of places there isn't, or it's actually positive transfer, but then like depending on how it's set up, like you can have some negative transfer.
John Collison
Yes.
Mark Zuckerberg
And so I don't know, there are all these interesting questions. But in general, one of the bigger challenges of scaling bigger models is it's not just about how many flops and how much compute. You want to be able to scale data with that. I mean, this is like part of when the original scaling laws for pre-training were written, I mean, this is sort of what came out of that is is that. So if the internet isn't growing that quickly and you want to scale bigger models that can be more intelligent, I think one of the interesting problems is, all right, where's the data coming from?
John Collison
Yes.
Mark Zuckerberg
So I think that probably gets to the end of what we're going to talk about on that. But we could go to the other—
John Collison
Excellent CEO energy over here.
Alexandr Wang
So post-training. Then I think post-training is kind of a very exciting field now, I think, and is an area where I think almost on a continual basis, the entire field is learning about the optimal recipes to be able to post-train these models.
John Collison
And again, for people, like, post-training is where you give the model a task that you expect it to do, grade it on how it does. And so if it's a chess-playing model, it plays chess games, learns that—
Mark Zuckerberg
Well, there's actually 2 stages. There's the SFT, so there's supervised learning. That's a lot of when you take this kind of raw thing and you give it both the question and the answer that you would want it to have, right? If like, if you're having a chat conversation, how would you want it to respond in this case? So when you're training things like personality, the SFT is actually like a very important part for molding it. But then I think most of the advances in the recent years have been in RL. Yeah.
Alexandr Wang
Yeah. So I think one mental model that can be helpful is an SFT is where you kind of initialize a lot of—
John Collison
SFT is supervised fine-tuning?
Alexandr Wang
Supervised fine-tuning, yeah. Where you initialize a lot of the behaviors that you want the model to be able to exhibit later on. And then maybe at the end of SFT, it has those behaviors kind of baked in, but it'll only do it like 10% of the time. And then in reinforcement learning, you really sort of give the model reinforcement in various situations to bring the elicitation of those behaviors dramatically up. And I think that the... Yeah, so I think there's a bunch of interesting things that are happening in RL and post-training. I think one is that just the scaling of RL and post-training has been one of the clear axes, just spending more compute and gathering more data.
John Collison
Just doing much more of it before you release a model.
Alexandr Wang
Yes, doing much more of it, especially as the pre-trains scale up, doing much more.
John Collison
It's like anything in scaling laws, like it has to go with the rest of the factors or it doesn't work.
Alexandr Wang
Yeah, in general. I mean, if I were to say like one broad statement about AI, it's just scale everything, you know?
John Collison
But in particular, you have to scale everything. Like scaling one thing in isolation does not work.
Mark Zuckerberg
You technically can.
Alexandr Wang
You can scale one dimension. You just will get much more optimal performance if you scale many things altogether.
Mark Zuckerberg
Well, so one way to think about Muse Wonder, right? It's like, all right, so we did the pre-train.
John Collison
Yep.
Mark Zuckerberg
We've done some RL. When you look at it like Muse Spark, right? So version 1 to 1.1, 1.2, 1.3. We didn't change anything about pre-training. We just continued doing more RL on top of it. So when you think about the early versions of a model of a new class, they're actually relatively less, like the RL to pre-training ratio is going to be lower for RL than the later versions.
John Collison
That is kind of what the point releases in models are.
Mark Zuckerberg
Well, I think every company has its own convention. For us, that's what it is meant to be.
John Collison
But for a lot of companies, like the big number is the pre-train and then the—
Alexandr Wang
I think it kind of lost that semantic value. Yeah, I think now you actually can't do that.
John Collison
Okay, but that is true at Meta at least.
Mark Zuckerberg
But we haven't even done it. We kind of decided to not even do it.
John Collison
It's true at Muse Spark at least.
Alexandr Wang
It is true at Muse Spark and Muse Wonder. It's true that Muse Spark and Muse Wonder are different pre-trains.
Mark Zuckerberg
And we decided to kick the can down the road on this question for Muse Wonder by not giving it a number yet. But when we have our first dot release, we will have to decide that.
Alexandr Wang
But yeah, no, I think the RL stage, part of what's very interesting is first is just producing the environments and producing the tasks. And I think that is one of the areas where there's a lot of intentionality in training these models. It's also one of the areas where you notice very funny things. Like if you... In a lot of cases, less is actually more when it comes to data in RL. And so I think I think the feel and the science of it is pretty flipped, generally speaking, from pre-training.
Alexandr Wang
The other is that I think it can give very magnified instruction to the model, if that makes sense. So I think that a lot of safety cases or alignment cases that we care a lot about are where you would see when you do large high-compute RL runs on frontier models, are where you apply a lot of RL or optimization pressure in scenarios where the only way for the model to succeed is to do something misaligned or do something unsafe. And so that's another thing that we had to be very careful about and thoughtful about in training Muse Wonder is ensuring that we curated the RL environments very thoughtfully to ensure that all of them were eliciting and driving towards more aligned behavior from the model.
Mark Zuckerberg
So the point that you're saying about quality, I think, is an important one because I think a lot of the previous questions back and forth were about the quantity and have the flavor of like, "Oh, you need to scale the quantity." But I mean, I do think like a huge part of this is the point that you just made of like even a smaller amount of high quality.
Alexandr Wang
Yeah, I think, I think in RL, as we've seen pretty big jumps in our new Spark line, I think these have been like accompanied with real meaningful improvements to the underlying recipe. And each of these improvements in the recipe have been like... I think the improvement between Muse Spark 1 to Muse Spark 1.3 feels night and day. I think probably to most people, if you ask them, do these feel like they came from different pretrains? I think they probably would say yes, because I think you can—
Alexandr Wang
I think it's happening industry-wide for what it's worth. It's just like we are learning to squeeze so much capability and elicit so much agentic capability from these existing base models. Now, that's not to say, like, we believe scaling is very important. That's why we have the Muse Wonder line in the first place. But this sort of like... These two things are happening at the same time, which is also, by the way, for us, this is why we're seeing so much progress in general. I think it's the fact that we're scaling both pre-training and scaling our recipes in RL kind of concurrently. At a nice clip.
Mark Zuckerberg
And then there's these other scaling axes too. I mean, like making sure that your agent harness is good and that you can co-optimize these things is probably just another dimension for scaling.
John Collison
But if I'm understanding you guys correctly, are you saying that alignment is not about coming up with some alignment-specific RL steps and instead about making sure the entire RL process is kind of steering the model—
Alexandr Wang
You have to be thoughtful about it in the entire process of RL. Yeah. So I think this is a holistic effort, I think.
Mark Zuckerberg
But also, I think different labs have very different theories. I mean, alignment is a very philosophical concept, of like, who do you want it to be aligned to? I mean, a lot of our theory is we care more about it being aligned to, you know, we're trying to give people agents, right? And it needs to... If you ask it to do something, you need to have a high confidence that it's going to do what you ask it to do. You know, like, you want it to do that with high reliability. You don't want a moralizing lecture for why you're wrong for asking it to do the thing. Obviously, if you're trying to do something that's catastrophic, it needs to gracefully, without pissing you off, not do it. But in general, we view alignment as more of a question of instruction and values alignment with the person who it's intending to serve. And I think that that's kind of a different mode of thinking about it. It's very different from, I think, how some of the others do.
Alexandr Wang
But one of the more interesting things I think that is maybe a question related to your original question is also just how long the process of training one of these models takes. Like Muse Wonder... In general, we've planned Muse Wonder since last August.
Mark Zuckerberg
Yeah, I remember the offsite when they first presented this.
Alexandr Wang
Yeah, so I think all of the stages that we just talked about are like, you know, you really work on them for quite a long time. And, and, um, it's not that we, we didn't start working on Muse Wonder when we finished Muse Spark. It's that we've actually just been planning and working on this for quite a long time. And we really view Muse Spark, if anything, as like I think we mentioned this in the first blog post where we wrote about it. That was a stepping stone to get to Muse Wonder. And that really was how we thought about it.
John Collison
Even at the time you thought about it.
Mark Zuckerberg
Yeah, and in terms of how long it takes, I mean, we're also even just talking about the research part. I mean, there's also the compute part where we've been investing in building up, first of all, a huge amount of compute overall. But a lot of, you know, part of our philosophy on this is that you can kind of measure the compute strength of a lab by how big its largest single computer is, supercomputer, for being able to do these large training runs. So, I mean, we built this Prometheus cluster, we call it, in Ohio.
Alexandr Wang
Mm-hmm.
Mark Zuckerberg
That's scaling to larger than a gigawatt. And now we're building this Hyperion cluster in Louisiana that's going to scale to 5 gigawatts. And those actually, predate MSL. Because like building a whole like data center site from scratch, at least the plans for those, they're coming online now. But that was—
John Collison
You guys thought it might be the feed, right? That you just have used for a lot of compute generally?
Mark Zuckerberg
No, no, no. This is kind of predating, even going back to the, to kind of the LLaMA program.
John Collison
I see.
Mark Zuckerberg
I mean, it was probably in 2023. or yeah, either 2023, 2024 was when we basically started making the plans to build those clusters. They're now kind of fully coming online. Well, I guess the Louisiana one will come online over a couple years.
Alexandr Wang
Yeah.
Mark Zuckerberg
But if you want to go really end-to-end on undoing this stuff, then, I mean, obviously the kind of full stack, and I mean, Meta is a full-stack tech company. I mean, we build our own data centers, we build our own chips, we build our own low-level infrastructure. And now this is kind of the next component, which, I mean, arguably is probably the most important component in all of this for the next wave of technology is kind of building really good models and then delivering them to people. But I mean, yeah, if you want to go fully end-to-end, I mean, this is a long process.
Alexandr Wang
I do think this is like a— I mean, maybe it's well appreciated by folks in the tech industry, but It really is, you know, it feels like models are coming out constantly. So maybe you'd have the mental model that like, oh, a company like launches a model and then they figure out how they're going to make the next model and, or something like that. But it's actually these like very long planning processes and these sort of like very long work streams that kind of all come together at the very end and then result in these beautiful models.
Mark Zuckerberg
Yeah.
John Collison
How do you build good products when the feedback cycle is so long? Like ultimately you start planning the model in August and you don't know until more than a year later. if it's good. And, you know, Silicon Valley famously likes to run on metrics, but, like, ultimately, do you just need someone who is the aesthete and says, I'm sorry, I don't like it?
Mark Zuckerberg
It's nice when you have metrics, but I think the, like, state of being an entrepreneur is, like, how well do you make decisions in an information-poor environment, right? So—
John Collison
But this is, like, that on steroids, right?
Alexandr Wang
That on mega steroids. Yeah, I think that this is— and I think Marc gets a huge amount of credit for this, like, Because of how long everything takes, like, there were many points along over the past year since I've been here, but also, you know, as you mentioned, like, long before that, where you have to make very large investments or make very big decisions about, you know, how you're going to invest or develop the technology or, you know, what decisions you're going to make long before there's proof points around whether or not even the thing is working. And so I think this is like, in some sense, it's like entrepreneurship on hard mode. It's like the opposite of Lean Startup. Like, Lean Startup is like, yeah, you like see if you get something working and then you double down and then you see if that works, then you double down. Like, you cannot do that in AI. You have to like weave— I mean, we've been investing into these data centers since, you know, for years and years and years. We've had to—
John Collison
And you'll be radically behind if you wait to see if things are working. Yeah.
Mark Zuckerberg
So we have these conversations where it's like, At the end of last year, we're like, all right, we are quite excited about the progress. Yeah, we haven't seen the models yet. Like, it's like, I'm going to, like, you know, we're going to write this check for, like, I don't know, like $100 billion, $200 billion. I'm like, Alex, are you sure this is going to work?
Alexandr Wang
It's like, um, yeah, so it was, it was so extreme that, you know, we had, we had like the whole exec team.
Mark Zuckerberg
I think Susan scared you a bit.
Alexandr Wang
Yeah, well, yeah, no, well, it was like The joke every time we talked about it would just be like, yeah, and we hope it's all going to work and be good. But like, there were literally, like, at the end of last year, other than me giving a vibe report on how our research was going and, you know, I could sort of show some plots and whatnot, but it was like, I mean, I was feeling good about it, but it was certainly not clear at all that it was going to work.
Mark Zuckerberg
Yeah.
John Collison
You guys are saying that the essence of training a frontier model is you have to push your chips all in before you've been dealt the cards. You just have to decide that you're going to develop a frontier model and you don't get to meter the inputs.
Mark Zuckerberg
At this scale. I mean, if we, like, back a few years ago, the clusters didn't have to be this big, but now they do. So I think this is part of the excitement of booting up a frontier lab now, right? Is basically like the ante is extremely high.
Alexandr Wang
And I think this is— and like, I kind of like viscerally remember this. We were in Hawaii at the end of last year. It was kind of like the whole exec team got together. And yeah, Susan pulled me aside many times and was like, you know, I just want to tell you, if this doesn't work, like, you know, this is going to be really, really bad. It's going to be very, very bad in financial terms.
John Collison
I really need to impress on you that this model has to work.
Alexandr Wang
it has to be good. If it's good, it's all good. We're great. But if it's bad, it's really, really, really bad. And Susan, who you know—
John Collison
And that was useful motivation?
Mark Zuckerberg
It was.
Alexandr Wang
Well, it was like, it was one of these very funny moments where I was like, you know, obviously the amounts of money are like so obscenely large.
John Collison
What are the amounts of money in training? Order of magnitude, whatever you can share, but just like how much are they costing?
Mark Zuckerberg
Well, you can do the classic math. Yeah, no, I mean, a gigawatt cluster. I mean, it's Like, I don't know, I think what you'd see out there would be, you know, to operate a gigawatt a year, call it on the order of $15 billion.
John Collison
Okay, so it's roughly a $15 billion effort to train a bunch of models.
Mark Zuckerberg
To operate a gigawatt cluster for a year. If you're talking about building the CapEx, it's probably closer to, you know, I mean, you pay for it upfront, so it's probably, you know, $50 or $60 billion. So, but then that's just kind of, one cluster for one part of training.
John Collison
Yeah.
Mark Zuckerberg
Um, then you have all the other parts. Um, research isn't just doing the run, right? It's like, that's like probably an equal amount of it is because for each change that you make, you're testing all the ladders all the way up to the big, the big scale. Um, and then they're serving the stuff. So, I mean, part of the stuff that in some ways—
John Collison
So tens of billions of dollars to train a Frontier model now.
Alexandr Wang
Yeah, definitely the whole, I mean, I've seen the financial rollups, tens of billions of dollars is a good estimate for—
John Collison
Yeah.
Mark Zuckerberg
And then, but then, but then you want to do something with it, right? So I actually think like some of the more complicated questions that we had were like, all right, so we're training this model and we know, we know we want to train the model cuz we, this is like key, but like how much product demand do we think we're going to have?
John Collison
Right.
Mark Zuckerberg
Because so now it's like, all right, well the product doesn't exist.
John Collison
Um, like, and as in you're trying to build data centers to serve Muse for a product that doesn't exist yet.
Alexandr Wang
Yeah.
Mark Zuckerberg
And we have to make it years in advance, right? So this is probably where, like, when we were talking about it last year, we were just like, huh, there was a moment where I think a bunch of people were like, yeah, maybe, maybe we should just like be a little more conservative. And it's like, no, we're doing this. It's like, it's like, we're going to make it work. We're going to make it work.
John Collison
We're going to burn the boats.
Mark Zuckerberg
Yeah, this is going to work.
Alexandr Wang
And But it was like, I mean, even, I mean, I too am an entrepreneur and founder and have, you know, at various points in my life taken risk. I remember the visceral feeling at the time was like, this is a fuck ton of risk. Like, I don't know if I really—
Mark Zuckerberg
Oh, it's only riskier. Yeah, yeah, it's only—
Alexandr Wang
well, I do think this is like one of the like things that I think, Mark, is very impressive about you, which is I think you have really internalized this concept of You know, it's about how big it is when you win, not like, you know, failure is okay. It's just like how big is the win when you do win? And this was, I think—
John Collison
Mark has fully internalized power laws in a way that many people have not.
Alexandr Wang
I think quite deeply, yeah.
Mark Zuckerberg
Well, I think the other piece is, I don't actually think it's quite as risky as this. I mean, the numbers are big, but I think—
John Collison
Meta has a backstop in terms of how it can use compute.
Mark Zuckerberg
Yeah, so I think there's 2 things. One is if AI works overall, plenty of other people want all our compute.
John Collison
Right?
Mark Zuckerberg
So they're willing to pay, like, I mean, if like the cost is whatever, $15 billion a gigawatt, we could get like a lot more for that. So I think that you, in order for it not to work out, it's not just that we have to fail, AI overall has to fail.
John Collison
Yes.
Mark Zuckerberg
And then even if that happens, we still have all these other uses to productively use compute. So in the grand scheme of things, we were definitely building out this compute for inference ahead of knowing that the product was going to be ready.
John Collison
But you were AI-pilled enough that you said GPUs online and powered will be useful to someone.
Mark Zuckerberg
Yeah, but also like we believed in what we were building. I mean, it's like, I think, and you can look at the kind of the trajectory where it's like, all right, first there's ChatGPT as the text-based assistant. Then, you know, Anthropic got coding agents to work. And no one had built the kind of personal consumer version of that. And it just seemed like this is the thing that Meta should build in the world. And I think that that was just like pretty clear to you and me and Nat, like from the beginning, right? That that's like, that's kind of what we're here to do. And so I don't know. I mean, there's always like the real question for me on the Muse launch even was you never know when you launch something how it's going to be received. Right? So there's always the case where you get a home run, which in this case it was, right?
Alexandr Wang
Yes.
Mark Zuckerberg
And it's just like extremely well received right off the bat. But the normal case would be you kind of— it's fine, and you have to like iterate a few times to get product-market fit before you really grow. In some ways, like, I actually think the product, we maybe could have even grown it faster, but we didn't have all the growth stuff really ready to go when we shipped it because we are like— our median expectation was like, all right, we're probably going to have to like iterate on this a bit, and like, that'll be That'll be fine. But I don't know, I think like if it takes a few iterations, it's like not the end of the world. That's normal.
John Collison
Yes.
Mark Zuckerberg
Um, you can have launches where it's like you get like, it's kind of catastrophic and then it takes a while to recover, but that's not good. But, but it's like, but I mean, even if it took a few iterations, it would have been fine. So I don't know, I guess it just seemed to me like this was going to be— yeah, the numbers were big, but I— it seemed like it was probably—
John Collison
Was it bonding for the executive team to be Holding hands, taking this risk together. You're talking about Susan taking UXi?
Alexandr Wang
Bonding or bonding moment?
John Collison
Is that a bonding moment?
Mark Zuckerberg
It's like, I mean, it's stressful.
Alexandr Wang
I think it was—
Mark Zuckerberg
Both.
John Collison
Both. Yeah. Being in a stressful situation together is generally bonding.
Alexandr Wang
Well, one thing that I think is quite notable about Meta is the degree to which, like, first of all, most of the exec team has been there quite a long time. Like, Nat and I were really newbies. And And, you know, most people have been through crazy situations many times in the past as well.
John Collison
Mm-hmm.
Alexandr Wang
And I think Meta as a company is like an extremely resilient company that has been through like, you know, many situations that also at the time would've seemed incredibly daunting. And so everyone's kind of like knows the program, so to speak, in terms of, you know, this obviously on paper, and especially when like Susan Li, our CFO, is sort of like pulling me aside in the meetings, like, does, you know, does seem like a particularly scary situation. But I think most people were like, yeah, yeah, no, but we're Meta, we're going to get through it. Like, I think there was like a real camaraderie around that and sort of like quite a nice energy around that, which really I did appreciate because I think it would be very easy for, you know, it's the other executives at the company who are operating and running—
John Collison
Who are paying the bills.
Alexandr Wang
Yeah, one of the greatest businesses of all time.
John Collison
Yes.
Alexandr Wang
And it'd be very easy for them to be quite annoyed by this whole endeavor. But I think everyone is also very excited about—
Mark Zuckerberg
They are.
Alexandr Wang
What the opportunities for the company with AI are. And I think, like, you know, taking a step back, I think one of Mark's jobs as, like, the founder and CEO of the company is to ensure that the company is always pushing towards the most ambitious and expansive version of what it can be. And I think that one of the great things about this entire MSL and AI push is that I think this is one of the rawest expressions of that is we knew that AI certainly is a very important technology and it's one of the bigger technologies right now. But if we were able to build it effectively and properly integrate it into everything that we do, as well as like, you know, as a full-stack technology company, like build it into, you know, these incredible products that we know will exist in the future, this could be like immensely expansive for Meta. And I think this is true. Like, I think Muse is an incredibly exciting product. It has like, you know, it can go in 1,000 different directions. I think it's one of like probably the most exciting products to just like You know, when like late at night and you just think about all the ways it can go.
Alexandr Wang
Like, I think Nat, Mark, and I just like, you know, shoot across messages late at night of just like, oh, we could do this and we could do this and, you know, we can make this move and we can like build this product. And I think it's just like, you know, very—
John Collison
It's a very exciting time in technology right now.
Alexandr Wang
Yeah, it's very blue ocean in that way.
John Collison
Yeah.
Alexandr Wang
Yeah.
John Collison
I'm struck by, as we talk about Muse and just how it clearly feels like the next platform, an agent operating on your behalf. You've talked for a long time about not wanting to be reliant on other platforms and how kind of the experience of building on Android and iOS kind of burned you on being dependent on other. And often when you would talk about, say, Reality Labs, is the importance of owning. the platform of the future and kind of many years working on Reality Labs with the intent of owning the platform of the future. And now this new modality comes along that feels pretty clearly like how you do a lot of productive work and is the platform of the future. And now Meta is really well set up.
Mark Zuckerberg
Yeah, no, I mean, it's less about wanting to own a platform for its own sake. It's about being able to build the best things. And at some level, you know, if you look at like other social apps, for example, I think part of the reason why Meta is so much better is because we don't just build the app, right? A lot of people like think about it, we're like this app company. We're not an app company, right? It's like we're like a deep technology company and we do this because we are world-class at machine learning and like all the recommendation systems are operating at a scale that basically few others are in the world. I think it's basically like Just us and ByteDance on that, and and then like okay we just kind of go all the way to the metal. And to me, being able to also deliver go the last mile through the device, it does come from a while of having these experiences where we couldn't build the experiences that we wanted on phones. But I just think for the future, the the models are going to be by far the most important part of delivering the technology experience.
Mark Zuckerberg
And I mean, this is something that I think we kind of bonded on and agreed on, like, very early, is this notion that, like, in every generation of technology companies, there's, like, no, there's usually not one company that kind of is the— is, like, wins everything. I don't think that that's how this is going to go either. But with the internet, there were, like, okay, 5 or 6 very big companies and then a whole ecosystem of other companies that are great. The companies that did the best kind of had the core technology to be able to serve people the best there. And I think in this next era, there are going to probably be a small handful of companies that have the ability to build frontier superintelligence. And I think that those companies are going to be the ones that effectively drive the agenda and can build most of the interesting, the most interesting experiences for consumers, for enterprises, hardware, these different kinds of things. So, and I just look at it as, okay, like, is it even really a choice, right? It's like, yeah, okay, Meta could, will probably exist into the future regardless of whether we do this.
Mark Zuckerberg
But if we want to be kind of an interesting company that is like inventing new things that matter to people in the future, I don't see how you can do that if you don't have the ability to build end-to-end your own superintelligence and build them into the products that people use. And I think that this is something that you just like believed from very early on, Is basically, it's like, all right, there are going to be like a handful of these. Like, it's like there's a kind of personal and consumer-oriented version of this. And like, this lane is open because everyone else seems to just be obsessed so much with just automating knowledge work, which, I mean, isn't even that charismatic version of the future, right? Because it's like, all right, like, a big part of our belief too is that the kind of highest use of superintelligence is invention and creation, not automation. And so, I don't know, it just seemed like there's like a big place in there, but you kind of need to own the stack.
John Collison
How is the Muse-mediated future going to shake out with businesses need to— okay, businesses variously need to build support for for Muse, where it might try to computer use its way in, but that might be janky in certain ways. So you kind of want everyone to have first-class support for agents talking to things.
Mark Zuckerberg
Yeah.
John Collison
Then the bigger question is they might actively try to block agents from interacting with them. And this might be because they have a misguided sense where, like, it's actually in their interest to support agents, but they don't want to. It might actually not be in their interest to support agents. You can imagine where maybe they have a very thin moat and it all comes from a business that only succeeds by consumers spending time on the site or only succeeds by overcharging consumers, then they should rationally not want to allow agents because the agents are like, why?
Mark Zuckerberg
But those are going to get disrupted.
John Collison
Yeah. So talk to me just through the second-order effects of how the economy works.
Mark Zuckerberg
Well, I think that it starts off with— I mean, we trained into the models very good browser use, right? So that way we want to make it so that no one No one has to do anything to integrate with Muse. And then, but then of course, for— we're kind of just going down the list of the most important partners, and for people who want to work with us, we'll just get direct API access, right? And then it's like, okay, then it ends up being way more efficient and faster for people to use Muse. It's more efficient because now you don't have an agent crawling your website. It's just going. So anyone who wants to do that, we eventually want to get there. Um, and then I think the fullness of time is that, you know, just like today every business has, um, a website and social media accounts and email addresses and all these things, every business is going to have agents too.
John Collison
Mm-hmm.
Mark Zuckerberg
So, and we're also building that on the other side of it. And, um, and so of course Muse agents will talk to other types of agents, but they'll work really well with our agents, right?
John Collison
Because I mean, we'll, we'll be able to integrate that Just, I mean, so Muse using websites is just a moment in time thing.
Mark Zuckerberg
Just to make it easy for everyone. You guys want to have it as a backstop so it's reliable for people using it. Yes, and easy for businesses to get bootstrapped. But we have, we have way more businesses that are calling us asking to integrate deeply, and very few really who have tried to block it. And my guess is that the ones that try to block it, it will end up not being the right strategic thing for them. But we'll see. This will kind of play out over time.
Alexandr Wang
It saves them a lot of time.
Mark Zuckerberg
Yeah.
Alexandr Wang
It saves them a lot of money. Like, I think these are, these are, like, clearly good things for the consumer.
John Collison
Yeah.
Alexandr Wang
So in the long arc, you'd clearly expect that demand will shift from current avenues more and more towards agentic demand. And, you know, there always will be demand that lives elsewhere, just in the same way that, like, there's still brick-and-mortar demand in commerce. But there's a lot of ecommerce demand. And kind of in the same way, like, I think agentic commerce demand will grow as a portion of the pie. The others will still exist, but will grow as a portion of the pie. And I think there's sort of these questions for businesses in the long run, like, do they want to participate in that slice or that portion of the economy over time? And I think it's an incredibly exciting area. I think it's, like, very, very exciting to imagine what will happen in this agentic commerce future.
Mark Zuckerberg
Yes.
Alexandr Wang
I actually think Patrick wrote a great tweeticle. Is that what we call it these days?
John Collison
I don't know. Patrick has his own thing going on with tweets. It's like they're buried amongst—
Alexandr Wang
I noticed that Brother Karamazov was quite a good literary analysis. But yeah, no, I think clearly this is a discontinuous event for how commerce will happen, broadly speaking, or just generally speaking, how the entire how all the business pipes will work, maybe even in a broader sense. But I think that, you know, you just go back to, this is clearly a lot better for the consumers. It's a lot better for the customers. So there's a natural gravity towards that reality.
Alexandr Wang
One of the exciting things, I'm curious what you guys see too, but I think Muse is one of the first products where you're actually people are fully delegating purchases to the agents. And this is like a— this is kind of like a, you know, a step change, right? Like, I think, you know, we've been talking a lot about agentic shopping and like the possibility of it, but now it's like really happening. And a lot of people are trusting the agents to shop for them. And I think the agents are finding them, you know, in my case, like great teas, or, you know, in whomever— whoever's case, like great products for them to purchase, which is quite exciting. And so I think that there's There's clearly an exciting thing happening here. I think, I think part of the value prop— well, one, one part of the value prop for the user is just reduced friction. Like, hey, I don't feel like click on all these websites and do all this research to try to find the exact products. Like, that step is taken care of for me. And also, like, criteria can just be automatically applied.
Alexandr Wang
Like, hey, I need something by Wednesday. So, you know, the agent will only look for things that could get to me by Wednesday.
John Collison
But that's, I mean, keyword search. It was insane that we ever relied on it because that's a very reasonable constraint. You just want to shop by.
Alexandr Wang
Yeah, exactly. And so I think the like old way of like searching for stuff just feels so obviously primitive now, now that you can do this. Then there's this sort of like back part of the funnel, so to speak, but just like the ease of friction reduction of just, yeah, filling out the form, just like getting it ordered is quite nice. And I think the long-term thing that we're quite excited about is greater personalization. Ultimately, I think one of the exciting things about Muse is that it has sort of like this holistic context into you as a person and all of the aspects of your life. So it sort of like can holistically understand things about your family or things about your coworkers or things about your friends. And then sort of like help you co-optimize across all of those to get you the exact right gift for your, you know, best friend in the context of your friend group based on what it knows about your friend group. Like, I think there's just like a lot of exciting opportunities that are somewhat new and novel when it comes to like these personalizations, these recommendations.
Alexandr Wang
Because historically, a lot of these recommendation systems are quite siloed and quite isolated.
John Collison
Mm-hmm.
Alexandr Wang
And so they're sort of like, they don't see like a holistic view of you as a person. And Muse has an opportunity to offer much greater levels of personalization.
John Collison
Yeah, I think people are already seeing that, where if they ask, like with Muse, if you ask it to buy something for you, quite frequently it'll add in some bit of context that it pulled in from a different conversation.
Alexandr Wang
You had a good example of this.
Mark Zuckerberg
Yeah. Well, so my 3-year-old daughter loves baking. She loves cooking. And so I'm like, all right, I'll start baking stuff. It's like I'm normally into smoking meats, but I'll do some baking in order to To make, to make Aurelia happy. And so we've gotten into this really cute routine where, um, you know, we, we, I've been playing with Muse voice mode, which isn't fully rolled out yet, but we're, we're playing with it. And, um, it does this great thing. It segments the, the voices. So it knows which one is me and it knows which one is not me based on that. And we just like, you know, throughout the week we're talking about what we'll want to bake that weekend.
John Collison
Mm-hmm.
Mark Zuckerberg
And she'll like give it ideas or like she'll kind of come up with crazy things that are not real recipes, but like bounce it off with Muse. And then Muse will just make sure that we, every Saturday when it's time to bake, I get the ingredients from Instacart and they kind of show up and they're on time, which I mean, the biggest thing is like if I were just starting the activity on Saturday, you'd have to get everything and it would take some time. But now everything's ready whenever we're going to go sit down and do it. This weekend it knew that watermelon was launching, so it threw a watermelon in the bag. It was very, very sweet. It was very heartwarming. It deduced that the recipe did not have enough— there was not enough, I guess, expense in it to get the delivery, so it needed something else. And it was like, you're going to—
John Collison
And it topped it up with a watermelon.
Mark Zuckerberg
It was like, do you want this watermelon? I was like, yes. I'm like, yes, that's very cute.
Alexandr Wang
Which I think you, uh, did you— is that the watermelon you ate?
Mark Zuckerberg
Yeah, that was the one I ate on the livestream.
John Collison
Okay, but the watermelon on the livestream was—
Mark Zuckerberg
That was just like, Muse just got it for me.
Alexandr Wang
That was a Muse—
John Collison
yeah, those are just—
Mark Zuckerberg
You just got it for me. But I think there's this, um—
Alexandr Wang
That was a very memeable moment of that very memeable situation. So we have Muse to thank for it.
Mark Zuckerberg
So, but here's another interesting thing about the whole thing is I didn't really use Instacart that much before. You know, it's like every once in a while, maybe twice a year, I would like like, use Priscilla's account or something.
John Collison
Yes.
Mark Zuckerberg
I went and got an Instacart account to give Muse, and now I use it like once or twice a week because like I'm talking with Aurelia with Muse, and we're coming up with these recipes, and it's delivering them, and now it's like a core part of my workflow. So I actually think that the opportunity here isn't just, okay, it's going to be 20% more efficient on the conversion rate. Like, I am a new Instacart like subscriber, and I just wouldn't have done this before.
Alexandr Wang
Yes.
Mark Zuckerberg
So, I do think that there's this question, and then there's this question, which I think is these sort of interesting— like, this is maybe the more interesting economic question, is that Instacart account really for me, or is it for the Muse? Because I think, like, over time, I think that the question is, is the total addressable market for all of commerce just increasing dramatically?
John Collison
Yes.
Mark Zuckerberg
Because it's not just the people anymore. Like, now agents are effectively going to have accounts.
John Collison
Yes.
Mark Zuckerberg
And I think that's where this starts to, like, become a really very exciting opportunity where anyone who sits this out, I think, is going to end up feeling sad over time.
John Collison
And why did you have to sign up for an Instacart account? Like, we're solving this at Stripe Projects for developer stuff where your coding agent—
Mark Zuckerberg
The Stripe integration, by the way, is great. You guys did a great job with us on Muse. I think we're very happy.
John Collison
We have been very excited about agent-to-commerce for a long time, obviously, because again, felt to us like clearly how people are going to buy things in the future. And that was both apparent because no one needs all that friction in their life from the actual buying stage, and how early product research with AI had product-market fit, even from kind of the earliest AIs. It really feels like computer use was the last big unlock that we needed to make it happen, because now you can ask your muse to buy anything, more or less. And it'll just figure it out with computer use, right? And so it's like the ultimate backwards compatibility layer.
Alexandr Wang
Yeah, it's pretty—
Mark Zuckerberg
this is a set of things that Nat really obsessed over.
Alexandr Wang
Yeah, yeah. I think, I think there was a lot of, like, I think a lot of magic of Muse is the fact that like a lot of this stuff just works. And so like clearly if everything was just like a little bit less reliable, it would feel like a total pain in the ass and nobody would use it and people just throw away the app. And so I think the fact that so much of this has gotten to the point of reliability where You know, it just works is like a huge part of the magic.
Mark Zuckerberg
Yeah.
John Collison
What were you saying Matt was obsessed with?
Mark Zuckerberg
Oh, well, I mean, well, Matt's kind of obsessed with everything.
John Collison
Yeah, well, that's generally a sexy character.
Mark Zuckerberg
Yeah, no, it's great, it's great. But this is, I mean, but yeah, I mean, this getting this experience to be good, I think is something that he and a core group of people on that team really focused on.
Alexandr Wang
It's kind of funny. I think I had a dinner with Patrick, like, I want to say it was 2023, where I was, he was like, what do you think we should be doing now? I was Oh, you're like, you're already way behind the 8-ball on this agentic commerce stuff. Like, you got to figure this stuff out. Like, agents are going to be buying things and like, you're not prepared for it.
John Collison
So you were saying this to Patrick or Patrick saying this to you?
Alexandr Wang
I was saying this to Patrick. Well, because Patrick was very politely asking, what do you think we should be doing? And I sort of had this like very like urgent spiel. Yeah, and I think it's very, uh, it's funny how that was like—
John Collison
Well, thank you for your provocation.
Alexandr Wang
Well, it was directionally correct, but, uh, but like 3 years early. 2023, I think.
John Collison
And so, um, no, it's very interesting to look at the agentic commerce stats at Stripe because we started developing all this stuff and we had a whole bunch of announcements at Sessions back in May, uh, and then it only as of September basically started going vertical. Meta is building increasingly capable models, including Muse Wonder, and their models are available on OpenRouter, Alongside hundreds of others. Stripe has agreed to acquire OpenRouter, who's the leading model gateway. OpenRouter already routes more than 10 trillion tokens a day and handles much of the complexity of working with different models through a single API, making it easier to experiment, find competitive prices, and automatically route around provider outages. Try it for yourself at openrouter.ai.
Mark Zuckerberg
Hey, could you order a watermelon to Stripe's office in South San Francisco? I'm doing a podcast here. This would be funny.
John Collison
Funny podcast idea.
Alexandr Wang
I'll check that watermelon delivery now.
John Collison
Great. We'll see how we get on. Well, okay. This actually raises an interesting question. People have been trying to make The HAL 9000.
Mark Zuckerberg
It's working.
John Collison
I really like this. I think this is an awesome touch where it's like clacking away at its laptop. I really like that in the app. We'll just leave them there and we'll see what we get. Yeah, it's tough to prop up. Yeah, there we go. People have been trying to make the speak to a computer AI vision happen for a really long time. Where Apple launched Siri in 2014, 2015, something like that. And obviously there's been Voice Mode and various AI products for a few years now. Why is it so hard? Like, it's still not the case. If you were driving your car and you wanted to check on the baking or move a meeting or something like that, you actually probably wouldn't do it with the voice. I mean, maybe almost now with the latest things.
Alexandr Wang
Yeah, well, there's a lot.
John Collison
I mean, so— But you wouldn't develop Siri like you're doing.
Mark Zuckerberg
Siri right now is actually a good example. As we're talking over each other. And I mean, I don't know, when you're using different voices—
John Collison
So is it just technical limitations? That's why we haven't done it?
Mark Zuckerberg
Well, I think that there's a few things. And one is you actually want it to go do stuff. So there's a question of when you speak with it, even if it understands, how's that going to work? Then I think that there's a lot of stuff about the way that people have conversations. It isn't typically, I will give a prompt, and now I will wait, and you will speak. And then, so it just goes back and forth. Um, how many of your sessions when you're talking to an agent in voice mode, do you basically, oh, it starts saying something, but then like, I don't know, it's like, I kind of feel like if I pause for 3 seconds, it's going to think I'm done and want to speak. And then if it starts speaking, but then I cut back in, is it going to think that it's like a whole new question? Um, I think there's a lot of the stuff about the conversational dynamics that you want to get right. But then there's this other issue, which is that For voice overall, it's quite high bandwidth for you to communicate with voice, but in terms of consuming information, having someone speak to you is very inefficient compared to having it be visual.
John Collison
Yeah.
Mark Zuckerberg
So I think like that's why, you know, partially when we do these, these interfaces, I think it's, it's kind of working on its task. I fired it off now. So I think the ideal interface is probably not just strictly voice-voice. Although I think voice output from you is pretty useful.
Alexandr Wang
Yeah, I think, I think there very clearly already is an explosion of voice input. Like, I think you even see this in coding where like—
John Collison
Yeah, people are speaking to the—
Mark Zuckerberg
Yeah, yeah, yeah.
Alexandr Wang
Most of the researchers just kind of like whisper into their keyboard because it is like, as Mark's mentioning, it's like the highest bandwidth way to get your like thoughts out, or usually the lowest friction way to get your thoughts out.
John Collison
Yeah, yeah, yeah.
Alexandr Wang
I think the most extreme version of this, by the way, is I think how college students now use AI is maybe the most interesting, where they just— I think a very common modality of how to use AI is that they just will completely word vomit for 10, 15 minutes, being full stream of consciousness, chain of thought of all the things.
John Collison
And then have AI clean it up and infer.
Alexandr Wang
Yeah. And have AI effectively do all of this organization of it and pulling things together and structuring everything. It's quite interesting. It's a very interesting, I think, modality of tech usage.
John Collison
Yeah.
Alexandr Wang
But yeah, I think voice input has been big. I do think agents is like, that's a big step change, because I think the mental model is just very different. Like, if it's just like an informational thing that you're talking to, that like, you will clearly get into situations where it just feels useless. Whereas now that it can do things for you, it's much closer to the experience of talking to a human.
Mark Zuckerberg
Totally.
Alexandr Wang
Or to a person. And so I think there's all these interesting uncanny valleys in interacting with AIs that until they've been able to take actions have felt kind of strange. And now that we have agents, I think one of the cool things about Muse is this confluence of a number of things. I think The fact that Muse is embodied and it like is this little guy that like you feel like is working on your behalf is also only possible and a coherent like representation and embodiment of the product because it has persistent memory and it has the ability to do things for you and the ability to take actions and has like a little bit of a personality. Like, is the confluence of all these things coming together that make this like embodiment concept— Yes. Even work in the first place.
Mark Zuckerberg
Then it turns out, okay, actually this is the real glasses, right? It's like, okay, a lot of people want it even with just like no display or a small display.
John Collison
Yes. The camera and audio is enough of a useful product.
Mark Zuckerberg
Yeah, yeah. And I— and then we started shipping the display glasses that aren't the kind of wide field of view full augmented reality, which, I mean, I thought would be awesome too, but even the small display is pretty powerful. So I think you'll start getting some interesting interactions there that are kind of fully multimodal, right? So it's voice in and out and It'll be able to see what you see, and it'll be able to kind of paint the pixels.
Alexandr Wang
And one of the— I mean, the other exciting thing that we've done here is Muse Gadgets, which is really, I think, unleashing the ecosystem to try cool device formats and form factors that could be really exciting. Like, since releasing that, we've gotten actually even a number of phone manufacturers, you know, talk to us about excited You know, we're like, we're just going to integrate Muse Gadgets into our phone and see how well it works. And like, we're really excited about that. Like, I think, um, I think part of what is cool about the current moment of technology is that like, it's very clear agents are really exciting. It's really clear that like, this should unlock lots of things.
John Collison
Will you guys come back to the household devices? Like, this is obviously a personal device, but, um, we actually had during COVID the, um, Portress—
Alexandr Wang
the Portal. Portal.
Mark Zuckerberg
There we go.
John Collison
Yeah, I mean, like, he's finding the Portal inside.
Mark Zuckerberg
One of the gadgets that we just shipped was the little HomeLink. It's like the size of a little, like, nickel or something, but with a USB, right? Yeah. Um, uh, so you can basically connect that, and then your Muse can connect to everything in your home. This is actually one of the most fun things too is just having it connect to your Sonos and connect to all your lights. I mean, I have it basically like when it's time for me to start winding down in the evening, I just have it like dim all the lights across the house and it's nice.
John Collison
How are we going to fund the AI buildout? Everyone's worried about, you know, Google went free cash flow negative. You know, you were talking about 1 gigawatt and then 5 gigawatts and everything like this. Meta's now free cash flow negative yet, but there's always more CapEx to fund. And so what's the answer?
Alexandr Wang
Where is Susan Lee?
John Collison
Exactly.
Mark Zuckerberg
No, I mean, look, I think that there are going to be a lot of ways to create value from this. So what we're trying to do with Muse specifically is make it so that it can be widely used, because I think that there are going to be some kind of valuable network effects at scale as people start being able to interact with each other's, and that's going to be great. So I think there are all these reasons why you want it to get to large scale. Also, you know, we were talking about partners. Partners are going to want to integrate more the more people are using it. So it will get more valuable if we can get it to scale. The best way to do that is to basically have a model where it is affordable and free for people to use. But there are going to be a lot of ways that I think we can scale revenue. Subscriptions is going to be a big one. You know, basically, just because the base version of it is free and we want the free version to be excellent doesn't mean that there are things that just naturally aren't going to be economical to include in that.
Mark Zuckerberg
So, you know, if you want the kind of— you know, we talked about this earlier. If you want to start using Muse Wonder in Muse tomorrow, then, you know, the way that you'll do that is you'll pay for a subscription. So I think that that will be a big business. There are a bunch of other things around commerce that we're quite excited about too, where we basically think— we believe that Muse is actually going to make people money and save people money by using it, and that over time there are lots of different ways to effectively take a small cut of what you are saving or making for people, and that will end up being a very large revenue stream. And then there's all the other things, right?
John Collison
So, you think Muse will generally just cost negative money for people? because it helps them out.
Mark Zuckerberg
Well, there are already all these sites that I think are interesting where people are like, go look through my emails, find a bunch of subscriptions I don't actually use, cancel them, whatever. Um, yeah, so yeah, I mean, that, um, I don't know if that's the thing that we would monetize, but, but like, but you get, you get, you kind of get the point and where it helps you make money. I mean, one of the ideas that we kind of talk about, I mean, this is, this is something that I think you've been kind of obsessed with, is making it like a business in a box for small businesses. So someone can just kind of— it can help them administer all these parts of what they're doing. And you can imagine that being very valuable and being very profitable, both for the business owner and for Meta. But I think one of the things that's interesting is, like, OK, if we're building the models to build the personal agents and the business agent, there is this large surplus of value because now the model, it's very general, right? So it doesn't just do those things. We'll also have the API that'll be profitable.
Mark Zuckerberg
A big part of why we want to start the Meta Enterprise Platform and have CJ Desai join us to build this is we have this view, which is like, all right, yeah, sure. So traditionally Meta hasn't been an enterprise company, but we believe that a huge part of the challenge is building the leading models in the first place. And they're very general. And once we do that, if we make a significant investment in building an enterprise group that can provide that in the way that enterprises need to consume it, then I mean, I think a lot of enterprises are going to want to use Muse Wonder. It's going to be an excellent model and affordable. And like, it's so, um, so I think that that is going to end up being a very big business too. And all of these things put together, coupled with the fact that we were already starting off as a very profitable business that can, that has good credit and can kind of underwrite these large investments, I think puts us in a pretty good position as a lab to be able to build out the compute that we need to build leading products, build up the network around this, and position the company to really lead in the area for decades to come.
John Collison
Okay. So, you think ultimately there'll just be a lot of revenue from these products, and so they'll be self-funding in fairly short order?
Mark Zuckerberg
Yeah. I mean, that's a much shorter answer than mine.
John Collison
It's like— But you're saying people are underestimating—
Mark Zuckerberg
The question was, Are we going to be able to make money from this? The answer better be yes.
Alexandr Wang
I think the history— Yes.
Mark Zuckerberg
The answer is yes.
John Collison
But the history is very front-loaded expense, right?
Alexandr Wang
Well, the history of AI as an industry, I think, to some degree is like, you know, people have always sort of been hand-wringing about, you know, how are these businesses going to make money? And I think, you know, every year people are sort of generally surprised at the upside.
John Collison
That superintelligence is Is a useful, incredibly valuable—
Alexandr Wang
Yeah, exactly. This is what—
Mark Zuckerberg
this is kind of Meta's whole playbook, right? Is we build something, we get it to a billion people, then we scale the revenue while growing our revenue primarily in the meantime by continuing to scale the things that are already kind of at scale and growing.
John Collison
Yeah.
Alexandr Wang
I think this is also one of the dynamics, like, I think, you know, I think Dario's talked about this as well, but like, just in general, because it requires so much advanced planning to build these models. Like, if you look at the R&D expense at any given time, that is like over— an overlapping R&D expense of multiple model generations at once.
John Collison
Yes.
Alexandr Wang
And, you know, these, they require a long time to work on. And then ultimately, you know, you see like immense payoff when, when you ultimately deliver them. And this is kind of what we saw with the Muse Spark family of models. Like, you know, again, we started working on these a long time ago. And are seeing like incredible consumer adoption through the Muse product.
Mark Zuckerberg
The other thing is that we use the models to accelerate our core business, right? So I mean, you have this core business that's hundreds of billions of dollars and still growing extremely quickly. And now, like, if you can grow it a few percent faster by using these models to accelerate these things, then that ends up funding a huge amount of this too as you're scaling it out.
Alexandr Wang
The Meta core business is a, is a like 8th wonder of the world.
John Collison
It's extremely good business.
Alexandr Wang
It really is. is quite incredible and is growing at sort of astonishing rates.
John Collison
Yeah, you guys really cooked with that one. Yeah, I think you're also getting at something, Alex, which is like there's an interesting accounting artifact where if you go build a data center, you have CapEx and you get to depreciate that over a useful life of many years. And people debate what that useful life is, but the point is you do get to depreciate it, whereas a lot of the AI costs are also just in researcher salaries, and those salaries are being spent on, you know, model training that also has a useful life of many years, but you do not get to capitalize generally.
Mark Zuckerberg
And that's also a financial advantage for us building out the data centers as opposed to a bunch of the other companies that are paying clouds, whereas they're paying the annual rate, but then at some point the GPUs will be fully depreciated, but we're probably still going to be using them for a while because they I mean, the value of even like an H100 at this point is higher than it was last year, right? Just because if compute continues being constrained. So yeah, I think it's— yeah, I think—
John Collison
Any advice for us with OpenRadar? We have agreed to acquire them. It has not closed yet.
Alexandr Wang
Yeah, I think part of the exciting— part of what we're starting to see in terms of the models, especially the frontier models that are being developed, like Muse Wonder versus, let's say, a Fable or an Astra, is that we actually are starting to focus in quite different directions from the other models. And I think there's a lot of intentionality that goes into the model development process that we've discussed as part of this, you know, in this conversation, that is genuinely different from how other models are focusing. And so, you know, when these models kind of get collapsed to the same set of benchmarks, that elides all of the—
Mark Zuckerberg
This is a good point.
Alexandr Wang
The interesting, um, sort of like, uh, texture that differs between the various models. And so I think one thing that is an exciting opportunity for OpenRadar, generally speaking, is like, you know, um, helping to Helping developers understand what these sort of subtle differences are so that they can make the most informed decisions about what models ultimately they select. And then also you can imagine the flip side of that, which is, you have all these developers using these models. And so those developers are great people to ask about what are the interesting textual differences between these models that are more than just tweet length. Like, this model's good, this model's bad, more than just that level of insight. And I think this is like getting to the wine sommelier version or form of understanding the models and their sort of relative strengths and weaknesses, I think would be quite exciting.
John Collison
Okay, so based on your experience, the OpenRadar view that there will be a diversity of models in the future and they're optimized for different tasks and—
Mark Zuckerberg
I agree with that.
John Collison
Exactly. That's just your guys' experience based on—
Alexandr Wang
Yeah, we are focusing On different things. We care a lot more about personal and professional agents, my observation, than the other major labs, which is good. There should be differentiation in the economy. This is the point of capitalism to some extent.
Mark Zuckerberg
Yeah.
John Collison
So you're going to say something?
Mark Zuckerberg
No, I just really agree with that point. I mean, I think that part of what I love about you guys working on OpenRouter and being able to scale it is just the more infrastructure that can get developed, that supports a diversity of players in the ecosystem, I think is good overall. I also, for the same reason, am quite a fan of some of the recent NVIDIA moves to support the open source ecosystem. I don't know, but this just kind of goes with our whole theory and philosophy about, like, what does a good future look like?
Alexandr Wang
Mm-hmm.
Mark Zuckerberg
We think that a good future is one where you achieve a, like, you basically get prosperity by empowering people. That's kind of the personal superintelligence thing that we're doing, but there's the whole business version.
John Collison
Yeah.
Mark Zuckerberg
But for a lot of the important safety topics, like, our view is that the way to get there is having a diversity and having checks and balances by kind of having a lot of different things out there and accessible to people, not like just having a couple of good models that are only in a couple companies' hands. I think that that would be— from my perspective, I think that is probably a riskier configuration for the future than lots of people having access to different things.
John Collison
Do you guys have a theory as to why that is? Where I feel like maybe people predicted more of a winner-takes-all dynamic in AI, or one person clearly— like, a story that people told back in 2023 is that one AI lab will start to develop an advantage, and that will allow them to get recursive self-improvement going, And then they'll just pull further and further away from everyone else. Whereas it has remained extremely competitive where there's kind of multiple people at the frontier. You guys with Wonder today are coming along with your frontier-class model. Open source models have done an impressive job of keeping up, all told. And so we had this story a few years ago that someone was just going to run away with the prize. Why hasn't that happened?
Alexandr Wang
I think there's a lot of underlying reasons and mechanics that drive this. One certainly is just execution. I think for any given frontier lab, there are periods of strong execution and weak execution. And I think that is true across all the labs. And so that sort of starts to average itself out over time. I think another—
Mark Zuckerberg
Very diplomatic.
John Collison
Sorry.
Alexandr Wang
Yeah.
John Collison
Is it that some of the Players in the lead screwed up?
Alexandr Wang
Well, I think if all players, when they were in the lead, executed perfectly, they would probably be able to stay in the lead. So I think that's one factor. I think the power and strength of distillation is certainly another factor that's certainly benefiting the open source ecosystem quite a bit. And I think this is just like a—
John Collison
There's an anti-hegemonic force in the phenomenon of distilling a model.
Alexandr Wang
Yeah. And this is just a quality of the technology. It's not like anyone planned for distillation to work super duper well, but it just turns out that you can get a lot of capabilities naturally through some form of distillation. And this is, broadly speaking, distillation is quite a good force for the entire ecosystem because it means that you can pack very strong intelligence capabilities into smaller and smaller models, which enables us to serve them at much wider scale. And that's certainly one of the very exciting stories is just like really small models today are like, you know, much more powerful than the biggest models a year ago or 2 years ago. And then I think the last is that there's just sort of this— I think there's like, I actually think a lot of people in the ecosystem want this future where it is not just this like, you know, definitely. One final company in all of capitalism kind of kind of end outcome. Like, I think this is one of the reasons—
John Collison
Not just individuals, but companies putting dollars against things.
Alexandr Wang
Yeah. And I think, like, you know, this is actually one of the reasons why a lot of people were excited to join Meta and Meta Superintelligence Labs in the first place, was that they were just sort of like, you know, I want to work at a place that is focused on giving this technology to billions of people. That is, like, not the stated goal necessarily of all the other frontier labs. And so, I think there's, like, a lot of sort of just, like, natural forces where many people in the industry are sort of, like, grappling with, okay, what is the actual world state that I want to live in in the future? And there's lots of sort of, like, decisions that people make at all levels in terms of where they work or who they do business with or what business deals they do that sort of, like, drive towards a more equalized outcome.
John Collison
You get into the Some of the MSL startup, which I also wanted to ask about, which is it just seems fundamentally very hard still to— again, we talked about what goes into developing a frontier model, but also from a standing start, like when was the standing start for MSL?
Mark Zuckerberg
Well, we launched Llama 4 on April 4th, I think it was, 2025. Right. And then I think, what was it, April 20th or 21st was, I think, when I was like, hey, when I started calling people for advice on what to do.
Alexandr Wang
So, I remember I was in a staff meeting and I got, like, the WhatsApp notification, like, have time for a call? I was like, huh, interesting.
John Collison
Okay, so MSL is a little over 2 years old.
Alexandr Wang
No, no, no, 1 year. 1 year.
John Collison
Is it '25? Oh, it's '26. Oh, sorry, I'm living in the future.
Mark Zuckerberg
Yeah, you are in the future.
John Collison
MSL is a little over 1 year old. Um, yeah, what are your reflections a little over a year in on what was load-bearing, as the IAs would say, in how it came together?
Mark Zuckerberg
Well, I mean, I think that there was this whole thing early on which is just like, first we had to pull together this core group. Right? So, um, so Nat and Alex and I, we were like talking a bit up front, um, and, you know, at first we were, you know, brainstorming. Yeah, like I was talking to Nat about, could we— is there a way that we can like structure this so that way it's like a win for Alex's shareholders and like we can like get him over here? And then, so then we were talking about, uh, like what would the plan be? And, and then basically I was just, I was in this mode where I was just like asking a lot of different people what we should do. And like every time that, that we would talk, I'd come away with like 3 more things of like, oh yeah, this is what we should do. It's like very like high signal, very like a lot of strategic clarity. And so we were brainstorming and working on this for a bit. And then Nat and Daniel joined and we pulled together a few of the the initial researchers.
Mark Zuckerberg
But like, in the beginning, it was sort of like the Supernova thing where we just kind of like jammed a bunch of people together. And I don't know, it was like, I don't think either of us slept that much for the first couple of months.
John Collison
No.
Mark Zuckerberg
I like, don't even, I kind of have this mental block almost where I like, I kind of almost cleared my calendar for like a month or 2 and was just meeting researchers and talking to them. And I remember even before we announced the deal where you were going to come over, we were basically like, you were recruiting people and people were like, what is going on? Like, it's like, what is this?
Alexandr Wang
But, um, but then there was the whole question, which was a really hands-on data labeler.
Mark Zuckerberg
But, but that was, that was this whole question too, where it's like, all right, well, like, okay, so if we're not happy with the trajectory of the current program, Like, who were the right set of people to lead the program going forward? And Alex and I had spent a bunch of time talking over the years. I mean, maybe like, whatever, once every few months or something we'd talk. And I was always just very impressed by how he thought about things strategically. And I kind of had this view of like, all right, yeah, currently he's doing this very operational data business. But it always felt to me like he was the sharpest entrepreneur of his generation, of the ones who I had had a chance to meet. And like, it always just struck me, it's like, all right, like, this is a guy who, yeah, he's like running this like kind of very grindy operational business, but like—
Alexandr Wang
Lovely business still, by the way.
Mark Zuckerberg
What's up? We're big investors. It's not a— it's a— there's more Commentary on the difficulty of running the business.
John Collison
When you run the 8th wonder of the world, Alex, a lot of businesses look like a grindy operational business.
Mark Zuckerberg
It's just like my view of things. It's like grindy, grindy. No, no, no. But no, so it's, so I just kind of had this view. I'm like, all right, like, even though this is sort of I think maybe when we got started, a lot of people were like, okay, like, how is this going to work? It's like the kind of like the social media guy is working with the like data guy.
John Collison
Like, is that like, are we going to build a frontier lab here or what?
Mark Zuckerberg
But, but I mean, but my, but I mean, my like really strong conviction, because I talked to like all the leaders of all the labs, like all the heads of research, like all this stuff. And like, I was like, no, Alex is the guy. Like, he's the fucking guy. And like, and we're going to, and basically, like, we're going to basically be able to, to, to go pull together a team and recruit the people that we need to go do this. And like, we need this kind of very, this kind of like strategic clarity that he has around the different ways that we want to approach this, the kind of extreme focus on talent density, on kind of like the clarity of communicating the different vision, which we kind of fashioned together over this period. And a bunch of things that I think were kind of extreme but have basically worked, like having this, this like very flat culture was something that you were kind of really obsessed with, right? It's, I mean, you have like, was it almost 200 direct reports? That's like, we have, it's not, it doesn't, it's not quite that extreme in practice.
Mark Zuckerberg
I think that there's like, like team leads and all, and all this stuff.
Alexandr Wang
My muse helps me manage that.
Mark Zuckerberg
Yeah, well, that's— But like, but there were these series of things that I think quite early we, you kind of articulated, and I was like, okay, this is like kind of extreme, but it seems like directionally right, and like we should go do this. And, you know, I went and like talked to other people who were more like the people you'd conventionally think of, and it's like, it was just very much like, I don't know what you'd expect expect a big company to operate. And I was like, no, we want to do something that's—
John Collison
As you describe this, I'll note 4 things that I think even other opinionated, effective founder CEOs would not have done. One, you just called the recent prior effort by a bunch of people at the company a failure and not good enough. I mean, you were more diplomatic than that, but the act of standing up off the trajectory— Yeah, but the act of standing up So, you know, people can read between the lines.
Mark Zuckerberg
So sometimes, yeah, sometimes you have to reboot a team.
John Collison
That's one. Thing 2 is you are willing to pay ungodly amounts of money for effectively talent. Like, you're paying a lot of money to bring Alex and some other people, but, you know, a lot of money for Alex in particular over. Thing 3—
Mark Zuckerberg
And 49% of this very good grindy business. Yeah, exactly.
John Collison
But let's keep the story straight. So there's that. Thing 3 is, I'll note, Alex had never trained a frontier model before. And so you were taking some amount of a bet on Alex's potential to do something where he had done stuff in AI, but no offense, he just had not done the specific thing that you were being hired to do. And then 4th, you were saying you cleared your calendar and just only did this for 2 months. I think a lot of CEOs would've had the business review and would have had the board thing and they would have had the customer stuff and things like that. Those 4 seem to me pretty interesting where I think a lot of CEOs would not have done those.
Mark Zuckerberg
Yeah, that's interesting.
Alexandr Wang
I definitely think— I mean, one thing you may not appreciate this about yourself, but I think one thing that is quite special about Marc is the degree to which, like, once he concludes that something is the right thing or something makes sense, he just like fully shifts everything to be consistent with that belief set. And it is quite like, to your point, I think like the inertia of most, even very successful founders who like are opinionated and edgy or whatnot, would just be to like, oh, I still have like my whole calendar. And like, so I'll just like talk about doing this thing and hope that somebody does it kind of situation. Whereas I think, No, I think when Marc concludes that something is the thing he has to do or makes a lot of sense, then he really does, like, he has, like, maybe very high learning rate in that sense. Like, he's very willing to, like, pivot entire things to be consistent with it. And I think that, like, you know, one very underappreciated thing is the degree to which so much of the company had to pivot to support MSL as well.
Alexandr Wang
Like, I think that while it was a kind of like this more startup-y effort within a larger company, you know, the needs of MSL were still quite large. Like, we needed a huge amount of help on infrastructure. We needed these large-scale data centers to be built. We needed like lots of support from many other parts of the company. And especially as Muse has been successful, we've also needed lots of—
Mark Zuckerberg
Yeah.
Alexandr Wang
Deep collaboration with the rest of the company. And this is one thing that I think kind of only is possible because Mark is the founder CEO, and he can ensure that we have kind of like a global optimization towards this sort of like AI maximalist version of Meta versus the natural inertia of an organization towards just local optimization ad infinitum.
Mark Zuckerberg
But it's also a lot of how we work together is it's important to me to be in the details of this work so that we can get everything around the company to go do what we need to do. Right. And so it's like, I mean, yeah, I'll have opinions about this stuff, but I'm not like, I'm not the lead researcher.
Alexandr Wang
Right.
Mark Zuckerberg
It's like, it's, you know, it's, but I think that's kind of a lot of, I think how we work together well is that, yeah, it's, I mean, we're, we're kind of like, quite in the details of all of this stuff, but running the lab is a big thing. Running the rest of the company in order to support this is like the effort is a big thing. You almost like, you can't really get this done, I think, without both of those pieces.
Alexandr Wang
And one thing that I think was quite important about when we set the whole thing up is Meta, the company, invested a huge amount to stand up MSL, and then we very clearly told everyone like, hey, there's no deadlines. We are going to take things at the right pace, which I think on the one hand could be like, you know, absolutely terrifying. It's like you've just like stood up this extremely expensive unit and you're telling everyone like, hey—
John Collison
Yeah, and you're worried inside a larger company that like, okay, 7 years from now maybe something will ship, right?
Alexandr Wang
Yeah, totally. And so, but I think that like, first, I think one of the cool things about, you know, focusing on talent density and hiring all these very brilliant people who have done so much already, that's very impressive, is, you know, everybody had like a lot of motivation to ship and everybody wanted to sort of like leave their mark, so to speak, on the field. And so everybody was like heavily, highly self-motivated to move as quickly as possible. And one of the things that's been funny and I've had to continue reminding the team of is like, you know, it is a team of a lot of insecure overachievers. So like at any one point, if I ask them like, hey, how good do you think— where do you think we'll be in 3 months? I always get the most like—
John Collison
Everyone always says it's awful. It's awful.
Alexandr Wang
We're screwed.
Mark Zuckerberg
Yeah.
Alexandr Wang
These like very pessimistic answers. And then I record it and I like 3 months later I'm like, hey, you know, we just shipped this model. Let me just show you what all of you said, where all of you said we would be in 3 3 months ago. And, uh, and this is—
John Collison
And has that caused them to get any more optimistic or no?
Alexandr Wang
No, they're still—
Mark Zuckerberg
they're very self-critical. I think it's, it's not, it's not that they're pessimistic about— I think that the average person seems quite optimistic about MSL and very self-critical about their own project. Yeah. Which I think is actually quite a healthy, a healthy dynamic.
Alexandr Wang
Um, but, but the trend continues. Like, I think, I think, and, and this is, I mean, in many ways Muse Wonder is like this incredible milestone for everyone because it is a—
Mark Zuckerberg
The team was very, very critical.
Alexandr Wang
Yeah.
Mark Zuckerberg
That was working on it along the way.
John Collison
When did you guys know that Muse Wanderer was good?
Alexandr Wang
Um, that's a good question. I think that we, uh, so there's all these intermediate drawings.
John Collison
You're peeking in the oven and it's like, I don't know, it seems kind of gooey still.
Alexandr Wang
And yeah, yeah, going back to it, I mean, there are all these intermediate intermediate signals that you can get along the way that things are good or bad.
John Collison
And what are the intermediate signals?
Alexandr Wang
So in pre-training, you have a lot of perplexity-based evaluation. So you have a different regime of evals that can help you in pre-training. And then you have the more traditional benchmarks that can help you towards RL. But no, I mean, this model, kind of as you mentioned, has been cooking for quite a while. And You know, a lot of— I think one of the funniest things about AI research is that, you know, there's a lot of times where people just sort of like are kind of squinting at a bunch of dots and being like, yeah, looks like it's good, or, uh, looks a little bit off. Like, one of the funniest things is the people have protractors, like analog protractors that they put against you know, a set of lines on their monitor to like measure an angle. And then they, um, so he's not come for this job yet. So I think there's, I think there's like a huge amount of like, uh, like it's very, um, there's more vibes, um, driving like, okay.
John Collison
But when did we have this conversation where, when did you know it was good?
Mark Zuckerberg
Um, well, we had this debate after it was done pre-training where it was okay. How long will it take in RL before it is better than Muse Spark? And I remember the folks on the pre-training team were just trying to set expectations. They're like, look, sometimes this takes a while. And then it was immediately. And so we were like, okay, I think we're on a good path. That to me was sort of—
Alexandr Wang
Yeah.
Mark Zuckerberg
I think when when people got— yeah.
Alexandr Wang
We've been excited. I think we've been pretty excited about it for a few months now. Yeah.
Mark Zuckerberg
But there's a lot to do there. But going back to the beginning, I think one of the things that was very interesting and challenging in standing the thing up was like, you really just wanted this like very high talent density effort. And we were just kind of pulling some of the best people from Meta in and recruiting a lot of the best people. And I mean, you were asking, it's like, all right, well, like, okay, for people who, like, if you were asking before, it's like, well, Alex hadn't necessarily trained models like this. But to me, one of the big things was like all of the best people were very excited to work with him. And I think like, it's, I mean, you, I'm sure see this in all of, in a lot of the things that you do, which is like, you know, it's like you, most people are not kind of one-dimensional, right? Yeah, maybe you do a specific thing, but your ability to grow and expand is kind of largely based on how effectively can you just do people who are excellent in different domains want to come work with you on different things.
John Collison
Yeah, and the great people have followership.
Alexandr Wang
Yeah.
Mark Zuckerberg
And so I think that there was a little bit of that, but also just It's even for the people that don't know you yet, or like, or aren't, you know, don't know what you're about. Like, are you like an impressive person and a compelling person that people want to come work with? Do people feel comfortable that you're going to create this environment? And I think you were like very thoughtful about this early on. We were kind of— Some of the stuff I think you kind of had fully formed as beliefs of how you thought we should do this. But I also remember when we were having the initial recruiting conversations, we were sort of in real time shaping— it's like, okay, well, this construct that we're talking about with this person isn't quite going to work, but like, they seem good, and like, they have this thought. And it was just like, like, really kind of coalescing this vision about what the shape of it should be in real time. I mean, I don't know, I mean, I'm sure you remember that. this, where it was like there were some of the initial concepts of like, I mean, I think probably the most extreme thing was I think you had the concept of the talent density and all and that kind of piece.
Mark Zuckerberg
But I think we weren't sure upfront that it was going to have to be a full reboot, right? It's like that basically started when we were out talking to a lot of the kind of best people. And we realized that, hey, like, in order to make this as attractive as possible, it should feel like, okay, we're starting this thing again from scratch. And yeah, we'll pull in a lot of the great people from the Llama effort and FAIR and all of that too. But that was a very intentional thing that we basically figured out and that you figured out as we were out talking to the initial set of people of like what was going to make this compelling. And it ended up being really smart and like making a huge difference in the culture of the lab and how it got and how it got started. But I don't know, it's like being able to be that dynamic about figuring out, like, what does the shape of this need to be? It's not like that. I think that's a very uncommon trait. It's a very entrepreneurial trait, right? So, when I was thinking about, like, all right, like, this field is going to keep on evolving, right?
Mark Zuckerberg
So, what is the thing that's going to be important for someone who's going to run a lab well? It's not like, are they an expert in whatever the latest modality is of, like, or kind of of training technique for a lab. It's, is the person a clear thinker and able to kind of dynamically update and move in, like, as kind of an entrepreneur would? I think this is just like an inherently very entrepreneurial space.
John Collison
Has this experience changed your view on the value of going external, where both your reboot, a bunch of the people you brought in were external, like Alex and Nat and Daniel and folks. But then also you're describing deciding whether you needed a reboot based on going to the outside intelligence of the aggregate views of the researchers in the industry and getting a sense for the— But just Do you feel like you'll carry forward that approach for future unrelated projects? Because obviously the risk of a large company is that everyone's like, oh, we got a great approach. Yeah, let me show you our 100-page document on it.
Mark Zuckerberg
I mean, obviously you can't— a lot of what we had to do here was rotate a lot of the rest of the company. And like for things like the infrastructure team that now, you know, it's like building these huge clusters to support this. I mean, that's obviously a huge thing. So huge parts of the company kind of had to pivot to do this. And obviously you can't do too many of those at the same time, right? Only one thing can be the top priority.
John Collison
Yes.
Mark Zuckerberg
But I think on balance, it has made me somewhat more optimistic. I mean, we brought in this guy Kunal Shah, who, you know, it's like, I think he's like one of the top entrepreneurs in India, building this great company CRED, and And now he's running WhatsApp, which I always thought should do more in commerce, and we were sort of underserved there, and India is the biggest country that we're in. But, I mean, that's— it's a bit of a different situation, because we're not, like, pivoting the whole company towards WhatsApp. We just wanted to bring more entrepreneurial energy there. This was, like, a level 10 out of 10 difficulty thing. I don't want to talk about it in the past tense, as if it's done, because it's still a level 10 out of 10 difficulty thing.
John Collison
Yeah, yeah, sure.
Mark Zuckerberg
But I think the number of pieces that kind of had to work, it was quite all-encompassing. And I think that it's both difficult for the people doing it, but I do think that that also puts some level of strain on the rest of the company, on the financial model of the company. I think like you— so I wouldn't want to conflate those 2 pieces. Right? It's like I think there's yeah, it's good to have new blood, and it's good to have people who understand the context. And you know, you were talking about people who've been around Meta for a while and understand the ups and downs. And it's like I don't know if you've worked at Meta for 10 years, you've seen some shit, right? It's like it's like we've gone through a lot of hard times, and and like I think that that's kind of important context too. But but no, I I think a lot of this is just. trying to, you know, you're— whenever you're building a team, you're— you inherently, you know, it's like Alex had a theory for what the team shape needed to be, and like, I guess I had a theory for what like the shape of the leaders needed to be for it.
Mark Zuckerberg
And, um, and, you know, it's like a year plus in, it like seems okay. Like, it's like, first of all, it was just good to try something different, right?
Alexandr Wang
Mm-hmm.
Mark Zuckerberg
And then different than what everyone else was doing.
John Collison
Yes.
Mark Zuckerberg
And yeah, I mean, that's—
Alexandr Wang
I think there's some other dimensions here that are worth mentioning too. Like, one of the— like, when deciding we had to do the full reboot, like, you know, a lot of it was, you know, one of the main inputs was external, but also one of the inputs—
Mark Zuckerberg
I don't think I've ever—
Alexandr Wang
actually, I can't remember if I told you this story, but one of the first meetings I had When I joined, I was in this meeting with some of the leads of the existing effort. And I was like, okay, I think the first thing we need to do is train a really good reasoning model by some timeframe. I don't know when, but this is clearly the most important milestone. And one of the key leads in the group who no longer works at Meta, but was like, oh, we don't need a reasoning model. Do any of our consumers need a reasoning model? I feel like we don't need to do a reasoning model. We could just do a normal model. And I had this heart-sinking moment.
John Collison
Oh, no, I've joined the wrong company.
Mark Zuckerberg
I was like, oh, I've got to take 10 steps back.
Alexandr Wang
And I think this is kind of a funny anecdote. I remember calling Nat after this meeting. I was like, Nat, like, this is what they just said in the meeting. But I think it was specifically an indicator of something that I think, like, Mark, when he talked to people, like, certainly also started to realize, which is like, we needed leadership in the company to be much more AI-pilled and much more sort of like forward-leaning and front-leaning on like where the— how quickly the technology's moving, where it's going to go to, what the possibilities are. And effectively dramatically more AI-pilled. And I think as Marc started pivoting and shifting the whole company towards this direction, and it was very clear that this is like, I think Marc got AI-pilled, and then everyone saw that he was AI-pilled and everyone realized, like, oh, this is where the company's going. Then I think you had a lot of the existing leaders like very quickly pick it up and change. And like Adam Mosseri, for example, who runs Instagram, has one of the most like aesthetic but also like effective kind of like personal vibe code and personal mini app setup of anyone that I've seen.
Alexandr Wang
Like, I think it's like he's like really gone quite far in the direction.
Mark Zuckerberg
Mm-hmm.
Alexandr Wang
And I think a lot of the leaders at Meta who have been at Meta a long time are, like, all very adaptive. Like, I think, you know, kind of to the prior point, they've been in—
Mark Zuckerberg
Yeah.
Alexandr Wang
A highly turbulent, like, period of technology. Now, AI is, like, maybe, you know, 5 or 10 times more turbulent and faster moving than those past waves. But I think they're, like, they have adapted to new trends many times. And so I think there's, like, I've been very impressed, actually, by seeing how quickly a lot of Meta leadership If I compare the kind of conversations I would have with them when I first joined versus the kind of conversations I have now—
John Collison
They're now super AI-built.
Alexandr Wang
Yeah. And it's very impressive, honestly, for that to happen at the large scale of Meta, which is one of the largest companies in the world.
Mark Zuckerberg
Yeah, we got them there pretty quickly. It was, yeah, but I mean, like you're saying, it's like if you start coding, then it's helpful. But I think a big part of what, a big part of the culture that we were trying to create is also, you know, when we were talking about how you have all these direct reports, the point of it isn't that you have the direct reports, it's that there aren't, it's the absence of non-technical managers, right? So, so I don't think you would say that you're like, you know, that you're providing a ton of managerial support to 200 people at once. I mean, that's not a feature of the thing.
John Collison
Alex, well known for— Career chats and emotional, how are you doing, checking.
Mark Zuckerberg
But I think it's just well known among researchers that, you know, there's this common pattern. It's like you go from being very technical to you go to a role where you're, you're not like in the details anymore, and then you very quickly become not technical or don't kind of understand it much faster than you probably realize. And it's quite dangerous.
John Collison
It's such a fast-moving domain, especially.
Mark Zuckerberg
I think that, like, this whole thing of just, like, you have to stay technical, people have to use the agents, people should be writing code. Yeah. I think that that's— that has been a really important part of this whole thing.
John Collison
Have you had any very funny disagreements between the two of you guys?
Mark Zuckerberg
We have a lot of disagreements.
Alexandr Wang
Yeah, I mean, we're working on naming up until very recently.
John Collison
Naming is a good bike.
Mark Zuckerberg
It's one of the things I appreciate about Alex. It's like one of his— I mean, what this is like, I think it goes back to your your philosophy, you're talking about this for recruiting back at Scale, is like hire people who really care, right? I think there's some expletives in it, but it's fine. But I—
John Collison
But Alex really cares, it turns out.
Mark Zuckerberg
I think that's like one of the things that I thought was, you know, I think that that's like really important, right? It's, so I think part of what you kind of get when you like have the supernova of like all these different talented but very strong-willed people together is like, you're going to have like fun fights, um, and not fun fights, but like, but it's like to some degree I'd rather have people push for things that they believe in and like argue over things than, you know, just have people think that they're like blindly executing something, you know. It's like people need to feel ownership. So yeah.
John Collison
How is Mark a different boss than you expected?
Alexandr Wang
Um, well, I don't have a huge comparison set.
Mark Zuckerberg
I think I can leave now.
John Collison
Yeah, exactly.
Alexandr Wang
Yeah, no, it's No, I think that like the— I think it has been very— one of the very cool things about like working with Mark over the past year has just been watching like the true degree of his boldness and like, and this trait that we were talking about before of like, you know, once he internalizes certain beliefs, like the— speed and magnitude of willing to change prior actions and pivot towards the worldview that is consistent with those beliefs. I think these are pretty unusual traits, generally speaking, even among this base of founders.
John Collison
Mark actually ingests new information and fully updates his worldview and then acts in accordance with the newly installed worldview.
Alexandr Wang
Very Bayesian.
Mark Zuckerberg
And then willing to take— Very Bayesian. Very Bayesian. I've never gotten such a nice compliment.
John Collison
What lovely priors you have.
Mark Zuckerberg
Yeah.
Alexandr Wang
Posterior.
John Collison
You're not allowed to talk about posteriors anymore after the whole incident.
Alexandr Wang
Yeah. No, but then I think also just the degree of boldness. And I think that, like, the— this is— it was, like, actually quite puzzling to me at first, the degree to which you were, like, willing to make these massive bets. It was something actually Nat and I would often, like, talk about. I was like, Wow, we're really going big here. And both of us would be like, would you go this big?
Mark Zuckerberg
Would you go this big?
Alexandr Wang
And so I think that this, I think it's like, I mean, that is quite a notable trait and something I really admire. And I think is very, I actually, if I look back at the history of Meta and I imagine what the scenario would've been in all these pivotal moments, I can see how like this is kind of the movie, so to speak.
John Collison
Yeah.
Alexandr Wang
Or like this is kind of how a lot of those situations go where it's like there's— I feel like probably the way I imagine is like there clearly is moments where like something very, very big is happening. And it may not even really be showing in the metrics or it may not be impacting user counts or impacting revenue or whatever it might be. It's not really impacting things. But clearly, if you believe that thing is going to happen, then it has massive implications for the company and the product. And I think my post hoc retelling of Meta's history is that I think Meta is a company where every time that happens, Meta takes it extremely, extremely seriously. And that has enabled the company to innovate and continue evolving with a very fast-moving and fast-changing tech landscape. Mark is very Bayesian. Yeah, very Bayesian.
Mark Zuckerberg
So Bayesian. Life is short, just do the right thing.
Alexandr Wang
Yeah. And I think AI is, I mean, I think AI is maybe that, like, you know, times 100 or whatnot. But, like, I think the, you know, a lot of CEOs would've told you last, you know, last summer, oh yeah, I think AI is the most important thing for my business. Like, I think maybe every CEO would say that, But then you look at, like, what they're actually doing and how consistent are their actions with that belief. And I think this is, like, a, like, quite, quite special thing about Meta and Mark.
John Collison
Last question, because we're past all of our bedtimes.
Mark Zuckerberg
I'm so flattered.
Alexandr Wang
Yeah.
John Collison
What do you make of all of Alex's deranged posting on Twitter?
Mark Zuckerberg
I mean, I think most of it is great, but, but I, I guess my main view—
Alexandr Wang
But it doesn't matter how How much you fail, it matters how win— how big you win when you win.
Mark Zuckerberg
I think with— so I think with communication, I think it's most important to just be authentic. And I think when you're trying to be authentic, some of the things don't come out the way you want. And I think that that is a— that it is better to make the trade in that direction than it is to be overly scripted and not be authentic. But I mean, most of it, I just feel is hilarious. So, yeah, what do you think about my deranged tweets? I'm not a heavy X user, but—
John Collison
But you are back on X now.
Mark Zuckerberg
I'm back to posting. I mean, it's like a bunch of the researchers are there and we want to be over there, but everything I post there, I also post on Threads and it's all good. But every time I load X, I have like 4 Alex posts and it just like cracks me up. It's like some of these, Like, it's like, where do you— we're just talking about this.
John Collison
Well, you're now a clearinghouse, I presume, where people send you good stuff.
Mark Zuckerberg
I was like, where do you come up with this stuff? It's like, yeah, this is—
John Collison
yeah. How does it work?
Alexandr Wang
Um, most of the— like, uh, we were, we were talking about this with my partner yesterday, and he was, you know, I think one observation was just, um, there was actually probably still to this moment, I— my, my, uh, thought pattern oftentimes is, oh, that's funny. I wonder what Jolly would think about that. And then, you know, you get good memes out of that. That being said, I have experienced the sort of like brutal metabolism of the internet and how quickly it like metabolizes memes. And so the sort of like, you know, you really have to stay on the cutting edge of memeing to—
John Collison
It's easy to fall behind.
Alexandr Wang
Yeah.
John Collison
Which is maybe a good summary of this conversation. It's easy to fall behind, you know, you gotta stay at the frontier.
Mark Zuckerberg
Stay in the details. Yeah, stay technical and stay at the frontier. Yeah. Yeah.
John Collison
Alrighty, we'll leave it there.
Mark Zuckerberg
All right.
John Collison
Awesome.
Alexandr Wang
Thanks, guys. Thank you.
John Collison
Okay, wait, so what is everyone drinking?
Mark Zuckerberg
Well, that.
John Collison
What is this?
Mark Zuckerberg
That's a keg of Guinness that I brought you.
John Collison
This is Guinness from?
Alexandr Wang
Dublin.
John Collison
Oh, wow.
Alexandr Wang
You were just there?
Mark Zuckerberg
Yeah, I was just there this week.
Alexandr Wang
Oh, awesome.
Mark Zuckerberg
Yeah.
John Collison
Well, thank you. We got the local stuff.
Mark Zuckerberg
Yeah, there you go. Yeah, no, I figured this would be a good occasion to Get the Guinness going. Yeah.