--:-- Host: I think Visa will enter, MasterCard will enter, PayPal will enter. Fidelity has already said they're going to enter. Bank of America said they're going to enter. I think you're going to have every bank probably issuing, I hope, a stablecoin the way you have them issuing credit cards. These all have users and customers. The banks will have a button that says "send a stablecoin." So what I'm hoping is that there's enough legitimate actors around this who create a network effect.
--:-- [Intro music fades in and out]
--:-- Host: Today on the A16Z podcast, Chris Dixon, general partner at A16Z, joins economist Tyler Cowen to discuss the ideas behind his book, *Read, Write, Own: Building the Next Era of the Internet*. They explore the evolution of web infrastructure, the economic consolidation of today's internet platforms, and how technologies like blockchain, open source systems, and AI could reshape the future. The conversation spans everything from creator economics and stablecoins to the decline of Stack Overflow, and whether AI will centralize power or help redistribute it. Let's get into it.
--:-- Host: 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. For more details, including a link to our investments, please see A16Z.com forward slash disclosures.
--:-- Tyler Cowen: Hello, everyone, and welcome back to Conversations with Tyler. Today, I am chatting with Chris Dixon, who is a general partner at Andreessen Horowitz. He has a longstanding history in the tech field. He started off, in a way, as a blogger, and we now have out the paperback edition of Chris's recent book, a very stimulating and provocative read. It is called *Read, Write, Own: Building the Next Era of the Internet*. Chris, welcome.
--:-- Chris Dixon: Thanks for having me.
--:-- Tyler Cowen: Let me ask you a very fundamental question so people can see where you're coming from. If I listen to music today, not on my stereo, I'll go to YouTube and Spotify. What's wrong with that arrangement? How is it you think we can make it better?
--:-- Chris Dixon: The core thesis of my book and the core thesis of my career now is that the internet began as a decentralized network. What that meant was that if you created a website, or let's say you were a musician and you created a website and you sold your music, you would sell directly to the consumers. There would be no intermediary in between taking money from that transaction.
--:-- The challenge with things like Spotify and YouTube, and generally the structure of the modern internet, is that you've had these services pop up which are very dominant. It's very consolidated. Roughly 90% of internet traffic runs through less than 10 services and companies. And they have very high what we call "take rates." A take rate is the percentage of money flowing through the system taken by the network intermediary.
--:-- YouTube actually is the most generous of the social networks. They give roughly 50% to the creators and take 50% for themselves. That's actually a low take rate on the internet. In any other area of the economy, as you'd know, 50% for an intermediary is generally a very high rate. Spotify is 30%. But then of course, musicians have many other layers of fees on top, including the music labels. Spotify's own statistics—it's on their website, I guess they're glad about this or they think it's a good stat—is that something like 8,000 of the 8 million artists make more than $50,000 a year. So a very, very small percentage make something like the average American living.
--:-- The question is, is that because people aren't paying for music? People are paying for music. Advertisers are spending a ton of money on YouTube. Meta and Google and these companies are making a ton of money. The problem, in my mind, is primarily economics. I go through this in detail in the first part of my book—the history of the internet and how this happened.
--:-- The reason I got into the internet was this very exciting vision. It came out of academia and government, but the vision was that you'd have an internet owned and operated by the people that use it. If you visualize a network, the money is flowing to the edges of the network. There were all these great ideas, like a famous blog post by Kevin Kelly called "A Thousand True Fans." The idea was that because you're removing intermediaries, musicians and creative people can now make a living with only a thousand customers. If someone's paying $10 a month and you have a thousand of them, that's $10,000 a month, $120,000 a year. That's a pretty good living for somebody doing something they love.
--:-- That was always the vision. That's what got me excited. I started in the internet in the 90s, and the internet was like that in the 90s and much of the 2000s. Then, because of the incentives of the way these services were set up, because of venture capital, because of a bunch of things, you had incentives to consolidate. Network effects, of course, make these services more valuable the more people are on them, which has a winner-take-all effect. About 10 years ago, we ended up in a place where there's roughly five to ten services dominating the internet, taking all the money, taking all the economics, and also having control. That's a whole separate topic—deplatforming and rules, demonetization on YouTube.
--:-- In the book, I go through the history of how this happened and the effects. There are economic effects and governance or control effects. More broadly, it creates real challenges for creative people. Now throw in artificial intelligence, which is obviously an amazing technology and incredibly powerful. I'm generally very excited about what it will do. But left unchecked on its current trajectory, it will likely lead to even further consolidation and reward companies with large amounts of data and capital. The reason I'm involved with blockchains and crypto is I see blockchains as a potential counterbalancing force to those consolidation trends.
--:-- Tyler Cowen: But there is a reason why the older internet dwindled, right? For me as a listener, Google, Alphabet—they've upgraded YouTube in some very significant ways. The search function is amazing. The algorithm to me is useful. It used to be you had to download a YouTube video and then you'd watch it much later. Now it's all seamless and great. Why aren't I just better off in this new world? I left the decentralized internet behind. My music listening today is much better than it was 20 years ago.
--:-- Chris Dixon: My argument structurally is kind of a thesis, antithesis, synthesis—and that's *Read, Write, Own*. The first era of the internet were what you're referring to as protocol networks. That was the World Wide Web and email. These were decentralized networks with no company behind them, not owned by anyone. They're really just standards, sort of like a language as a standard.
--:-- The case study I use in the book is RSS and the fall of RSS. RSS is still around, but in a very niche way. At one point, let's call it 2008, I was there blogging about this, it was a legitimate contender. You could have imagined a world where you were getting your YouTube and Twitter and all these other services through a decentralized protocol—not RSS as you imagine it, like a bad visual interface, but as the underlying structure, and then you build graphics and streaming and all these other things on top.
--:-- So why didn't that happen? To your point, it wasn't good enough. It didn't offer the same user experience. You're absolutely right about that. These services won because they were better. One of the big reasons they were better was subsidization. YouTube for a very long time, and to this day in some areas, is subsidizing video hosting costs, which is very expensive.
--:-- I've seen this from the other side. I work in venture capital, I've been investing for almost 20 years. These companies like YouTube and Twitter raise a lot of money. What do they spend that money on? A lot of it is subsidizing the hosting costs. The basic idea is these are winner-take-all markets with network effects. So you subsidize along the way.
--:-- Imagine YouTube came out in 2005. You're sitting there in 2007. You're a video blogger and you have two choices. You can set up RSS and pay hosting costs, or you can go to YouTube and do it for free. That was one of the big value propositions of YouTube early on. If you used it back then, I did. That subsidization is something those protocol networks can't offer. There's no company behind them. RSS couldn't offer subsidies.
--:-- Tyler Cowen: But isn't there some a priori reason to expect the centralized service to spend more innovating and innovate more rapidly than the decentralized service, precisely because there is some monopoly profit there?
--:-- Chris Dixon: I think there are a couple of things. If I could get to blockchains briefly—I think of blockchains as a new architecture for building internet services where there is no intermediary. "Decentralized" is another way of saying no intermediary, where you have very low take rates. The way I think of blockchains is they offer the benefits of protocol networks, the old networks, the societal benefits—money flows to the edges, control is by the community. But blockchains are able to do some of the things you're describing.
--:-- A blockchain can offer incentives. A blockchain can have a treasury. This is the architecture of a blockchain. They have something called smart contracts. A smart contract is something that's hard for people to get their heads around, but it's not a company or a person. It's literally a piece of code that owns tokens and money and can use that money to do things like the YouTube subsidization.
--:-- My core argument is that protocol networks are better for society. Corporate networks, as I call them, like YouTube, have a lot of advantages, as you're describing—competitive advantages. Blockchain networks done right can ideally be the best of both worlds. They can have the societal benefits of protocol networks, but the competitive advantages of these corporate networks.
--:-- Now, to your point, anything that's decentralized has some coordination costs. A good case study there would be Linux versus Windows. Linux is now, I don't know what the percentages are, but I think it's well above 90% of operating systems. If you rewind to the 90s when it was starting out, it was kind of a mess. It was on a listserv with a bunch of uncoordinated people. You did have a very important kind of leader, a bully pulpit leader in Linus, but it was sort of an uncoordinated mess. You're right, there is overhead and coordination costs. You don't have a hierarchical management structure that can be very efficient.
--:-- The flip side, and why I think Linux won, is that open source benefits from what's known as composability. That's the idea that anyone in the world can write a piece of software once, put it on GitHub as a public resource, and then anyone else can use that as a Lego brick to build another thing. I say in the book that composability is to software what compounding interest is to finance. It's this thing where you just keep building—one person builds something and you never have to build it again, you reuse it and build the Lego bricks up.
--:-- I think that's one of the main reasons open source has taken 90% plus market share—AWS, every embedded device you have, every data center, Android devices, and so forth. So you're right, there's a different set of tradeoffs. It's tougher to coordinate building a decentralized protocol or a blockchain protocol. On the flip side, you have other benefits: lower take rates, the ability to offer incentives through tokens, and composability. Everything is open source and reusable.
--:-- Tyler Cowen: But there is legal free entry in the sector. People can go to Andreessen Horowitz, they can go to you personally and ask for VC money to do something. I recall a bunch of people told me I should try Mastodon, which was decentralized. I did. I didn't like it. Most people didn't like it. That was a bunch of years ago. No one has come to me in at least five years and said, "Oh, Tyler, you need to try this new decentralized service." Why isn't that happening? What's the benefit the service might be offering me that's concrete?
--:-- Chris Dixon: I have a section on Mastodon. I think there are architectural flaws in the whole thing. For example, if you've used it, there's this whole concept of servers and you have this very challenging problem of coordinating across servers. It's very fragmented. That might be a separate topic.
--:-- Look, when I think about where blockchains are getting adopted, if you look at where they're successful right now, it's mostly in financial applications. Stablecoins, for example. The numbers are shockingly high. If you Google "Visa stablecoin dashboard," Visa is, I think, a pretty neutral party. They have a dashboard that tracks it. It's three and a half trillion dollars in stablecoin transactions last month. That's been steadily going up.
--:-- A stablecoin, just for your audience, is something like a dollar or euro, but it's backed by an asset in the bank. There's Tether, USDC. There's a bill that just made it past the Senate Finance Committee and is going to be voted on in the House hopefully in the next six weeks and hopefully passed. It's congressional legislation to put full rules around stablecoins. All of this growth has been happening before there was that kind of clarity, and I think that could really accelerate it.
--:-- The reason I'm saying this is that I think there are really interesting things you can do with blockchains and social networks. But the reality in 2025 could just be that we have social networks, the network effects are strong, and the areas to focus on when you have new architectures like blockchains may be other areas. In software, there's a phrase "greenfield, brownfield." Greenfield is you have some breakthrough AI or crypto or whatever—do you go after existing use cases and make them better, or do you go after new use cases? What you're describing is more brownfield. It may just be that some of the wars are fought and they're over. There are three billion people using Facebook apps. At some point, the network effects are so strong that even if you come up with something much better, it may just be a challenge. I'm sure some of the challenge with something like Mastodon is just that your friends aren't there. Network effects.
--:-- I'm very excited about blockchains and this new architecture. We make investments in a lot of different areas because we like to experiment and try a bunch of things. But right now, the most likely areas where we're going to see increased adoption are in areas where things are more broken. I would say that's in payments and financial services. And then I think there's a lot of interesting stuff happening at the intersection of crypto, blockchains, and AI.
--:-- Tyler Cowen: With stablecoins, once the Trump people are no longer in office—it's four years from now—what do you think the regulatory counterreaction will be like? Stablecoin issuers are subject to runs, right? There's nominally 100% reserves, but there's no guarantee.
--:-- Chris Dixon: There are 100% reserves in the legislation, and USDC has 100% reserves. Tether is debated because they're not audited in the U.S., and people don't know. They say they do. But the bill would actually require 100% reserves.
--:-- Tyler Cowen: But why isn't the equilibrium that I have my stablecoins overseas where they're not regulated, and de facto the sector is not 100% reserves and there are a lot of runs? Why isn't that what ends up happening?
--:-- Chris Dixon: What we're hoping to do, and in the bill, there'll be all sorts of things. For example, if you go to Coinbase, there'll be restrictions on offering unregistered stablecoins.
--:-- Tyler Cowen: But that's a U.S.-based firm. The lowest bidder will be foreign firms—Estonia, Argentina, wherever—that will offer me a better deal, precisely because they're not 100% backed. It'll be a higher return. I won't care about the social risk I generate by putting my funds into a more run-prone firm. Why don't we have to worry about that?
--:-- Chris Dixon: What my hope is, and what often happens, is—first of all, I'm not an expert on every aspect of the proposed stablecoin bill, but there are reciprocation kind of requirements. If other countries want to access them and participate in our financial services and KYC and AML, they're expected to have similar rules. Historically, our financial services regulations tend to propagate to a lot of other countries. My hope would be that the U.S. does something here and a lot of other countries follow that lead, and that becomes the norm. There'll be rules like you can't call something a stablecoin, you can't get it on a registered exchange if it's not. That would be my hope.
--:-- Look, in crypto, there's no question there are scams and there are bad things. I'm sure there will be in the future. I think the best solution is to have smart regulation that incentivizes or requires things to be built in the proper way, in a way that avoids things like bank runs. By the way, the bank run thing—just to be clear—there were a bunch of stablecoins in the past, like Terra Luna, which had a bank run and collapsed. That was not asset-backed by dollars. That was a sort of circular thing backed by its own token. That has happened. The hope would be that these existing ones and the new ones and the regulations around them would require full one-to-one asset backing and therefore make bank runs impossible.
--:-- Tyler Cowen: But circa 2025, we've seen a lot of international cooperation break down. The WTO basically doesn't work. The UN has not worked in some while. Intellectual property law still works. Agreements on trying to end various wars—those are not working. It seems unlikely to me that the U.S. could convince the world, in essence, to copy our stablecoin regulation. Even if we tried to use SWIFT, we've pushed the SWIFT incentive on countries very hard. It's shown its limits. Again, why don't we just end up with these significant corners where the most profitable stablecoins are outside of what the U.S. wants to do with them?
--:-- Chris Dixon: I think it's partly also what do you want to do with the stablecoin? Who's using the stablecoin? I'll give you an example. The Stripe founders—I think of Stripe as a very smart financial services firm. They had done crypto like 10 years ago, and then they really soured on it. I think you know Patrick and John pretty well. They had actually really soured on it. When I saw them, I would say, "Oh, crypto's not..." I just watched them on the All In podcast a few weeks ago. They acquired this company called Bridge, the stablecoin company. In their annual letter they just put out, they called stablecoins "the room temperature superconductor," which is a holy grail kind of thing.
--:-- I don't have their financial statements, but as they describe it, they're using it for things like treasury management. The example they gave was SpaceX moving money from one jurisdiction to another. A very popular use case is international invoicing. You're an importer and you have to send out 50 invoices to various countries. You can now do it in a fully digital way with very low fees and very quickly.
--:-- One of the interesting things the Stripe founders said is it's not just the lower fees. To give you a sense on the fees, as of a year and a half ago, the infrastructure in crypto has gotten better. You now basically on things like Base, which is an L2, and Solana, have hit what we thought of as a long-term target: one second, one penny to transfer things. That's where we are now, technically. Your viewers, if they want to check me, can go download the Coinbase wallet and try it. You can see it.
--:-- One of the big benefits that Stripe founders talked about is that because it's fully digital end-to-end, sort of like email, you can fully automate the whole thing. A big problem with invoicing is invoice fraud. People send you an invoice and it's a fake place to wire it to. Now it's fully digital. Stripe has effectively what's like a reputation network. One computer sends a request to the other one, checks it against Stripe's database—is this a whitelisted address?—and does the whole thing end-to-end with low fees, internationally.
--:-- To your question, sure, will there be people at the fringes who want to maximize yield? Sure. There's always that kind of behavior. It's the internet. The internet has edges. The question is, can we marginalize it? Will these legitimate companies like Stripe—and look, what I'm hoping, stablecoin bill passes—I think Visa will enter, MasterCard will enter, PayPal will enter. Fidelity has already said they're going to enter. Bank of America said they're going to enter. I think you're going to have every bank probably issuing, I hope, a stablecoin the way you have them issuing credit cards. These all have users and customers. The banks will have a button that says "send a stablecoin." So what I'm hoping is that there's enough legitimate actors around this who create a network effect that, to your point, yes, there will be that stuff, but it will be marginalized.
--:-- Tyler Cowen: In that world, should we infer that the Federal Reserve loses control of the money supply? Create a stablecoin. It's backed by a T-bill. In a funny way, it's like a private open market operation. I'm fine with that. I'm not sure the Fed controls the money supply today. But does that become a macro issue?
--:-- Chris Dixon: I feel like I'm talking to a famous economist. I'm on your territory now. It's dangerous because I'm not an economist. But I haven't figured this out myself either, to be clear. I'm genuinely asking various people. I asked Austin Goolsbee the same question because I don't know.
--:-- Tyler Cowen: [Laughs]
--:-- Chris Dixon: I believe the last stat I saw is stablecoins hold 4% of treasuries. If this grows a lot, it will be meaningful. It already is kind of meaningful. I think of it more in terms of payments and all the adjacent things you can do around payments versus a bank account, and sort of replacing folks like banks and the Federal Reserve that control the money supply. That's at least where I think the focus should be.
--:-- But in the current draft of the bill, you can't offer yield to the consumers. I think that's something the banks are fighting for. This is still a political thing going on. They don't want that because they think if you can offer essentially a treasury bill yield, and the consumer gets the yield, that will become more attractive than bank accounts. So this will also be a question: is this optimized more around payments or more around savings use cases?
--:-- I would also argue on the other kind of national security interest here. Others have written about this in various editorials. It's another way to popularize the dollar. It's a way—we've already seen the demand for it. If we legitimate it, it presumably will make the dollar more popular and increase its status or maintain its status as a reserve currency. Fractional banking stuff, I probably have to defer to you on that. I can speculate a little bit, but it's from an amateur perspective.
--:-- Tyler Cowen: When the AIs trade with each other, do you think they'll prefer Bitcoin or stablecoins?
--:-- Chris Dixon: That's a good question. We actually just made an investment in a project that's doing agent sort of stablecoin infrastructure for AI agents. We haven't officially announced it yet, but I guess I can talk a little bit. The idea is that a lot of things with AI—there's the technical side, but there's also the legal side. How do you deal with KYC, AML, and liability and all those kinds of questions that come up if you have an AI going out and doing stuff and spending money? That turns out to be a really interesting area of innovation. The person we just invested in is a veteran of that world of the regulatory money world.
--:-- Tyler Cowen: To your point, one of the knocks on Bitcoin as a payment system is the volatility. You don't want to have something that changes value a lot when you're paying for something. But if it's a machine, presumably they can do that in milliseconds or microseconds, and volatility becomes a mood issue.
--:-- Chris Dixon: There's a guy, David Marcus, who ran the Libra project at Facebook and spun out. We funded his startup. It's called LightSpark. It's doing exactly what you're describing—a payment system built on top of Bitcoin. His argument is Bitcoin is preferable because it's not the dollar that a lot of countries will want. It's the only crypto asset that is credibly globally neutral, that no country sees as a U.S. thing or some other country thing. Therefore, it should be the currency of the internet. I think that's an interesting argument.
--:-- Tyler Cowen: I wonder sometimes if the AIs won't create their own crypto tokens and say, "Well, the seniorage from Bitcoin has gone to early humans. We can design one better because they're already smart, but they're going to be really smart." We'll have to learn how to trade their stuff. People have done experiments like that. Once they have emergent capabilities, I guess it all depends on what we allow them to do and what systems we plug them into. But if there's anything where there's a clear cookbook and you can do it digitally online, no one built crypto for the AIs. But if you had wanted to build a money for the AIs, I'm not sure you could have done much better than crypto.
--:-- Chris Dixon: I think that's right.
--:-- Tyler Cowen: How will AI change how the web is organized? Will it just be totally different in five years?
--:-- Chris Dixon: That's a great question. A couple of things I'll say on that. And all of this, obviously, in the context of our firm doing a lot of AI investment. I think it's generally a great thing. The technology is amazing and impressive. Every day you see new amazing things. There are these debates around whether it's going to slow down. My guess is it's not going to slow down. There are so many smart people and so much money and so many different ways in which it can improve. It was pre-training and now it looks like it's reasoning. They'll probably come up with some other method in a year that does that. We haven't even begun to scratch the surface around humanoid robots and haven't gone deep on video and other forms of media modalities.
--:-- I think one thing I think a lot about, though, is the economics of the Internet. I have a chapter in the book on this. I think of the Internet as having an implicit covenant right now between these five to ten companies—Facebook, Google, Amazon, et cetera—who control 90 percent plus of the traffic and the money. Just to take Google as an example, there's the rest of the Internet, the long tail: my blog, your blog, a cooking site, a travel site, whatever it might be.
--:-- For 25 years or so, there's been this implicit covenant between the distribution centers like Google and the content providers like us. That covenant is: you let us fair use, you let us take a snippet of your content, index it, put it in our search results, or let people share it on our social network. The trade is we'll put a link there and you'll get some traffic back, and then you can have ads or subscriptions or make it free or whatever you choose to do. That's been the equilibrium state that evolved. It wasn't like people sat down and decided that, but it evolved. And it's worked reasonably well. It's worked more to the favor of the distribution side, to the Googles of the world. You see that in Rupert Murdoch and Fox suing Google over it, and Facebook just removed news in Canada because they're upset about it. So it's not that happy, but it sort of worked okay.
--:-- What happens in a world where you just get the answer and there's no need for a link? Which is kind of the world we're in now. I'm probably using Claude and ChatGPT and all these things more than I'm using Google.
--:-- Tyler Cowen: And I read fewer books also. It's not just media sites. Why read a book when you can ask for a 10-page printed report on the history of Edward III?
--:-- Chris Dixon: Yeah, exactly. Books are a whole other thing, but the end of the rest of the internet—if I'm not going to click, why am I going to click through? Look at Stack Overflow. I actually used to be on the board of Stack Overflow before they got acquired, and I love the service. It's a programmer Q&A site. The traffic is down, I don't know, 80 percent now, because of these new services. What basically happened is the AIs went out and trained on Stack Overflow and GitHub, and then created these amazing services like Copilot and Cursor, which are amazing. I use these on the weekends. I use Cursor. It's an unbelievable product. You can type an English sentence and get code out of it. But it obviates the need to go to these websites. The traffic is down.
--:-- My concern is Stack Overflow is the canary in the coal mine. This is the first thing to go. What are we, in five years, ending up with the rest of the internet atrophying and we have five to ten services? I love the AI stuff. It's better. I use it. But I worry that we're going to look around in five years and realize we just recreated a structure that looks like broadcast TV in the 1970s. You have four channels. The beauty of the web—the serendipity, the diversity, the ability for somebody to just spin up a blog—this is how a lot of my career got started. You've been blogging for 20 years or something every day. Is that ability to do that going to be lost because there's now no way to get discovered anymore, since links aren't really needed?
--:-- On the one hand, it's great for people. I'm by no means anti-AI or technology. But I do think we need to think about—and I'm a little bit surprised more people aren't thinking about—what kind of internet do we want? Are we creating the right incentive systems to maintain that? Look at the new ChatGPT 4.0 image thing that came out, which looks amazing. It's gorgeous. There's going to be more great stuff like that. But what's our plan for graphic designers? Maybe it'll be emergent and new jobs will pop up. In two years, we're going to have the ability to make a Hollywood movie with one person. What's our plan for all of these other jobs, parts of the Internet, parts of the—you mentioned books. It's easier to get that thing out on Henry III. On the other hand, do we want a shrinking book industry? I think it's already shrinking too much, in my opinion. Do we want that to go further? I don't know.
--:-- Tyler Cowen: Let's say a media company calls you in—somewhere like The Atlantic, The Economist, The New Yorker. I'm sure you know these products. They say, "This is happening. We'd like some commercial advice from someone who knows the sector." What would you tell them to do?
--:-- Chris Dixon: It's hard because I think it's a systems problem. It's a tech problem and a systems problem. I'm not sure that a single node in the system like The Atlantic has a lot to do. It's really a structural issue.
--:-- Tyler Cowen: But does personality-driven content survive? Is that the innovation rather than so-called facts?
--:-- Chris Dixon: One very common pattern with technology is death of the middle, barbelling. To give your listeners context: before the Internet, you had a lot of midsize retailers like JCPenney and Sears. The Internet pops up and you get this barbelling effect where you either want to be really, really big like Amazon and Walmart to take advantage of scale, or you want to be a boutique, high-touch, high-brand-value Gucci, Hermes, and so forth. The two biggest winners of this era from a stock perspective were Amazon and LVMH, which is a private equity style roll-up of those high-end brands. Who lost? The people in the middle—Kmart, Sears, all those companies that went bankrupt.
--:-- Why does that happen? Because technology tends to unbundle. Kmart and Sears were this artificial bundle of distribution and merchandising. You could bring stuff to them, but not all the way to them the way Amazon does. Once you had this new tech, you could unbundle it and go all scale or all high touch.
--:-- I think the same thing has happened in media. We've had a barbelling effect. The winners have been either 30-second dopamine hits of TikTok and Instagram Reels, or three-hour podcasts. Or high-end, long-form content like *Severance*. What suffered has been the 30-minute sitcom, the middle—the 30-minute game show. That matches human needs. You're in line to buy something and you want a couple dopamine hits, you watch TikTok. Then you want to lean back and get something, you want a three-hour podcast. I think that surprised a lot of people, how popular three-hour podcasts are—Joe Rogan, Lex Fridman. Those 30-minute things were there artificially because of the constraints of the medium of TV.
--:-- So to your question, my default answer for media would be: you're going to have a barbelling effect. You're going to have highly automated, scaled-out AI-driven content. Somebody comes up with a new *Star Wars* thing and you have a whole Reddit community building AI-created serial movies. On the flip side, you have rising demand for very bespoke human things that could be live. We've seen live concerts become more popular in an age of machine-created music. Fine art, The Sphere. Books. People say this all the time: chess is now more popular than ever, even though AI is better than humans at chess. In the end, we're monkeys who like other monkeys. There is this fundamental human nature that doesn't change. Machines can do very powerful things and can be useful. That would be my naive default assumption.
--:-- The other really interesting thing with media that I think about: when film came along, the first cameras and photography, there were these famous essays—"Art in the Age of Mechanical Reproduction," Benjamin. There was all this discussion about the future of art now that we can take a picture. What's the role of the representational artist? The rise of photography coincided with art becoming abstract and all those modern art movements. The camera did replace representational art. But another thing happened: a new medium was formed. You take a photo, you take a series of photos, and you make a film. You had two things happen when the camera developed. The old medium got replaced, and high art pivoted into abstract art. But then you had a brand new medium that simply couldn't exist before—film—and a whole industry created around it. So when I think about media, maybe AI will be like the camera. It will replace illustration; the role of the illustrator will go away. But on the flip side, maybe it will enable a new, AI-native form of media that couldn't have existed before.
--:-- Tyler Cowen: What will AI competition for a16z look like? You have an AI competitor. What are they like?
--:-- Chris Dixon: Well, it's a good point. What is venture capital? Venture capital today is a bundle of services. There's the choosing what to invest in—the picking aspect. In that picking aspect, I think AI could do a very, very good job and probably beat humans in a lot of ways in the near future. There's another aspect, and I think this is one of the smart things the founders of the firm, Ben and Marc, did: there's a very human aspect to what we do. There's a high-touch human aspect. A lot of how we succeed is we create a culture and a group of people where entrepreneurs decide they want to work with us. We spend a lot of our fees on a set of people that help them.
--:-- Tyler Cowen: But can't AIs do that? They're great therapists.
--:-- Chris Dixon: They can, but at some point you always need—for some foreseeable future—someone to help you build and design the robots. Think of us as playing the role of advisors to the people that are building the robots. Hopefully there will be some role for that, at least for some period of time.
--:-- Tyler Cowen: If an AI applied for funding, let's say it was legal in terms of the bank transfers, would you consider the proposal?
--:-- Chris Dixon: I think right now they're limited. They can't open a bank account. They can't send money. But yeah, I think it's super interesting.
--:-- Tyler Cowen: They'll do it with crypto, right? Maybe this is in two years from now. You'll get a proposal from an AI and you'll get the AI on the phone, on the Zoom call. I don't even know how you would do it. Maybe you just refer it to your AI.
--:-- Chris Dixon: We've been talking about it semi-seriously. We haven't done anything yet. Mark and Teresa and I were talking about whether there should be a nonprofit built on a blockchain. Were you involved with the FastGrants thing?
--:-- Chris Dixon: I thought that was really interesting and inspired by that. Imagine FastGrants, but it's a blockchain version that uses AI to give out the grants as an example. That's actually something we were just brainstorming. It's not a serious idea, but that type of thing. You could call it FasterGrants.
--:-- Tyler Cowen: Right, yeah.
--:-- Chris Dixon: I actually wrote out a business plan—I haven't done it—but it was like you have a DAO, an entity on a blockchain, an autonomous entity. Think of an endowment. When I first raised venture capital, I walked into the Ford Foundation and thought, "Aren't they this organization that gives away money?" Well, they have this other side that invests money. Endowments are two things: invest the money and give away the money. They're very different. The idea was: could we create a blockchain autonomous AI entity that is like an endowment? Half of it is investing in tokens and this and that, and the other half is giving the money away a la FastGrants. I think it's pretty cool. If someone did it, I'd be excited to help out. I haven't gotten around to it. But I do think that might be a future thing. What's cool about an endowment is if it invests well, it can just snowball and have more and more money. Then you use some portion of the returns every year to give that out. Hopefully it becomes this self-sustaining thing, the way university endowments have.
--:-- Tyler Cowen: If I ask you, how will AI change politics? What's the most confident prediction you have?
--:-- Chris Dixon: That's a good question. A couple of things. One, I think there's going to be a lot of political drama around AI in the next couple of years. This is not AI changing politics—it's politics potentially changing AI. I think there are going to be issues around copyright, safety. I believe these are not going to be settled in the court. They will be settled in Congress. They are too important. I just went through this with crypto—we had four years of regulatory struggles in the courts, and now it's going to probably be decided in Congress. I think ultimately the same thing will happen in AI because it's just too important.
--:-- Another thing: I wonder how many problems in the world are intelligence problems versus political or coordination or regulatory problems. Building housing is a hot topic right now. Does more intelligence help us build better housing? Probably the opposite, because the smarter people lobby harder to stop it. Housing is more of a regulatory or collective action problem.
--:-- Tyler Cowen: Sure, yeah.
--:-- Chris Dixon: I think it's a collective action problem. Does AI help with that? Maybe, I don't know. I think AI can do a lot of really interesting things. But I do think a lot of the most challenging things we have in the world right now—the gating factors to productivity growth, economic growth, and human well-being—are collective action problems, coordination problems, getting people to agree on something.
--:-- Look at the Internet. How did the Internet change politics? It changed politics by changing the way information flows. I think we're only beginning to see the effects of that. Technology always has first-order and second-order effects. The first-order effect of the automobile was you can get from point A to point B faster. The second-order effect was suburbs, trucking, the highway system—the next 50 years. The first order was "go somewhere faster" from 1900 to 1930. Then the next era was all the second-order things.
--:-- So the Internet: the first-order thing was social media. "I can tell people I had a burrito for lunch." The second order is you suddenly have a whole different political dynamic. You can have an insurgent political movement like Trump or Bernie Sanders that bucks the system and counters the establishment. You can have a whole narrative universe that counters—you have COVID and the establishment's world, and then you have the podcasting version of it. It restructured how information flows. People will debate whether that's a good or bad thing, but it's clearly had that effect. We're in the very beginning of that. Social media didn't become a mainstream thing until smartphones—mainstream in the sense of three billion people, probably until 2012 or so. So we're still relatively early in that development.
--:-- Now AI: a big question is, for example, it's very possible in five years that people get all their news and political information from an AI. Who created that AI? Was it created by someone with a political agenda? Was it open source? If I can pick one issue that will make the AI future better, my own belief is that we have very strong open source AI so that people can have a choice, it can be audited, it can be open. Our kids will be raised by AI. The news will be formed by AI. Politics is downstream of culture and information flows, and AI will reshape that. Who controls that reshaping seems like the key question.
--:-- Tyler Cowen: As you know, not many countries have serious AI companies, and even those in Europe may or may not last. They're not obviously mega profitable. Let's say you're the government of Peru. You can turn over your education system to some foreign, maybe American AIs. You can turn over how your treasury is managed to the AIs. You can turn over your national defense to the AIs. None of these are Peruvian companies, most likely. In the final analysis, are we even left with the government of Peru? Or has it in some sense been pseudo-privatized to the companies that are running the structures and indeed to the AI itself?
--:-- Chris Dixon: This goes to open source. I think the answer to your question—
--:-- Tyler Cowen: But even open source is managed by someone. A version of DeepSeek is embedded in Perplexity. That's worked great. It's still someone's company.
--:-- Chris Dixon: I think it's different if they can get the open weights of DeepSeek and the Peruvian people and government can fine-tune or change those weights and decide on things. I do think that matters. I'm not trying to dodge your question, but I do think that architecture really matters. Do you have—maybe not every country has the ability to do that. But maybe without open source, we only have two countries left in the world: the U.S. and China. It's possible. That's an argument for open source.
--:-- Tyler Cowen: Economic planning seems like an obvious thing that you would involve AI in. That just seems like one of the first things you'd want to do in government eventually—have AI set your interest rates and, to the extent you have a central planned economy, decide on your production schedules and pricing. If you bought an economic planning AI module off the shelf, presumably it would be pretty opinionated on how to do that. The person who creates those opinions would have immense influence.
--:-- Chris Dixon: I think those countries will insist—one of the reasons that open source will eventually be a dominant model and will be competitive with proprietary AI is partly because of questions like this. I think these governments and a lot of companies will demand open source.
--:-- Tyler Cowen: But even then, there seems room for intermediaries. If you just passed DeepSeek or Llama over to a relatively poor government—again, take the case of Peru—they need a McKinsey-like entity, which is maybe the AI company itself, to come in and tell them how to use it, how to integrate it with their systems. Does this mean American soft power has just won? It's either America or China.
--:-- Chris Dixon: These are hard questions. I think there's a—
--:-- Tyler Cowen: It might be a good thing, too. If everything in Peru is run by American AI companies, the quality of life will be much higher, I think. They may not like it.
--:-- Chris Dixon: I think Peter Thiel said crypto is libertarian and AI is socialist or communist. I don't think either of those is right. The twist I would put on it is AI tends to be centralizing. It tends to be consolidating power. I think you're making that point: the people that produce these things, if they're very powerful things that manage much of the world, being one of the countries and companies that produces them seems to give you immense power. That's what makes the China move into DeepSeek very interesting.
--:-- There are a number of interesting things about that. One is the fact that they are so competitive so quickly and with a relatively small team. I guess we shouldn't be surprised. There are a lot of brilliant people there. The second is the strategy of taking this counter strategy of doing open source. We'll see if that lasts. I would expect India is going to have some interesting stuff. Russia. There are a lot of smart people in the world. The fact that DeepSeek was 150 people—
--:-- Tyler Cowen: Right.
--:-- Chris Dixon: There may be another thing we're trending towards: AI is incredibly important, but it's also just kind of a commodity. The foundation models—it doesn't mean there won't be businesses. There will be businesses at the—the emerging consensus among the people I talk to is that you're going to have cloud providers like AWS, and you're going to have end-user applications like Cursor, and every vertical will have their product. But it may just be that the foundation models end up being like PyTorch or math libraries. It's essentially large-scale statistics. It requires heavy engineering to do the training, but the tricks and secrets end up propagating and a bunch of people know them. It becomes very large-scale statistical analysis of data sets. A lot of people have it.
--:-- Tyler Cowen: If that comes about, what do you think is then the dimension that determines which companies dominate? One argument could be that Meta and Elon just have a lot of resources. They can be in this for a long time. Another argument is that OpenAI has the most powerful brand name with ChatGPT and so on. What's your commercial intuition on what other factor steps into play?
--:-- Chris Dixon: One framing: if you compare the internet to mobile. Both were massively important tech movements. A big difference was with the internet, something like 90% of the net new market cap value went to startups—organizations that didn't exist before the internet: Amazon, Google, and such. With mobile, on the flip side, something like 90% went to incumbents. The biggest pure mobile startup was probably Uber, at whatever it is, $100 billion. There was Snapchat. Instagram got bought for a billion. But the vast majority of market cap value went to incumbents. If you look at what it did for Apple, Google, Facebook, Amazon—they all did fine in that era. They created mobile apps. They leaned into it. Their user base went from desktop at 400 million to 4 billion, a 10x. By the way, it's very hard—almost no venture firms over the 2010s beat the strategy of buying Apple stock. The incumbents turned out to be the winners of the mobile era.
--:-- So one way to frame your question is: will AI be like mobile or like the internet? The concern—I work in venture capital, I'm biased—the concern would be it's more like mobile and it just ends up reinforcing the strengths of the big companies: Google, Microsoft, Facebook. They have more data. They have more money for the training. They have the infrastructure. They have the distribution. How is somebody going to create a competitor to Facebook?
--:-- Host: They'll just make, you know, Lama's already great and they'll pump it through all their channels. And so will Google and they have more data and so forth. So that would be the sort of the, I don't know, the degenerate outcome from the venture capitalist perspective. Look, it could be great for the world, though. You know, it could be, I think it's, a lot of this will be consumer benefit, consumer surplus, I think. And that's good. And I'm happy about that. And if, you know, and we'll always find things to, you know, there'll always be some vertical thing or some other thing for VCs to invest in. So I think that might actually be a really good societal outcome.
--:-- Tyler Cowen: Do NFTs have a future?
--:-- Chris Dixon: I think of, so I think of the, so let me say with an NFT. So one of the things a blockchain allows you to do is to create digital assets. This is one of the unique things: you can create a Bitcoin, which is a digital asset that exists on a blockchain. And those digital assets can be either fungible or non-fungible. Fungible means interchangeable, like a currency. Any Bitcoin can be exchanged for another. Non-fungible means it's a distinct asset. So do blockchains have a future? Of course, I believe yes. And on those blockchains, will you have some assets that are fungible and some that are not? I think definitely.
--:-- When people think NFTs, they think of people buying JPEGs. That may or may not come back. The idea that people will have digital assets that, for example, could represent a movie ticket, an NFT. It can represent a physical object. A lot of people are doing this now where they represent a pair of tennis shoes or a piece of fine art or a housing deed. If we have a future, which I hope for and I'm pushing for, of people transacting with digital assets on blockchains, I think very much that some of those will be non-fungible. And I think some of them will be also purely digital. We're seeing a bunch of people making video games where you have a sword or a gun you can buy as an NFT, and you can resell it and do other things.
--:-- As the digital world becomes more and more important, you'll have virtual worlds, metaverses, worlds like Roblox and many more variations. You will have digital assets in those worlds. And I think a better architecture is to have those assets on a blockchain where the user can move them around and control them, rather than locked inside of the game owned by the game maker. So the answer is yes. But in the true sense of NFT, not in the kind of caricatured sense that people associate with the 2021 bull market.
--:-- Tyler Cowen: Won't NFTs become the property rights system for the AIs?
--:-- Chris Dixon: I think that's going to be a good outcome. I think it's the right way to do it. The reason my book is called *Read, Write, Own* is that "read and write" is a common phrase referring to the first two eras of the internet. My argument is that if blockchains are successful, this era will be about ownership, digital ownership. Blockchains enable digital ownership, and that ownership can be something digital, or something digital that represents something non-digital, like a stable coin or something in a bank. I think it's the right way to represent digital ownership.
--:-- One way to think about the internet today is it's like imagine the offline world where you could never own something. Every time you go to a new venue, you have to change clothes and use all their stuff, and you can't take it anywhere else. That's the weird structure on the internet now. You go to Twitter and you gain followers, yet they own your followers and you can't take them with you. You go to play a game and everything's stuck inside that game. One of the ideas in the blockchain world is we can shift that power back, and the user can have this persistent inventory of things that they own and control. That's what a crypto wallet is. You can take it to different services, and you unbundle the ownership of the data from the provision of the service.
--:-- Tyler Cowen: What's your favorite book in philosophy?
--:-- Chris Dixon: I've actually been getting back into philosophy lately. I did philosophy years ago in grad school. Favorite book, man. Do you read philosophy?
--:-- Tyler Cowen: Of course, yeah. Plato's Dialogues, Quine's *Word and Object*, Parfit's *Reasons and Persons*, Nozick. Those come to my mind right away.
--:-- Chris Dixon: Yeah, so I did analytic philosophy. I was in a grad school program and dropped out. Quine was one of my favorites, and *Word and Object*, and "Two Dogmas of Empiricism," all those kinds of things. I like Donald Davidson. Nozick I loved, *Anarchy, State, and Utopia*. Reading that with Rawls is a great pairing. I used to love Wittgenstein, both early and later. I was into logic, so Frege, Russell. But now that I'm out of grad school, I'm actually trying to finally understand continental philosophy. I never understood it. I've spent the last three months in a philosophy phase, watching a lot of videos. I highly recommend Brian Magee—amazing. I watched all of his videos. This guy Michael Sugrue, a Princeton professor, has great videos on continental philosophy. I've been reading—that sounds pretentious, I'm not saying I understand it or I'm an expert—but I'm struggling and reading it. I'm trying to read *Being and Time* right now, Heidegger.
--:-- When I did analytic, I would say probably Wittgenstein. I really like Kripke, Saul Kripke. I liked his books a lot. Nelson Goodman was one of my favorites. *Fact, Fiction, and Forecast*—funny enough, I just bought it again.
--:-- Tyler Cowen: Kripke's *Naming and Necessity* is a kind of legendary book on reference and language. I've never been persuaded by that one. It always felt like sleight of hand to me. He's very, very smart, he might be the sharpest philosopher, but I like the book on Wittgenstein better.
--:-- Chris Dixon: He basically invented modal logic. I don't know if you know that story. He was like in high school, he was 15 years old.
--:-- Chris Dixon: So he's a true prodigy. But *Naming and Necessity*—like a lot of philosophy, you have to take it in context. I think of it as a response. I'm forgetting the whole history of it, but it was a response to the descriptive theory of reference, like Russell. You have to read these things in pairing. I was actually last night with a group of people, and I got a lecture on philosophy. It was great because he went through Hume, Kant, Hegel, Nietzsche. I don't want to go too much into it, but I've always struggled with Kant. And then he went into Hegel and explained how Hegel struggled with Kant in the same way that I did and improved on it. For me, a lot of it has to be taken as a dialogue between thinkers over multiple periods.
--:-- Tyler Cowen: Are you getting anything out of Heidegger? Because I sometimes say I've looked at every page of that book, but I'm not sure I've read it.
--:-- Chris Dixon: It's a good question. I have a friend who's really into it, and we spend time together and he's trying to teach me. I guess—if you want, I'll send you some videos that have helped me a lot. I get it from an intellectual history point of view. If you want to follow the history of postmodernism, there's Heidegger and then Derrida, and what's going on in the academy today with relativism, discourse, hermeneutics, and the modern political implications that were probably kicked off by Nietzsche and then Heidegger. I've always understood it in that sense. What I struggle with is understanding him as a theory of psychology. I think of describing the experience of the Dasein and being in the world. To me, it's an interesting theory of psychology—how do I experience being thrown into the world? That whole idea is very appealing to me. The idea of "ready at hand" versus "present at hand," and knowing how versus knowing that—different kinds of knowledge—is a very interesting idea.
--:-- Tyler Cowen: Do you watch John Vervaeke?
--:-- Tyler Cowen: I highly recommend. He's a philosopher and cognitive scientist, really smart guy. He has a 50-part video series called "Awakening from the Meaning Crisis." I've watched like 30 of them. The idea is that in the modern world, we lost religion and lost philosophy, and now people are seeking meaning in their lives through drugs or video games or whatever. We should go back and look at all the great thinkers and see how we find meaning. He's very sophisticated on the philosophy stuff. He goes through 50 parts covering Aristotle, Plato, Buddha, Jesus—literally every great thinker—and how they think about finding meaning in life. Very interesting.
--:-- A big part of his thinking is that there are many different ways of knowing. One thing we've done in the modern world is forget that there's more than just propositional knowledge, more than just "knowing that." There's "knowing how." Religious traditions are very good at embracing multiple kinds of knowledge. That's why you don't just—this is the mistake the Redditor makes about analyzing Christianity, looking at it as a set of propositions when in fact it's a form of life. It's about community, co-development, meaning, rituals. So I think the Heidegger stuff is very interesting on those topics of different kinds of knowledge.
--:-- I guess I'm just very interested in—we all, you and I, have this default naturalistic, physicalist, scientific worldview. And I'm interested in sophisticated thinkers who have a different worldview and trying to understand it. John von Neumann had this great phrase. Someone asked him about set theory. He said, "You never understand set theory. You just get used to it." I think that's true of a lot of these alternative philosophies. It's not like there's going to be one argument you read in Heidegger and you're like, "Oh my God, my worldview changed." You just have to submerge yourself in it. That's what I'm trying to do. Over a year, you keep seeing it from different angles, and hopefully you eventually get a gestalt switch and start to see the world through a different lens.
--:-- Tyler Cowen: If you had to say, not what you like in philosophy as philosophy, but what in philosophy has stuck with you when you make venture capital decisions, what would you cite?
--:-- Chris Dixon: Well, I think the mode of—one thing I like about philosophy is the moves they make. It's almost like a chess game. They have a very nuanced argument style. I only recently hesitated. I don't know how much you want me to go down this path.
--:-- Tyler Cowen: No, be as nerdy as you are.
--:-- Chris Dixon: Okay. Are you familiar with Kant's synthetic a priori?
--:-- Chris Dixon: Right. So this is just really interesting. Hume comes along and says, "Let's analyze how we get knowledge." He says there are really just two ways: relations of ideas or matters of fact. Math is relations of ideas—you can sit in a room and think about numbers. "All bachelors are unmarried" is a famous example; it's embedded in the words. That's relations of ideas. Then there are empirical matters of fact, things we go out and see in the world. Hume asked the famous skeptical question: if that's the case, where does something like induction fit in? Where does causality fit in? Where does time and space fit in? They're neither purely true based on relations of ideas, and it would be circular if we learned induction through induction. How do we know the sun will rise tomorrow? It's always risen in the past. Well, how do we know the future will be like the past? He goes through the whole skeptical argument.
--:-- Then Kant comes along and says, "That's right. But what that means is there are other preconditions for knowledge." That's a really interesting move. A transcendental argument is not going from premises to conclusion; you're going from premises to preconditions to those premises. That move, that kind of plasticity of thought—when I first saw that, I was like, "Can you do that?" He says you can't fit causality, induction, and a whole bunch of things that are our modes of intuition into Hume's two buckets. So he argues, over 300 pages and very nuanced, that there must be another bucket—this other precondition of thought, the modes of intuition. Then he derives this whole beautiful system from that. Then Hegel comes along and changes some of the preconditions.
--:-- I just feel like I learned, when I did it in grad school, this kind of freedom. I use the plasticity. You learn how to do these moves and meta moves intellectually that I find really interesting. It forces you to really think. I've spent an hour just sitting there trying to understand these ideas, and sometimes I feel like I don't. It forces intellectual honesty on you.
--:-- Honestly, I've read philosophy for years, and I don't know if I've ever actually gotten any conclusions out of it. I think you can sort of see both sides, and I feel like I can argue both sides. I feel that way about a lot of things in politics too. I can give a pretty good argument for a bunch of different things. I think you get better at analyzing and understanding all the things, and philosophy helps you with that. I don't know if you actually get answers.
--:-- Tyler Cowen: Before we say goodbye, I'd like to recommend Chris's book again, *Read, Write, Own: Building the Next Era of the Internet*. Chris Dixon, it's been a pleasure. Thank you.
--:-- Chris Dixon: That was a lot of fun.
--:-- Host: Thanks for listening to the a16z podcast. If you enjoyed the episode, let us know by leaving a review at ratethispodcast.com slash a16z. We've got more great conversations coming your way. See you next time.
