Meta CMO: How Advertising Works Under the Hood

Ads pay for the internet—and they’re about to change again. a16z General Partner Erik Torenberg, entrepreneur and author Antonio García Martínez, and Meta CMO Alex Schultz dive into growth and performance marketing, privacy myths, retail media, and the AI future of “audience-of-one” advertising—plus Instagram what-ifs, WhatsApp as a super-app, and how Meta’s feed shifted from social graphs to AI-ranked content.

Speaker names marked ? are inferred from speaking time and the show’s known hosts — they are not voice-verified.

• 0:00Speaker 00

  I'm sick and tired of all the negative talk about online ads. It drives the entire internet economy. You wouldn't have all of these companies without online advertising. It has democratized all this stuff so the startups can do little things that before you'd have to pay for a soup bowl ad to do. I've learned to stop worrying and love the automated ad campaign. I am full in on algorithm. I'm sick and tired of no one saying the positive thing. Ads is the damn business.

• 0:25Speaker 01

  Antonio Garcia Martinez, Director of Ads at Coinbase, picked up Click Here: The Art and Science of Digital Marketing and Advertising by Alex Schultz, CMO at Meta, and called it "The Definitive Book on Growth and Performance Marketing." That moment led to today's conversation. We cover why ads pay for the internet, what people misunderstand about data and surveillance capitalism, and how retail media and first-party data are reshaping the game. We also dig into AI's "Audience of One" future, where ads, agents, and even transactions happen inside the ad. And along the way, we explore Instagram's alternate history, WhatsApp as the stealth super app, and what Meta got right and wrong in its shift from family and friends to AI-ranked content. Let's get into it.

• 1:11Speaker 01

  Antonio, this conversation was inspired by you, appreciating Alex's work and thinking it'd be good to discuss his book. Why don't you share what captivated you about it and what inspired the conversation?

• 1:20Speaker 02

  How did this get started, Alex? I think either I DM'd you or you DM'd me on LinkedIn or something, or I noticed that you were publishing a book and I'm like, hey, congrats. And somehow I got a review copy. One of the great things about Alex's book is that he really understands marketing funnels, which seem simple in theory but are actually super complicated in detail. It's really hard to reconcile the super high-level thinking of someone who's like CMO at Meta or running growth at Meta, et cetera. Like, high-level, like, the theory of this, how does it work? And then the super detailed thing that's like, oh, if the email server gets backlog, it doesn't send out the email. It's going to completely hose your analytics. Or if you do server-side versus client-side logging, you're going to see completely different conversion rates for like X, Y, and Z region. And then that's the reality of working in marketing.

• 2:04Speaker 02

  It's like the super high-level strategery. And then the most detailed technical breakages you can possibly imagine. And so I think your book is brilliant at showing the full funnel, so to speak. Alex, why don't you go deeper? Why did you write this book? What are you trying to say that's not already understood by people about the company? Say more.

• 2:21Speaker 00

  I mean, the obvious banal reason why I wrote this book is there's a gap in the market. Everyone asks me for a book like this and there isn't one that I can recommend. And the publisher agreed. And my team tells me to do thought leadership all the time because I'm a CMO of a company that sells ads. They want me to do thought leadership bad. So this seemed like a really good fit. So that's the banal reason. The less banal reason, which works for you guys, and I think both of you in particular, is I'm sick and tired of all the negative talk about online ads, about our industry as a whole. And all the books that get written, especially about us, are the people who like, don't like Mark and don't like Facebook and don't like Meta. And I just actually want to have a book out there that's like, online advertising rocks. It drives the entire internet economy.

• 3:07Speaker 00

  You wouldn't have Google, you wouldn't have Meta, you wouldn't have all of these companies without online advertising. It has, you know, democratized all this stuff so the startups can do little things that before you'd have to pay for a Super Bowl ad to do, that they can try it small and then build up. And I just think it's awesome. I'm sick and tired of no one saying the positive thing. And so I wanted to do that too. So that's the less banal reason.

• 3:28Speaker 02

  Yeah, yeah, it's like what Mark said in TPN, right? Thinking ads are bad is like a luxury belief of people in wealthy countries who can entertain the notion of paying whatever Facebook's ARPU is that Alex knows and probably can't say. But let's say it's something north of a hundred bucks or whatever. Maybe someone in the United States will pay that. But if you're going to get three to four billion people online, there's no way they're going to do it. And ads is an indirect business model that works. It's not the only business model in the world. Like I used to be on Substack, I used to charge subscriptions, there's other business models. But if you want to pay for the internet at scale, ads is a way to do that. And I totally agree with Alex. It's time to end the stigma of ads. It's how the internet gets paid for. And the reality is good ads experiences, like everyone cites me the example, Alex, like even in crypto that I'm in now, Instagram ads are amazing, right? People like Instagram ads, actually. The CTRs are crazy.

• 4:14Speaker 02

  And I think one thing that we both learned at Facebook, Alex, is that like people's stated preferences and revealed preferences are very different. Right? And it all comes out in the CTRs and the conversion rate, which is one message you have in your book. And that kind of is the only thing that actually matters. And so, yeah.

• 4:30Speaker 00

  Yeah. Yeah. In general, I think as an industry, a lot of people in tech kind of think ads is like, oh, it's that thing over there that pays for everything. But actually, like ads is the damn business. And anyway, I want to say positive things about ads.

• 4:43Erik Torenberg?

  Before we do that, why don't we just steel man, just so we understand the critiques better. Maybe even Tony, you could take a stab. It's been a few years since the critiques were in full swing. But what were people so concerned about with ads? People had all sorts of outlandish claims about it. Can you sort of recall and steel man their side and then we could respond back?

• 4:59Speaker 02

  I think people feel a sense of mistrust and violation that their personal data is being used. I mean, I even wrote a Wired piece about it because it so annoyed me, like the rumors that Facebook is listening to you through your microphones, which is like completely ludicrous. A, you'd be able to tell on your phone that it's like constantly talking to the mothership and it would consume so much network bandwidth. Secondly, the reality is like, even in this conversation, what are the chances that anything I say could actually be used in ads targeting, even assuming perfect godlike knowledge? Not very much. And the reality is, what actually happened, and people are still asking themselves a question, it's like, you search for something, or maybe your spouse or partner actually searched for something on the same machine, and maybe you're actually in the same lookalike cluster on Facebook, as the case may be. So there's other things going on that you actually do have control over that aren't as nefarious as anyone listening to you in your microphone.

• 5:45Speaker 02

  And so they just don't understand how it is. I do definitely think that Alex, like a certain level of transparency in what data is being used is often not clear to most users, like they just don't understand it. And I think a lot of web two ad tech, I'm not putting Facebook in that bucket, by the way, there's other way more sketchy forms of ad tech that I do think do sell data. But like, it's not even true that Facebook sells data. It doesn't. So yeah, I'll stop there. Alex probably has more informed comments here.

• 6:09Speaker 00

  I think that steel man is like the Zuboff surveillance capital. Like, that's what they're describing it as surveillance capitalism. You know, you are the product. You don't understand what these companies are doing. That's all you're doing. That's the steel man. Oh, and by the way, you're manipulated by all of these marketers because people don't like average advertising either. It's not just online performance advertising. It's just like, it's advertising as a whole. So then there's the meta critique of the marketing industry as a bunch of people who drink martinis and convince you to buy things you don't need.

• 6:38Erik Torenberg?

  I think there was also another critique in addition to the ones you're making, which Is that it shapes the content itself that because it's a business model, it sort of it like dumbs down the content. It makes the incentive to create, you know, sort of the cat with, you know, sort of BuzzFeed lists, et cetera, the cat videos and stuff.

• 6:55Speaker 00

  This actually is part of the core of the Zuboff surveillance capitalism thing. That's actually part of the core is because your incentives are to get more time spent in attention and clicks, therefore, you know, you are going to do stuff that's nefarious, that incentivizes bad content and so on. The thing that blows my mind, well, anyway, we have, we are not, we're not attacking the steel man yet.

• 7:15Erik Torenberg?

  Yeah. Well, I mean, one thing I'll say is it seems almost the opposite in the sense that I think we have some nostalgia for like subscribe, you know, New York Times moving to a subscription model didn't quite make the New York Times more sort of, you know, detached and objective. When people, you know, when you're charging, you know, a certain demographic, you're going to want to tell that demographic what it wants to hear.

• 7:34Speaker 02

  If you look at the history of advertising, particularly newspaper advertising, which you go into a little bit in the book very briefly, it was actually the New York Times was originally a platform for advertising, right? If you look at 19th century American journalism, it was overtly partisan, right? The press Democrat was the Democratic Party's newspaper. And if you were a party member, you'd subscribe and that's fine. And everyone understood that if they read that version of reality, it would be very biased. If you still go to Europe now, most European countries have a left and right wing newspaper. In the UK, they have 12, but you can still see a sort of partisan media here. A lot of why American press became objective was that the New York Times could sell ads for Macy's without offending anybody, right? Because the Macy's didn't want to run an ad alongside some screaming editorial for or against this or that president. And so in many ways, ads actually kept media honest

• 8:19Speaker 02

  in the sense that you couldn't take an overt position. But once you have a subscription model, and I think, what's his name? Ben Smith even said this, the New York Times then becomes a home for compelling narratives that people are paying for. And that in some sense, that actually skews media more than an ad model would, in my opinion.

• 8:36Speaker 00

  Completely right. I actually think that is completely correct in terms of the incentives of the models. But this is why we, like, these are the critiques. And I think a lot of the critiques, when you actually get into them are negative. The biggest piece for me about attacking like the critique is growth is good. Growing the economy is good. Growing the economy gives people better lives. It provides taxes. It improves the world to have a growing economy. Now, look, if it's out of control, if you've got rivers catching on fire, none of that stuff is good. You do need some level of like regulation and you need to figure out the balance. But I think we've got to a place where online advertising has been so attacked. It's as if it's Satan and actually it just helps people get slightly better ramped ads for trainers,

• 9:18Speaker 00

  which is a good thing because it sells more trainers, people get products they love, companies employ more people, the economy grows, it's just good. So that's the point I want to make out there. Advertising is a good thing. It's useful.

• 9:29Erik Torenberg?

  Alex, you're in the UK, so your pro-growth arguments are thought crime at the moment.

• 9:35Speaker 02

  I wasn't going to say that, but I was thinking it, Alex.

• 9:37Erik Torenberg?

  I'm just teasing.

• 9:39Speaker 00

  Look, I get the point. I have seen CoinBolt's latest work in the UK. I think there's a lot of very pro-growth people here who are fighting for it, who I'm pretty supportive of. And I encourage you to read the ukfoundations.co essay. That's a very good essay.

• 9:54Erik Torenberg?

  I know, I'm just teasing. Yeah, a lot of great companies, pro-growth ethos. Antonio, so you also wrote a book about Facebook, obviously we referred to Chaos Monkeys, you know, over a decade earlier. If the book had sort of a new foreword or an updated edition based on what's happened at Meta and the advertising industry since then, what do you think would be different?

• 10:12Speaker 02

  I mean, I don't know that the ad side would actually be fundamentally different. I think, I don't know how much we want to get into this, but I think the attitude towards, I mean, what's changed in the past five years is the attitude towards content moderation, not of ads content, but of organic content, right? And to what extent should platforms put a thumb on the scale and say, this is truth or not? Or you know what? We're going to be an open forum for ideas and other than, you know, bad content that nobody wants like porn and violence, we're not going to do anything. I think that pendulum swinging is what has changed in the past 10 years, not really the ad side of things.

• 10:43Speaker 00

  Yeah, I mean, I think that the biggest, so yeah, I agree with Antonio's point. A pendulum swings and it swims. I think like the revolution in social media in the last five years has been the, it was all connected content five years ago, like it was people you follow. And now the combination of the modern. AI systems that allow you to have semantic understanding of content and do ranking based on semantic understanding of content, combined with short form video means the majority of meta, and I spent a lot of time testifying in court to this fact, the majority of meta at this point is actually unconnected content that people are engaging with, the majority of people's engagement. And I think the TikTok factor is the other massive thing

• 11:27Speaker 00

  that's happened to the industry in the last five years. It's totally different. We have been complete, I mean, people don't understand how completely disrupted as a company we've been. Fundamentally, five years ago, it was all connected content. Now, the vast majority of engagement is on unconnected content.

• 11:43Erik Torenberg?

  You define unconnected and connected for the listeners who maybe don't know what that means?

• 11:47Speaker 00

  Yeah, I kind of think of the world as a sort of two by two with one cell hopefully never existing. So connected and unconnected. Connected content is someone you follow, a group you've joined, a page you've liked, whatever. Unconnected is something you've not followed, joined, connected with an N O I friend, did whatever. And then private and public is obviously private is it's restricted to some small group of people. Public is it's open to the world. And if you think Facebook started in connected private, so it was friends, it was private, even, you know, small public networks of your university, but no bigger. So connected private. If you think about it, like Instagram started as connected public, right? It was like photographers and celebrities and whatever. Twitter started there as well.

• 12:33Speaker 00

  And then TikTok started as unconnected public. And each of these companies have moved and now play in every one of the zones, like TikTok has its friends tab. It does an awful lot on message sending and DMs. Instagram has reels and Facebook has reels, which are completely unconnected content. And so what's happened is each of these social media companies started in a different quadrant and they're all competing across all quadrants now.

• 12:59Erik Torenberg?

  It it seems, yeah, social networks in general, but we also matter have shifted from, you know, primarily or mostly, you know, friend to friend to to content feed over time. I'm curious what that, you know, means for switching costs and network effects and and more broadly, the way I sort of read, or one way to read sort of meta's big push into in AI recently is is just the sort of the threat of a world in which people are communicating less and less to their friends and more and more to to chat G P T or or etc.

• 13:34Speaker 00

  How do you react to that? I mean, I think there's two or three things. One thing is, I mean, we just posted a very significant earnings result. That could not have happened without the modern AI systems in two different ways. The first way is they drive engagement. So getting more time spent and more engagement with reels is really, really what's driving our usage growth at this stage. But the second element is ad ranking. And so, you know, I mean, 10 years ago, I would not use automated ad campaigns to save my life. I'm like dead set against them. Today, I've learned to stop worrying and love the automated ad campaign. I am full in on algorithm first.

• 14:15Speaker 00

  And the place where the creativity lies is what data do you feed to the algorithm once you teach the machines? All of that relies on the modern AI systems. And so if you don't, when people say we don't have a reason to invest in these AI systems, I'm like, first and foremost, our core business is all AI ranking and has been since I've been there. You know, it's always machine learning after it shipped and it's AI until it shipped, until like the last two years. And so like, it's our core business would be the number one point. The number two point would be, um, yes, I mean, I think people are going to be talking to AI chatbots a lot. And I think down the line, there will be a feed of AI content. You're already seeing a lot of it on TikTok and Instagram, Facebook, Twitter, some on Snapchat.

• 14:58Speaker 00

  People are really interacting and liking the right AI content. And that's only going to get better because these things are definitely good. Even if scale and laws end, these things are good at media. Um, and so I think there's two elements to it. One is like, it's our core business, ranking things. And two, yeah, there's really engaging AI stuff out there and we need to be on top of it. Just like we were the shift to reels, just like we were the shift to stories, just like mobile, just like Messenger.

• 15:26Speaker 02

  Yeah, I went coming on that, Alex, because you have a you have a section at the end of AI that I thought is really interesting. And one of the concepts you introduce is, I forget how you frame it exactly, but it's a world where the there's an audience of one, right? Which is like the media I'm consuming is literally hyper personalized, like literally me, as if the person. And you always had a feed that was ranked differently person to person, but you're still engaging with not exactly mass media, but media that is being seen by more than one person, while AI totally changes the dynamic. And, you know, in the past thirty years of ad tech, we've gone from something that it felt like newspapers, like everyone got the front page of the New York Times, to something that feels like, you know, Facebook, which is a ranked feed, but it's the same content. To one where, no, no, literally, nobody else is seeing this except me, right?

• 16:09Speaker 02

  I mean, the underlying influences that bubble up through an L M model or through search indexing, you know, um, might be common, but what you're actually visualizing is different. So I'm curious if you thought about it, because I've gotten interested in what AI ads will eventually look like. Having some sort of sponsored offer or surfacing or reranking content inside your AI result seems like it'd be really interesting. And to the extent the web browser goes away, like I almost don't even use Google anymore. I'm really a chat G P T guy. Obviously, it's gonna be a channel, you know, you you you mentioned A I D A, right? The the I side of it, the intention side of it's happening inside AI. I'm curious if you thought about how that plays out and how advertising kind of gloms onto to AI.

• 16:48Speaker 00

  Yeah, I mean, I think I actually think this is gonna be super, super interesting. There's, I think there's two major interesting sort of things to think about. One is what happens to search, which, you know, Google's very on top of, like, they're very worried about the innovator's dilemma they find themselves in right now. But if if search ads goes to a paid model like chat G P T, um, and people start asking their questions in that way and they pay for it, that is very disruptive to online advertising, um, and who knows where that lands? Or does it go the other way where people are using more and more search and there's a freemium model and there's ads associated?

• 17:29Speaker 00

  In which case, the AI ads will have a lot of principles in the search ads. And so which way that goes is one big question. The second big question is, yeah, what happened to the creative intersected with the target? So my perspective is it gets more engaging, more personalized. You can convert in the advert, like in a world where you can tell an agent, hey, I want to buy this thing, take a picture of me or look at my camera, this is my size, like evaluate it. By the way, you've got all of my past purchases associated with me and you can therefore make sure it's the perfect fit for this particular retailer because you can mirror it with everyone else. You can get the conversion experience right in the interface where you're in, start the

• 18:09Speaker 00

  buy flow, complete the buy flow from an A genetic perspective or honestly just a you interacting with the other company's agent perspective. So I think you're going to find a completely personalized ad will happen, targeted entirely to you, that's customized to you, and you can convert in the ad by interacting with the company you're trying to buy from through an AI interface. I mean, the crazy end world of this is ads to AI agents, which I think is definitely also going to happen within our lifetime. But that can work, right? Yeah, the AI agent is doing a search, there are ads for the AI agent to click on and pay for placement where you're promoting stuff to someone's agent.

• 18:51Speaker 00

  I mean, this is all very likely and the underlying principles of online marketing are actually going to hold like AI optimization versus search engine optimization, really actually a lot of similarities and a bunch of companies trying to do the same thing there. AI search ads versus search ads, they don't have a lot of similarities. And when you talk to a chatbot, why is that that different than SMS marketing, push notifications or snail mail back in the day? There's a lot of similarities. It's gonna be wild. I mean, you just think about, I think the way you think about it as like intelligence too cheap to meter. Like, there's a lot of people who are afraid of loss, right? You look at the ads industry and every agency like freaked out when Mark said, oh, someone

• 19:33Speaker 00

  will, you know, use AI as their agency and will use a credit card to buy. Now, big ad agency clients don't use credit cards, so it clearly was talking about SMBs and not big clients, but, but. You're going to be in this world where I think a bunch of stuff that today is not ROI positive because it's not worth staffing it suddenly becomes possible and becomes ROI positive and AI will unlock a bunch of abundance in terms of us being able to make things possible like creative for SMBs, like having a chatbot. I mean, we're in this place where we're replacing our content moderation systems with AI from old ML and we've just been watching the curves

• 20:14Speaker 00

  get closer and closer and closer and they finally crossed over on a few cases and they're doing better, better than experts even when we use expert training data. The same is true for customer support. You can now actually get customer support in the US for Facebook. It is possible via a chatbot and that is entirely AI and at the end passes over to human beings. It's suddenly possible. We could never staff it with human beings. We can now actually start something that wasn't possible before. And yes, there's incrementally a few more staff, but the workload is that of 100,000 people. The workload isn't that of like 100 extra people. So I think there's going to be just a ton of stuff that's unlocked that today isn't ROI positive, but you add in an AI tool

• 20:55Speaker 00

  that makes you as a human being more efficient and you're going to suddenly make things ROI positive that were never ROI positive before.

• 21:02Erik Torenberg?

  When we think about the future of advertising, like is there going to be a world where I have the option to just get a ton of things for free if I just ingest ads? I go into an Uber and I don't have to pay for it, but I'm just looking at ads. Then I go to a restaurant and I don't have to pay for that, but I'm just seeing ads everywhere. Like how more integrated can you see ads to the they experience more so than it is now?

• 21:27Speaker 02

  I mean, the classic terrible web two ad tech idea was paying the users for their data. I mean, that's flawed for a bunch of reasons. A instantly gets farmed and you have massive fraud. Also, the amount of the value of the users' data that they actually own isn't nearly as valuable as they think it is, right? Like it's...

• 21:46Speaker 02

  It just isn't that valuable in the scheme of things if you're actually shot. Like data is not the new oil. It actually isn't this commodity product that you can just sell. Like even if I managed, let's cite a third party company who get it so we don't get into trouble. Let's say I managed to take Amazon's conversion record for users and steal it and tried to sell it. I mean, somebody might find it meaningful if they managed to match the identities, but it's not actually. That is worth billions of dollars within the context of the Amazon buying experience and the customer relationships that they have. On its own, it's actually not worth a huge amount of money, right? And that's, that applies even more so for like literally the past 30 days of my browser behavior. I don't know, how much would you price the past 30 days of my browser behavior? A few bucks, 10 bucks maybe at most.

• 22:28Speaker 02

  It just isn't, it isn't enough money to make anyone switch off a browser, say, and go to this browser versus that one typically.

• 22:35Speaker 00

  Well, I think what's interesting, a few things is interesting in this. I agree, debt is not the new oil, data is sand. And sand is only valuable if you're refining silicon chips, right? So, and by the way, data isn't destroyed. Data exists forever. Oil gets burnt once and it goes away. So there's a ton of reasons I really agree with what you're saying there. And then the, the thing you're saying, I've literally got an example in the book of incentive affiliates, where at eBay, when we paid people to get confirmed registered users, incentive affiliates would go out and pay people to sign up for eBay and confirm. And then they would give them money back. And then we changed it to activated confirmed registered users. And that caused us a huge amount of credit card chargebacks.

• 23:17Speaker 00

  Cause activation was buying or bidding on something and people would convert, activate, and then they would charge back once they got paid back by their affiliate partner. And so a hundred percent here lies a lot of issues. On the other side of it, I really do think this is the point about online ads being awesome. You live in that world today, Eric. Like Google search, Facebook, Instagram, TikTok, Snapchat, Twitter. Like these are incredibly YouTube. These are incredible products that you get completely free. Not like cable where you pay and get ads. You get them completely free and you get them completely free because ads makes that possible. And so I think it will go further. I actually agree it will go further. At the same time, I think there's a lot of risks with it.

• 24:02Speaker 00

  If you don't think about the fundamental principles in my book in dealing with affiliates, because like there's a lot of ways you can be defrauded and I have a lot of examples of fraud.

• 24:14Erik Torenberg?

  Hey Alex, why hasn't Meta built a one-integrated super app like WeChat? Why are Facebook, WhatsApp, Instagram siloed?

• 24:27Speaker 00

  Well, two thoughts. One, I would actually say that WeChat doesn't really integrate Instagram and Facebook in that meaningful way. So I wouldn't smash all of social media together into one super app. I think there's a few comments here. One, actually WhatsApp is disarmingly close to that, and people don't understand it. So WhatsApp is the place, the single largest broadcast-sharing channel on Earth. The WhatsApp status tab is bigger than Instagram or Facebook in terms of the amount of content shared. It just so happens it's not in the US, so Americans

• 25:06Speaker 00

  don't notice it, but it is massive. WhatsApp has payments in India. It just got awarded the best payments platform in India. The government fought us for a long time on that one, and now is really supportive of WhatsApp payments, which is a wonderful turnaround. So WhatsApp actually has payments in India. They're growing very rapidly. The interesting thing, though, about WeChat is going back to growth. Why did WeChat turn the corner so dramatically on payments? And it comes down to two things in the Chinese market. Thing number one is China has red envelopes, and nowhere else on the earth had the red envelopes, cash in envelopes that you get with the... Yeah. And thing number two is China had not the credit cards. And so what

• 25:51Speaker 00

  happened was you had to connect your bank details to WeChat if you wanted to partake in WeChat red envelopes. And WeChat allowed people to give their friends red envelopes, but also gave people red envelopes. And one in 50, 100, whatever, had a large chunk of cash in it, but most of them didn't. And WeChat got people to connect their bank accounts before they could receive the red envelope. So WeChat wired up the payment system in China in a way it had never been wired up before through this viral growth mechanism that couldn't exist anywhere else on Earth. So we are building a super app. It's called WhatsApp. It is the largest place for broadcast sharing on Earth, and it has a lot of success in commerce and, in particular, in payments growing.

• 26:35Speaker 00

  But it's a lot harder to do it anywhere else on earth than it is in China.

• 26:39Speaker 02

  Yeah, I was gonna say, I mean WhatsApp really is like the business communication layer. And it's funny, Americans don't use it because they all use messages, which is funny because it's like only Americans would flex using an inferior product at my time and say, oh, no, no. Like there was this viral tweet, it's like, oh, I don't live in a third world country. I don't use WhatsApp. It's like, bro, messages sucks. I can't even synchronize across two devices, the message thread and WhatsApp, I never had that problem. But like I was in Lisbon randomly for a conference and like the dry cleaner on the corner only did it via WhatsApp. And like I had to WhatsApp them and like somebody magically appears, the clothes came back all via WhatsApp. Like no other touch point WhatsApp, the whole thing, right? And I assume in India it's probably even more sophisticated than that. But I think people in America don't quite understand like calling a business is like backwards or like even text like,

• 27:22Speaker 02

  no, no, no, no, no, there's whole business WhatsApp accounts where you can just transact business with WhatsApp. And that is the communication layer. And like you said, also as the social thing, right? Like group chat started as like a WhatsApp thing. And it's like, I also did a Wired piece, you know, it's like Facebook got so much flack during the whole post 2016 thing for like feed and ranking. It's like, bro, there's no ranking in feed in WhatsApp and it still causes both good and bad. Like technology always has positive and negative externalities. And so of course there are some negative externalities that come out of tech, but that all happens on WhatsApp too. It's not the ranking. It's just putting a network computer in everyone's hand and connecting everybody to everybody is obviously kind of a shock to the system. Right? And so that's, it's always gonna have, you know, it's gonna convulse the system a little bit. So you see some negative outcomes, but net, I think it's positive.

• 28:07Speaker 02

  And again, having a communications layer that everyone can communicate on, that's just, to me, it's crazy, right? And WhatsApp is that, is that layer. So one thing that actually has changed in the past 10 years of ad tech is retail ad networks. And you mentioned a little bit inside your, inside your book, Alex. And for those who weren't familiar with it, and I haven't played in that pond too much, but one of the unique things, and it actually reflects another topic that we haven't gotten into, which is privacy and GDPR and all that. But basically inside large retailers like Walmart, like Amazon is probably what? The number three ad system after Meta and Google at this point, or like number three, number four? And it's. And it's unique because what Retail Ad Network means is you're showing ads inside

• 28:48Speaker 02

  a retailer that already has a first-party experience with the user, which again is somewhat unusual, right? Like typically, the classic ad network is advertiser A, advertising and publisher B, and A and B have like nothing to do with each other. But in this case, advertisers are actually paying for placement inside like a native Walmart shopping experience. And part of it is it's very efficacious, right? Like you control the entire, like you say in your book, friction kills everything. And so having a single form of payment, whether it be Amazon Pay, Walmart, or Shopify, for example, who also does the same thing, is absolutely key. But the other thing, of course, the way privacy law works, if you have a first-party relationship with the user, you can do things with data that you can't as a third-party. I mean, arguably, it's a good reason for that, but it does skew the market

• 29:31Speaker 02

  a little bit. Walmart can sit there and construct a certain segment around users because they can legally do that because they have a first-party relationship, but then they leverage that data for a third-party who's buying an ad inside that system. And so it's a combination of both easing the funnel and frankly, you know, getting through existing privacy law that's kind of driven that. But I'm curious, you know, what you think about that? Because that world, I think, didn't really exist when I was at Meta, and it's only come recently. I'm curious what your take is on that.

• 29:59Speaker 00

  Yeah, there's one other thing that you're missing off, which goes on deeper in the Retail Ad Network funnel than necessarily than Walmart and Amazon. But like the smaller retailers that do it, they also buy off other companies with the data they have on first-party. So the Retail Ad Network becomes a third-party ad network as well, which is space for opportunity for things to go awesomely and space for opportunity for things to go really badly in terms of fraud attribution and your ads showing up in the wrong place. Yeah, I mean, I think when you look at it from the perspective of an FMCG company, and I'd add in for a lot of quick-serve restaurants, QSRs, DoorDash is a Retail Ad Network as well. Uber Eats

• 30:43Speaker 00

  has a Retail Ad Network as well. And so they're seeing this situation where they're getting a double bite at the cherry from the people who are retailing their goods, both at Walmart and Amazon, and also with Uber Eats and DoorDash, where they have to pay for placement to promote themselves, and they have to pay a commission to the vendor. So this sounds terrible, right? This sounds like it's a real squeeze from the distribution channels onto the FMCG companies and the quick-serve restaurants. But it's also just the history, and this gets back to like a lot of what I feel and have put in the book. This gets back to the history that Benedict Evans likes to say, Amazon got the CS catalog from 1950 and just implemented everything in the CS catalog for the internet.

• 31:25Speaker 00

  You paid for placement in the shelves when you were on Walmart in the big mega store. You paid for placement in the CS catalog. Retail ad networks have existed for as long as retailers basically have existed. And so all it's done is modernized that and brought it online. It's going more to below the line media budgets. So kind of direct response, placement related media budgets, discount related media budgets, and less to above the line typical media buying. So it's competing for different budgets, but it's a major, major factor in terms of the ad industry like now. And the whole thing does boil down, as you say, to data and algorithms. And first party data having advantage over third party data.

• 32:07Speaker 00

  So these big companies have a really good legs up to be successful in this space.

• 32:13Speaker 02

  Which if you're being charitable, you would say, well, that was a good outcome of GDPR, right? I mean, it's, it's true, like the sketchiest part of ads was always the third party data vendors. I won't name names, but we know who the companies are or were for that matter, because the world has died a little bit. But like that part of it always kind of sketch me out a little bit. I think the reason why I think you should privilege first party data is because if you have the user relationship, you'll, you'll safeguard that user relationship because you have a long-term incentive in making sure that user is comfortable and doesn't, you know, bail or turn out of Amazon Prime or stop buying on Walmart and buy some targeted instead, right? They're not just a random user that you have a very transactional relationship with. There's actually like an arc of a relationship there. And if they were to churn out, that same growth leader is going to get dinged in the next report. Like, well, why did their, why did their retention rate go down to 75% or whatever?

• 32:58Speaker 02

  It's like, well, because you burnt them out on ads or the ads were terrible or what have you, right? So I think it actually aligns incentives very nicely.

• 33:05Speaker 00

  Hundred percent it is, but it is directly incentivized in the privacy laws, but yeah, hundred percent.

• 33:10Erik Torenberg?

  Is there anything else worth sharing about how the industry is going to change with the platform shift today? I also gave a couple of examples earlier, but anything else high level or concrete that's worth mentioning?

• 33:22Speaker 02

  Yeah, I think ads to AI agents are going to be nuts. I think that the interesting question is you have to pass that personal context somehow. Like you mentioned, like, it's funny, you know, one of the interesting exercises is go ask your favorite chat GPT, like your open session when you're logged in. So, but what do you know about me? And it's amazing how much it knows about you. And it's not even scraping the internet. It's just like literally, so what do you think my political opinions are? What do you think my taste in clothing are? And it just sits there and rattles off the most informed opinion that knows me better than my fiance does. And, you know, how you how you activate that in a kind of ads sponsored sort of way, I think will be interesting. I think there's going to be billions of dollars of billable hours for privacy lawyers. So they have to figure this out. But at some point, you're going to get to the point where, in fact, you have even better personalization than you have now.

• 34:07Speaker 02

  And again, in a way that you yourself control and is probably centralized in one place and all that good stuff. But it'll be interesting to see that play out. It's funny, though, cookies are still around, Alex. Everyone's been predicting the death of cookies for 10, 15 years, and yet here they still are.

• 34:21Speaker 00

  And Google was reverse course. So that that's like actually quite pleasing on that one. Yeah, I think this point is is also like, if you think about which ads are going to be disrupted, you've already got a bunch of disruption happening for Google Search, for Meta Advertising. You know, there's going to be more disruption, there's going to be more creative stuff happening. We talked about that already. I think the thing we haven't talked much about is direct mail. And now direct mail was a letter, and then direct mail was an email, and then direct mail was an SMS, and then it was a push notification, and now it's a chat thread. And has a lot of stuff in common across all of those different different areas. But when your direct mail is sentient and can interact with you, that is going to completely change the value of direct mail marketing

• 35:05Speaker 00

  and having a chat thread with your client, with your customer. And so I think that is the probably the field inside marketing that's going to get revolutionized the most. Because I mean, definitely, I mean, I was talking at one very large, very large luxury cosmetics company at Pan. And they were saying in one country, like of their D to C sales, 30% online were coming through Chatbots now, a large company. And the startups in the beauty space. There are some that are totally chatbot first in terms of sales. And if you look at India, if you look at Thailand, so much of what's happened in tech came out of Southeast Asia and Asia in general.

• 35:46Speaker 00

  So, look, Kakao was the first real breakthrough in messaging. Yes, there was WhatsApp line, WeChat, all the others, but Kakao was like the first one we looked at that woke us up. Kakao Story was there before Snapchat brought it to the US by a year. And then Snapchat made it disappear, which was really smart. But Kakao Story was first. You know, TikTok came out of Southeast Asia. You're in this place where, at the moment, you're looking at TikTok Shop. You're looking at some of the other stuff that's going on that's come out of Southeast Asia. And the stuff, or Asia in general, sorry. And the stuff that I'm looking at a lot is, for businesses, they are now replacing a chat thread with a human being, with a customer, with a chat thread, with a chatbot, an AI chatbot. And in the US, that's not working.

• 36:31Speaker 00

  And in Western Europe, that's not working. But in Asia, that is working right now. And in the developing world, Brazil, Africa, that's working really well. And it's scaling. And so I think that's something that will be really big that comes across is direct mail will be completely revolutionized by AI.

• 36:48Speaker 02

  It's funny you mentioned direct mail because it's, I mean, direct mail is where this all started. I mean, actually, in Chaos Monkeys, I kind of draw a line between, at the time, I don't remember, Alex, when we were doing like custom audiences and stuff. It was like, yeah, Bed Bath & Beyond, and there's like 20% off coupons or whatever. And the reason why direct mail, people were wondering like, what the hell is direct mail about? The reason why is because the primary key to use database language for the American consumer was the address, right? And that was the way that you would address, like a lot about ads, to get a little philosophical about it, is about identity and addressable media. Like, how do I actually reach out to that person and then follow that identifier throughout the entire conversion funnel? It used to be literally the address, because then you'd order something, it'd arrive at your address. And that would be, that was the attribution system, right? You received the thing, and then that would be cataloged by that.

• 37:32Speaker 02

  But of course, everything going on the internet meant a total reshift, whether it be mobile device identifier, Facebook user ID, cookie, what have you. It's just interesting that how the notion of who you are as an individual actually informs a lot of. The marketing and how in the crypto context, just to go to what I do right now a little bit, your on-chain wallet address is kind of that identifier, right? And that's a whole different namespace for the user. And so it's just interesting to see how, I think you say this in your book, the details, the details change, but the fundamentals always persist, right? So like the primary key on the database shifts from mailing address to cookie to mobile device identifier, to Facebook user ID, to blockchain address. But the actual principles don't change, actually. They're exactly the same across all these, across all these channels.

• 38:17Speaker 00

  Could just say that to myself, 100%.

• 38:20Erik Torenberg?

  So I want to play a little bit of alternative history, but it's like four or five questions. What if Instagram was an independent company and wasn't bought by Facebook? Antonio, because I'm not sure if Alex can comment on it. Antonio, imagine-

• 38:35Speaker 00

  I can comment on it, the antitrust charge is over and I just spent two months in DC. I can comment on it.

• 38:41Erik Torenberg?

  Great, so maybe, imagine a world where Instagram said, no, play the Snap route. What would that look like?

• 38:49Speaker 00

  They look like Snap, maybe smaller. I mean, I think like if you look at what Kevin has testified to, like he's explicitly testified to, he said it could go either way. You know, on WhatsApp, I think WhatsApp was tremendously successful. And I would not say that I, as a growth leader, did much for the success of WhatsApp. We helped them a lot with infrastructure. I think we kept them free. I think there was a bunch of things that we enabled to happen for WhatsApp. But bluntly, WhatsApp was a great, great product with higher retention. Instagram, I think, was touching go as to whether it could be either a Twitter or a Snapchat or just a complete also-ran like path. Like when, when we announced we were buying it, there was a massive spike in their growth just caused by the PR.

• 39:33Speaker 00

  Like literally the global news that Meta was buying, Facebook was buying Instagram, caused their growth to accelerate. And there's a whole stack of things I can go into in the details of like, their infrastructure was on the verge of falling over. They were in a position where they had real, real issues with spam filtering and massive issues with comments that they were failing to get on top of. Now, the flip is obviously it could have gone either way. And that's exactly what Kevin said. And I, I can see how it could have gone either way and other companies have got there. But, and Mikey and Kevin were astounding. We can see it from what Mikey is doing today. But Mikey and Kevin were just so brilliant. And I loved working with them. At the start, we fought so much. Like Kevin and I fought so much. And by the end, I'm like, he's one of my favorite people to work with.

• 40:18Erik Torenberg?

  You fight Alex? No way. I don't believe it.

• 40:22Speaker 00

  I know. But we did it. So we fought. And by the end, we were like powerfully aligned. And so I think it could have gone either way. My gut is it would have been maybe Snapchat, maybe Twitter. It wouldn't have been zero, but it certainly wouldn't have been the Instagram it is today.

• 40:38Erik Torenberg?

  What were some of the biggest, uh, what were the crux underlying the fights?

• 40:42Speaker 00

  So there was some organizational stuff which literally came up in the tribal. About, um, my team really, uh, so you need to have a product and engineering team on growth. You can't just have a marketing team. And the magic in Meta is that we have really made it a killer, a capability that is so ingrained in the company that every product and engineering team gets growth now. And it is really special about the company and it's different about Meta. And I have a product growth team, but it's only like a hundred, 150 people for an 80,000 person company. It's not a massive, massive team. Who then go in and help teams implement growth best practices throughout the company.

• 41:23Speaker 00

  At that time, Instagram didn't have a growth product and engineering team. And so we had a big disagreement over that. In the end, they built one out. It was really successful for them. Um, and I'm really, and actually it's the place where we ended up being super aligned and super happy with each other, but we don't have to fight. The other place that was massive was called the connections pivot. So Instagram was focused on public figures and it was not focused on friends and family. You remember it was public photographers and cool celebrities and everything. And we said, no, people you may know, which on Instagram is accounts you should follow or suggested users, is really important to get that happy blend of friends and family with public figures, you know?

• 42:06Speaker 00

  And Instagram is a modern social network where there wasn't a distinction between business accounts and friend account, whereas Facebook, MySpace, others, they had some of these distinctions back in the day. And so we had a disagreement over like importing contacts, focusing on friends, all of those things, focusing on average, make follows and all those. So those were the two, I think, larger disagreements that ended up being quite front and center in the antitrust case. And so I feel like I can, I can mention them because they have been covered in the press and they are literally a matter of court record at this time.

• 42:40Erik Torenberg?

  Okay, continuing our alternative history theme, and Antonio, why don't you take this one? What if Libra was allowed to sort of exist in its full vision without regulatory challenges?

• 42:51Speaker 02

  Yeah, I mean, I wasn't there for the Libra period, to be honest, but obviously I'm at a crypto company now. It's amazing how much the regulatory landscape changes everything. I think recently Stripe announced it is building an L1. Every company and their brothers used to be announcing a stablecoin. We had major crypto legislation passed that was favorable to crypto. We've had members of the admin say that they want the US to be the capital of crypto, which obviously is something I'm supportive of. It's just, it's, yeah, I mean, I, I think if Libra were to launch now or something, it would be a very different conversation. And it seems like it was doomed from the start at the time. It's a shame. But yeah, I think you're just start seeing crypto be, be sort of the rails under the hood.

• 43:36Speaker 02

  People are going to be on chain and not even realize it, actually, which is fine. And actually arguably is the way you should build these things. And I think it's, again, what's the positive shine on all that? I think I tweeted about it yesterday, right? The fact that you're going to have investors around the world who can hold dollars freely or who can invest in American equities but have them on chain with all the, all the rigmarole that it takes to buy a US stock in some other country, that's going to be magical in a way. And it's going to open a level of economic freedom that you haven't seen to date. I know that's probably a little bit beyond the purview of Whatever Libra's goal were back in the day, but just to give an example of how it actually opens up the world and makes it freer and drives more economic growth. That's kind of a good thing. And that's kind of part of the reason why I got crypto pill. It's like, man, it just seemed like a lot of traditional institutions are actually choking innovation rather than fomenting it.

• 44:21Speaker 02

  And sometimes you kind of need a different paradigm to kind of kickstart that process. And I'm old enough to remember web one and web two and now web three, if we're still using that term or on chain or whatever. Yeah, I think obviously Libra now would be a very, a very different scenario.

• 44:37Speaker 02

  Maybe Alex has thoughts on this one.

• 44:39Speaker 00

  I had one thing, which is, look, it's public now that one of the places we go when we are sourcing terrorist material to take off of Facebook and Instagram is Telegram. Like we go to Telegram, we subscribe to the terrorist groups. We take the data and for prescribed organizations in the United States and United Kingdom and so on. We take down the terrorist information. That is illegal content by anyone's standards. It's not sort of one of these debatable areas of content iteration. If you were in a place where Libra had succeeded, you would have WhatsApp, Messenger, Instagram,

• 45:17Speaker 00

  front ID, US friendly messaging apps be the successful ones with crypto and instead you don't. And I, I think that probably matters to national security and matters to a lot of things that the people who beat us up on this actually care about.

• 45:33Erik Torenberg?

  Continuing our alternative history, Antonio, we'll start with you. What if Hillary had won the 2016 election?

• 45:40Speaker 02

  Oh my God, I'm not even going to go into that topic. I have no idea.

• 45:43Erik Torenberg?

  Absolutely no clue. Is there anything? Totally different.

• 45:48Erik Torenberg?

  Totally different. I have no idea. Going back to sort of AI for a second, it seems that mobile was a more of a sustaining technology than it was a disruptive technology. Sure, we got Uber and Snap and Instagram, but more of the gains went to, to Facebook and Google than, than the entire startup industry from the time. Do you think something similar is going to happen in AI? Maybe Antonio will get your first take where companies, you know, invented pre 2016, pre open AI are going to just accrue so much more value than the, than the companies after it.

• 46:25Speaker 02

  You know, I don't have a hard answer. I find myself somewhat unopinionated. I do think it's it's one of the great questions. Are you going to have the incumbents? I mean, Microsoft seems to be playing the AI game pretty well and actually have it, you know, bootstrap or like boost their growth. Are they going to be totally and utterly disrupted? I mean, I think another way of framing the same problem is, is OpenAI and ChatGPT and some of the other foundational models who currently own the consumer experience, are they going to own the consumer experience forever? Are people going to actually be building apps on top of a more API-ified version of, and it's like Alex said, it's intelligence too cheap to meter and it just infiltrates every product, you know, bots, whatever. I really don't know what the answer there is, to be honest. I think it's going to be, it's a, it's a standard yin yang thing of like, can Amazon become Walmart before Walmart

• 47:07Speaker 02

  can become Amazon thing, right? In which can the old become the new faster than you can eat the old? And that's, we're going to see that play out again.

• 47:15Speaker 00

  Yeah, and I've challenged you a little bit on mobile, how sustaining it, how sustaining it was versus disruptive it was. I think, I think maybe I'm being a Chinese dipper here. I think it's too, too soon to tell, because a lot of stuff couldn't happen without mobile in the current wave of destruction, of disruption that's come in with, with the chat, the chat face interfaces for AI. So we will see. The thing for me that's been the realization is when I got into this with 20 years ago and sort of, you know, Google was maybe the previous generation of company and then you had Apple and Microsoft and Yahoo and eBay were the most valuable companies on the internet.

• 47:56Speaker 00

  And you know what? Like, Google, Amazon, Microsoft, they made it through. Yahoo and eBay are much less relevant today. And the question is, are we the leadership team that does the Google, Apple, Microsoft, Amazon? Or are we the leadership team that does the eBay, Yahoo? Is a question that really does keep me up at night and makes me, makes me think about this, this moment of transition. We've gone through a lot, but there's another one coming. And now we're the little age company leadership, which is interesting.

• 48:24Erik Torenberg?

  I think Meta's recruiting strategy, from how I understand it as an outsider, is just absolutely brilliant. And I'm surprised more sort of, you know, big tech companies haven't done it sooner and have more just sort of antiquated sort of understanding of, of, of sort of aligning compensation to true value. Antonio, do you have any, any, any perspective on whether, or any thoughts on whether that's going to be the norm going forward?

• 48:49Speaker 02

  That's so above my pay grade, Eric.

• 48:52Speaker 02

  I wouldn't know what to tell you there. I mean, I mean, obviously, you know, I think one of Mark's, you know, famous saying is that we we judge you based on your outputs, not your inputs. Right. And that was like the party line even at Meta like in 2011. And I assume it hasn't changed.

• 49:05Speaker 00

  No.

• 49:06Speaker 02

  And I think every, every sane, you know, successful company I've worked at has had some version of that as a, as a, as a company principle.

• 49:14Speaker 00

  Yeah. I mean, I think the thing that, that certainly has been talked about in conversations I'm having is

• 49:21Speaker 00

  people who can successfully use AI as individual contributors in particular, it is a force multiplier. Like you give a 10x engineer access to AI, like they become a 100x engineer. And it is showing therefore the talent, the best talent, when they can use the tools that currently exist, becomes way more valuable. And then being able to produce the tool is way more valuable too. If that continues to play out, that will mean there will be a massive differentiator in how pay works out for the 10x, 100x engineers than there is today, which is already like they come

• 50:02Speaker 00

  in a BP level, they're paid significantly more than a typical IC, but clearly, clearly there's going to be more. The other alternative is these tools make it easier. Now, my, my perspective is the tools help you implement a lot of stuff that's been implemented before, but they don't solve logic for you. They don't solve creative thinking for you. And so they push things up the value stack in terms of where the work is happening, whether you're talking about marketing or I run data science and data engineering for the company. You talk about DS and DE, they take away routine, routine data pipeline coding, but they don't get you to the creative analysis that actually does something that changes the direction of the company.

• 50:43Speaker 00

  So I think there's a good chance that value accrues to the most creative, best logic thinkers. And value gets taken away from the routine work, which therefore would say compensation is going to go up for the best employees.

• 50:59Speaker 02

  Yeah. The best encapsulation effort is you're not going to lose your job to AI. You're going to use, you're going to lose your job to a human who uses AI better than you do. That's what you're going to lose your job to.

• 51:07Speaker 02

  I mean, it's still kind of a harsh lesson, but that's, that's the reality of it.

• 51:11Erik Torenberg?

  Do you think these companies in the future have fewer employees or more employees? Or what's the way to think about it?

• 51:19Speaker 00

  So this is a really interesting one. Like, I, two things. One, it depends how weird you think it's gonna get. Like, do we actually get to a S I A G I?

• 51:30Speaker 00

  Let's not, you know, AI 2027 is not the table.

• 51:35Speaker 00

  Look, this comes back to three different things, which we are talking about a long term. Thing number one for the framing is doing existing stuff more efficiently. Existing stuff is gonna be done more efficiently with AI as it exists today without more breakthroughs. Like, a lot of stuff can be automated, a lot of stuff can be done more efficiently. So that work, there will be less of it. Thing number two is, things that literally weren't possible before will become possible. That can open up a lot more jobs. And thing number three is, things that were possible but just were really effing expensive, suddenly become over the threshold of, you know, reasonable to do. That's the support example I used recently. So,

• 52:20Speaker 00

  thing number one, let's just, you know, rank cats better. We could already rank cats. We can now do it more efficiently. We can rank them better. Thing number two was semantic understanding of content to lead to content ranking to enable short-form video and unconnected content could not exist without these models. It now exists. Thing number three is, support chatbots. You could do it with human beings before. It was just prohibitively expensive. And now AI does a hell of a lot better than a phone, a phone trip. So there are those three different categories, and what matters is, jobs will go down in category number one. They work. How many jobs can't be number two and category number three create will determine, some of these companies may even get bigger, like literally, because so many things are now ROI positive to do.

• 53:05Speaker 00

  So this is just, I think it's still up in the air, but what is certain is all three of those things collide.

• 53:10Speaker 02

  Yeah, no, I think in the short term, you know, one of the biggest problems here is what's called a lump of labor fallacy, which you assume that does a certain fixed amount of work in the world. And it's not true for manual labor. It's not even true for now. It's certainly not true for knowledge labor. I think in the short term, you might have AI might have a negative impact on hiring ever so slightly in certain places, maybe interns or junior engineering. But overall, I think I agree with the latter. I think it was the last scenario that Alex set out, which is no, it actually increases jobs. I mean, take the example of stagecoach drivers and introduction of a car, right? It's like, yeah, the poor stagecoach driver who had a team of horses through, you know, New York, whose streets used to be full of horses with horsemen on the streets. I'm looking at New York, which is why I'm citing the analogy. Cars created

• 53:52Speaker 02

  hundreds, thousands of new roles that didn't exist in the past, from motel owners to gas stations, everything you can possibly imagine. So the thought that we're just going to sit around with 99% of people just getting paid UBI from OpenAI, like, I just, that's like the least likely scenario I see playing out, actually. Though, to be clear, I think there'll be disruption between here and that point. But I just don't see, no, I think there's going to be more demand, there's going to be more code, there's going to be more companies. Like, like Alex said, there's going to be things that weren't even physically possible, or possible, just so expensive, you couldn't possibly do them, which are suddenly feasible. And so, no, I think it's going to be actually a net positive on society. It's just, it's going to change. And change is scary. And, but that's, that's it. I mean, generally, kind

• 54:34Speaker 02

  of positive AI. I think if anything, like, it's funny, because you think, oh, there'll be this cognitive elite of AI builders. Like, dude, if you have intelligence on demand, does that make intelligence more or less valuable as like a thing that just exists in the world? It's like, no, no, the human touch EQ rather than IQ. And I hate those buzzwords, because they sound very LinkedIn. But you know what I mean, right? Like, actually, the raw intelligence is just going to be a thing you buy, just like a server on AWS. Like, that actually just matters a lot less than actually framing the problem or the human touch to it.

• 55:03Erik Torenberg?

  What if Snap had sold to Meta?

• 55:08Speaker 00

  I think Snap would be Instagram. I genuinely think we have the tools and the ability and the know-how to make things more successful. So I genuinely think it would be significantly more successful if it had. But, you know, that's a woulda, coulda, shoulda. And I'm sure they might not agree with me, but that's, that's what I believe.

• 55:29Erik Torenberg?

  Five plus years after the acquisition of, of, you know, well after the acquisition of Oculus, after sort of the big bet on, on the metaverse, help us conceptualize where we are, you know, five plus years later, and, and where we're going.

• 55:45Speaker 00

  Yeah, I mean, the thing I would do is we spent an awful lot of time and an awful lot of money on a video that was the mother of all demos of what the future looks like. And that video contains AI assistance, it contains audio glasses, it contains glasses like our Orion glasses we put out last year, which have sort of the visual overlay. And what's really interesting to me is despite everyone saying the metaverse is over or whatever, a lot of stuff in that video is coming true. And so my perspective on this is these are a smash hit. So the glasses, total smash hit. It's AI with you at any stage, you can ask it questions. I now ask it all the time to give me a reminder.

• 56:27Speaker 00

  I think of something and I say, hey, Meta AI, remind me in three hours of X, which I know is a minor use of it, but I don't have to take my phone out and send myself an email. And I'm finding myself just because it's with me all time, all day, I'm thinking of things to use it for. These have a lot, a lot of active users at this stage. And that is one of the things we feature in the video. So my perspective is, you know, where are we? It's a research and development space. Actually, amazingly, the AI jumped forward far faster than we thought, which means that it got ahead of the wave guides that are needed to do the overlaid augmented reality experience that we think is the end game.

• 57:11Speaker 00

  So yeah, that's going at the pace that we expected it to. The AI is going far faster than we expected it to. And I think you'll probably see 10 years from now, people walking around with these glasses with always on AI, with a visual overlay of the real world, helping them navigate their lives better.

• 57:28Erik Torenberg?

  What is your prediction of Antonio as a now an outsider of what the legacy will be of Meta's big bet, rebrand and sort of big financial bet on the category?

• 57:42Speaker 02

  I mean, this is definitely an outsider view because of all the companies we talked about, the VR thing didn't exist at all when I was at Facebook. So this is a completely uninformed view. It sold to me the big investment. I mean, and again, you know, Alex has a deeper knowledge here, but it seems to me that Mark's obsession was always mediating human social behavior in different forms. And how many touchpoints could that possibly take? I think Facebook played its cards with the mobile revolution very well. And I guess the thinking there was like, well, VR is the next way that humans are going to interact with each other. And so we're gonna, we're gonna build a social grass around that. It hasn't quite played out that way. I'm not quite sure why, because I used to be an AI skeptic as well, because I've, for the past 20 years in tech, it's like AI is the constant technology of the future and it seems like the future is finally here. But similarly, virtual reality has always been this like perpetual technology of the future.

• 58:28Speaker 02

  And somehow that ship hasn't really quite come in yet. That said, over the longer term, I suspect something like the VR headsets that I think Alex modeled for us a second ago, probably will be in production and at scale. But you know, sometimes things are just too early. I think people, if you actually read technological history, you realize that technology has developed a lot longer than you think they did, right? There was 20 years between the transistor and the integrated circuit. Another 20 years between the integrated circuit and the PC. Another 20 years between the PC and the internet and like 10 years between the internet and mobile. So these are like decade long things that play out. So it might just be too early, I think, in the case of VR.

• 59:05Erik Torenberg?

  Is there any world which Facebook, Amazon, Microsoft could have made an actual competitor to the iPhone? Or was it just too complex and a better try for the next platform, VR?

• 59:20Speaker 00

  Maybe if everyone had spotted it early enough, but everyone said it was the year of mobile for years and then, then stopped saying it. And then suddenly it actually became the year of mobile and we all missed it. Because it actually happened around 2010. And so I think the issue is everyone missed it. Yeah, maybe, maybe, but I think probably not.

• 59:39Erik Torenberg?

  Gearing towards closing here, the book is Click Here. It's a history of the industry, it's a history of the company. Meta, Facebook, obviously, it's one of the most, if not the most iconic and important company of the last 20 years, not just its technology, but globally. And it's had a lot of different evolutions and phases in its substance and its perception. What do you think is not fully appreciated about where the company is now? Or what's a misperception or misunderstanding that's worth correcting?

• 1:00:08Speaker 00

  You know, I'm going to come back to what I said at the beginning, but this is obviously my personal take on it. I'm sure if you're stock, he'd have a different, he'd have a different opinion. The biggest thing is that advertising is good. And I think there's this whole, this whole thing about like, you know, surveillance capitalism, ads are bad, you know? And it's all protecting people from slightly better ads for personalized trainers. And I actually think like the way that it gives people access to things that they like, like people prefer personalized ads to non-personalized ads. We have tested that again and again and again. If we just personalize the ad based on your demographic, you will not like what you get. But if it's based on what you truly like, you'll get a better experience.

• 1:00:49Speaker 00

  And I'm still struggling to see what European customers have been protected from, that US customers are suffering from, when it comes down to a lot of this privacy regulation and a lot of this hatred of ads. And so the number one perfection I want people to shift is ads are a good thing. People like personalized ads. And, you know, yes, the book has a bit of history in it of the company. It has a lot of context on each of the channels. And I think it is really important that we as an industry care about advertising and learn how to do it well. A lot of startup founders just still believe in like, oh, they clicked on my ad and they converted, therefore the ad drove that conversion. There's some really obvious stuff that people can learn.

• 1:01:30Speaker 00

  And this book is, I think, at a good level to get people to know how to do the basics really, really, really well.

• 1:01:37Erik Torenberg?

  The book is Click Here: The Art and Science of Digital Marketing and Advertising. Alex, it's been a fantastic conversation. Thanks for coming on.

• 1:01:44Speaker 00

  Thank you very much, and thanks, Antoni.

• 1:01:48Speaker 01

  Thanks for listening to this episode of the A16Z Podcast. If you liked this episode, be sure to like, comment, subscribe, leave us a rating or review, and share it with your friends and family. For more episodes, go to YouTube, Apple Podcasts, and Spotify. Follow us on X at A16Z and subscribe to our Substack at a16z.substack.com. Thanks again for listening, and I'll see you in the next episode. As a reminder, the content here is for informational purposes only, should not be taken as legal, business, tax, or investment advice, or be used to evaluate any investment or security, and is not directed at any investors or potential investors in any A16Z fund. Please note that A16Z and its affiliates may also maintain investments in the companies discussed in this podcast.

• 1:02:32Speaker 01

  For more details, including a link to our investments, please see a16z.com/disclosures.