Epicenter - Learn about Crypto, Blockchain, Ethereum, Bitcoin and Distributed Technologies - Ali Yahya: Andreessen Horowitz – A16z Crypto Investment Thesis

Episode Date: May 3, 2023

When it comes to top-tier VCs, very few equal Andreessen Horowitz' (a16z) impressive portfolio, which stretches over both Web2 and Web3. Their crypto venture arm includes investments in most industry verticals, from infrastructure projects to NFTs and blockchain gaming. Constructing such a solid investment thesis requires conviction in the long-term prospects of the industry, as well as an influential network.We were joined by Ali Yahya, GP at Andreessen Horowitz, to discuss about a16z structure, operations and investment thesis, as well as the broader crypto landscape and value proposition.Topics covered in this episode:Ali’s backgroundWhy Google didn’t embrace cryptoa16z’s team structureThe role of a GP in a VCa16z’s thesis when it comes to crypto investingWhich crypto niches Ali is most bullish onGovernance vs. platform riskValuation-based investmentsProtocol revenue and monetary premium for L1sToken vs. EquityHow to prevent banking crisis contagion for crypto projectsEpisode links: Ali Yahya on Twittera16z cryptoa16zThis episode is hosted by Sebastien Couture & Meher Roy. Show notes and listening options: epicenter.tv/494

Welcome to Epicenter, the podcast which talks about the technologies, projects, and people driving decentralization and the global blockchain revolution. I'm Sebastian Khrushu, and I'm here with my co-host Mayor Roy. Today, we're speaking with Ali Yafia. He's a general partner at A16Z. Of course, they are a massive fund, which bears know, I don't think they need any proper introduction. Everybody knows who A16Z are and the investments they've made. but we're going to be chatting with Ali today about a bunch of stuff, including A6NZ thesis, also taking a step back and looking at fundamentals of crypto networks and some of the broader banking and political crises that have been happening around the world in the last few weeks and

months. But first, yeah, Ali, thanks for joining us and maybe starting off by giving us a little of a background and how you became part of the team at A6CNZ. Yeah, it's your thing. Thank you, Seb and mayor. It's great to be here. So yeah, maybe I'll give you a little bit of a little bit of my history. I was originally born in Mexico City. I was born and raised in Mexico. And throughout my childhood, I was fascinated by technology. I was very into engineering and to the things that humans have built. I was always in awe of things like, you know, fast cars or planes or like tall buildings, all of the superlatives that could be kind of the great things

that humans could build. And I was in particular fascinated by computers and by robotics and prospect of AI and kind of that whole world. And very much I wanted to pursue a career in engineering. And so I kind of saw, I looked like toward

the U.S. as a place to study as a place to kind of make a career. And in the end, I ended up going to Stanford to study computer science, focused on systems in particular. I really like to write code. And that was like the track that would expose you the most to large like programming, programming projects and programming experience. And so focused a lot on distributed systems, operating systems, computer networking, computer security. Those were all of the fields that were most interesting to me. And this was by the way,

like my undergrad started in 2006. This was before crypto was a thing. At the very tail end of my undergrad degree in 2010. I was doing research with David Mazzieras at the computer security lab at Stanford. And that's when we all became aware of the Bitcoin white paper. I still have a few emails in the lab of us being like, oh, like, isn't this a cool thing? We should just like play with it. And we all, we all like downloaded it, played with it. We mined it. But I think most of us, especially certainly I did, I missed the implications of how important it was going to be. So even though I was playing with it and mined it, I didn't track anything. I didn't track any of the private keys that I was mining with.

And so that ended up being later on, like three years after that until 2013, when it started to become more clear what the implications really were. That was my first very visceral lesson in the space. And I kind of, that kind of like became seared in my memory as an experience early on. And I feel like many people who've been in this space in a while have. similar stories. But I was fascinated by it. So I kind of was following it closely since then. At the time, there weren't really many ways of working on it full-time.

And I was also, as I said, I was also interested in AI. And so I actually joined Google. First, I joined Google X, and I joined this robotics project called Replicent, which had been started by Antibuven, which was this agglomeration of a bunch of different robotics projects and the mission was to build an operating system for fleets of robots. And that was nice because it kind of married my interest in distributed systems and then also machine learning and robotics. So it was a very fun project to work on for two years. That ultimately, it was a project that was at X. It actually became the everyday robot

project, which is now a real thing. It's like it's live, it's now public. You know, how X is super secret about everything, but it's now a project that people can talk about, which is nice. But after that, I moved over to Google Brain, where I did a little bit more of the same. It was distributed systems for machine learning generally, and I was working closely with the robotics team at Google Brain. It was a core contributor to TensorFlow, to make TensorFlow more capable to work in that kind of environment. And then finally, in 2017, like all throughout this time, by the way, it was still kind of fascinated by crypto. Google, of course, will never touch crypto with a thousand football. So I could only really follow it and play with it in my own time. And my roommate

at the time was Juan Bennett, who's the founder of Protocol Labs. So we were very close. He's maybe my best friend. And we were kind of contemporaries in school. So we go away back. We were living together at the time and talking about all of this stuff. So I was fascinated by the space. At one point, they got connected to Chris, to Chris Tickson, your firm. And we kind of hit it off. And he essentially just said, like, hey, you should join. We should start a crypto fund. Because back then the firm was investing.

Chris was investing in crypto. He's been investing in crypto since 2013 or so. But it was all entirely within the context of the main firm. So the thinking was we should build a dedicated team for crypto and really double down. And that was his pitch to me. And to me, that felt like an operator. I love the space. It's super interesting. I was frustrated that Google is just such a big company. It's very hard to actually get things done. So I dropped, I joined the firm in 2017. It was Chris and me at first, and then we started to build out the team. We recruited a few other people. And then since then we built out the entire crypto vertical here at the firm. And we're now up to 80 people or so across a bunch of different functions. And the idea is to not only make great investment, but, but to also support every portfolio company in every way we can.

And I'm happy to talk a little bit more about the way that we tend to work with founders. But that's a story. It's been five years now, and it's been quite a ride, and it's been awesome and really great. That's a fascinating story. I didn't realize that your roots into crypto went so far back, and that, like many of us, you had been burned early by not realizing the implications of what you were getting involved. did. What, you mentioned that you think that Google will never touch crypto with a 10 foot pole or perhaps even a longer pole. Why do you think that is and why do you think Google hasn't,

you know, embraced crypto because there's such like a forward thinking company on so many other verticals and industries, but crypto seems to be something that they are incredibly hesitant to touch on. Although, although recently there were some announcements that they were implementing like if they're cloud offerings, like some MPC services and things like that, that seem to indicate that they are open to providing infrastructure services to crypto companies, but at least like on the payment side or on any of the higher level aspects of the stack, they're quite absent. Yeah.

Well, I could go into an hour-long rant about Google that would consume the entire podcast. But I think that maybe the core reason that Google will. will likely never do anything significant in the space is just the classic innovator's dilemma. It's just the fact that crypto is such a disruptive technology. The fact that crypto works by being fully open, by being open source and by decentralizing power as much as possible, the fact that that is at the core of what crypto is about is entirely at odds with the entire business model that Google, that Google leverages to survive. And so in order for Google to Google to actually truly move into the space and actually do something meaningful, it would have to

disrupt the core business models that drive it today. And because as a company, it's no longer really founder-led, you don't have Larry and Sergey there as involved as it used to be. And because of the fact that it's atrophied in all these ways and that has become, I mean, I think it's fair to say, it's become a more complacent company than it used to be. It's just really, it's just very unlikely for them to do anything that's struck. Everything that they do now, it seems to be much more incremental and seems to be in service of the core business model that has powered them this far. So yeah, I think I'm not holding my breath for Google to do something meaningful. They do have all of these little experiments because Google does still have this kind of freedom where

you as an engineer, you can use it quote unquote 20% time and play around with things. And there are a couple of teams that are maybe experimenting, but it's hard to imagine that Google will actually invest heavily and make it a core part of what they do for a part of the above reasons. And maybe that Google is like the IBM or the HP of our generation. Oh, no, for sure. I agree with that completely. I mean, at this point and sort of this size, yeah, the innovator's dilemma is probably the biggest culprit here. And that's the case for many, many companies, not just Google, you know, we could say the same thing about Microsoft or or Facebook or, although, you know, there have been some, I think, other types of interesting

experiments there that, you know, might leverage crypto at some point, but in a very different form than what, you know, we know and sort of thriving for, I think, as people have been in the space for long as we have. So just maybe zooming in a little bit on A16C and the, and the, the, the crypto team, you mentioned it's 80 people. What does that look like? And, you know, what are the, you know, what are the key roles there? I'm curious because I'm also starting a fund. And, you know, there's a lot of things that I think I have to learn from, you know, other funds. And, you know, even though ECCC is a massive organization and like, you know, is nowhere near the size that I'll ever, you know, aspire to get to, you know, understanding sort of what the core support structures there are and,

and support functions. Yeah, give us a bit of an overview of what the team looks like. Yeah, of course. So the whole vision from the very start was to build a vertical within the firm that mirrored the original A16Z form factor, but was very tailored for crypto specifically. And so originally, I mean, A16C as a firm, the whole vision from the start was that it's essentially a platform.

the firm that not only makes investments, but is truly a partner end-to-end across many different functions to help the companies in our portfolios succeed. What that means for crypto is that because crypto has certain big differences from traditional companies, there are things in ways that we can help that are fundamentally different that require building out the team differently. So as an example of that, one of the observations that we made was that because most of the work in crypto tends to be open source, because the work that's done tests to be out in the open, it's actually possible for collaboration to happen between organizational boundaries. And that means that as a venture firm, we actually can get involved in some of the core technical challenges that some of our portfolio companies are facing, which is something that would be much more difficult to do. do in any other sector of technology. Can you imagine, for example, in the world of biotech,

or in the world of, like, I don't know, deep enterprise software, in order to get access to the code base or to the intellectual property that the company is working on, you really have to become a full-time employee. And so it would be difficult in those worlds to contribute technically, to contribute to the core science of technology

that each of our portfolio companies are working on. Whereas in crypto, So it's a much more collaborative environment. The fact that it's all open source actually changes the game and makes it possible for us to build the research team, which is led by Tim Ruffgarden. He's a former professor of Snapper. He was actually my professor. I took all of his classes, and these were some of the most interesting and hardest classes that I ever took. And he's a pioneer in the world of algorithmic game theory, and he's also extremely deep in the space.

So in a sense, he's a bit of a bit of a unicorn. There are very few people who have the knowledge and breadth of the traditional literature, the academic literature around distributed systems and game theory and mechanism design, but who also deeply understand crypto and blockchains and bring those two together to do a pioneering research in the space. He's great. Oh, incredible, yeah, of course. And he was actually the person who did the formal analysis of EIP-1559. and actually got the community, or helped get the community more comfortable with incorporating

the IP-159 to Ethereum. So he's the head of research, and under him, we have a team of world-class researchers across cryptography, distributed systems, economics, game theory, mechanism design, and we're still kind of open to continue to expand that team. And the idea there is that they can really help our portfolio companies, at the very least, understand what's been done before. And then also we can help them think through some of the open-ended problems that are actually research problems. And that team works very, very closely with our engineering team, which is headed by Eddie Lazaran. You may also know. So Eddie Lazarin is our

CTO and head of research. And under him we have a team of like extremely senior, very sort of they, they, they, they, like, a smart contract engineers, security engineers, people who build, who build things. And they work closely with our research team and our portfolio to, to actually produce production grade code. And so much of the work that they do tests to be open source. Again, the idea here is that we want to contribute. We want to help our portfolio solve open-ended problems, but we also want to help the community at large. So all of the work that we do is fully public. We publish all of the research papers

and all of the research content, both in our website and then also in a peer-reviewed conferences on the research side, and then we publish all of the code that our engineering team produces in our GitHub repo, right? So it's all open.

So those two teams were closely with our portfolio, and there's a third important team, which is very different from like the traditional venture side, kind of the main firm, and that is the regulatory and policy team, because as you guys know, that's obviously like a huge open, open kind of question in the space. There are no bright lines as to what is regulatory compliant and what isn't. And as a result, we have to take, we have to be very smart

about how we navigate the regulatory landscape. And so we have a team that includes that includes people who have been extremely senior just about every regulatory agency, including, for example, Bill Hinman, who was the person who actually established the framework for decentralization at the SEC, who established, for example, that Ethereum should not be considered a security,

in part because, or largely because it fails like the fourth prong of the highway test, the fact that it's, that there's no managerial and entrepreneurial efforts of others because it's fully decentralized. And so that whole framework came from him. And so he works closely with us and works for their portfolio companies to think through some of the regulatory constraints. And so then it's afforded to have both of those three teams, the research, engineering,

and regulatory policy teams in the same room, often with the portfolio company to be able to explore at the same time all of the possibilities on the research and engineering side, but then also the constraints. Like how do you design a protocol that also is also compliant and doesn't run afoul of like securities laws or email, like other kinds of regulatory constraints? So those are three important teams that are fairly unique to crypto and no other vertical at the firm has those teams just yet.

And then beyond that, we have like the more traditional, the traditional teams that help our portfolio companies and the ways that you'd expect. So, for example, we have a big recruiting team to help our portfolio companies recruit at every level, recruit both individual contributors, people, like software engineers, product managers, generally product or UX people, as well as executives.

So hire, like if you want to hire like a general console, for example, which is, again, super important given the regulatory piece. So the recruiting team is an important function. We also have the kind of the biz dev, market dev team, team which helps with partnerships. And that team is connected to just about every team, every entity, every individual, important individual in the space. And also outside of the space. So we want to connect our portfolio companies to potential partners, to exchanges, maybe also partners from the traditional world that we want to connect them to a company like Stride or Visa or companies like

that. That team is basically the connector of everything. And then beyond that, we have the marketing team, which helps us basically produce all the content we do, but it also works closely with the portfolio companies to shape the narrative. It basically help them tell the story of their company. And we help, especially around important moments, like announcements, or, for example, if there happens to be some kind of crisis, we can really be helpful at managing the PR and the cons around. around moments of crisis. So that's a very important team. And then finally, we have the network operations team, which is the team that allows us to participate actively in all of the

networks that we were involved with. And so that involves staking our tokens, for example, for layer one blockchains. It involves voting our tokens for protocols that have on-chain governance. It involves delegating our tokens to other players in the space so as to increase the decentralization. of these networks. So for example, we have this delegate program that involves many university blockchain groups that are very eager to learn a lot about space and who would like to participate in some of the governance conversations that happen for protocols like UDISWOP and compound and protocols like that, like that. So we very actively, we have a whole process for vetting the various different delegates that we can potentially delegate tokens,

too as a way of including more voices in the conversation and making the space more decentralized. So that was a bit of a rant. That's maybe the full rundown of the various different ways that we work with the portfolio. That's how the team is structured. So, Annie, for those of us who don't know a lot about the venture capital business, you started off as partner in the crypto arm of A16C, and then you became general power. I think a couple of years back. That's right. What does a partner do and what does a general partner do in this organization structure?

General partners are responsible for ultimately making investment decisions. And then the whole investment team works together at finding the best founders, being connected to everyone in the space, thinking through what our thesis for the space is. It's always an active conversation, and we have many meetings internally to discuss what we believe to be true and what we think might be interesting to invest in. So it's a fairly flat organization that everybody has a very active role in making any one particular investment. And then the one distinction is that ultimately a general partner is the one who makes the final decision of whether or not to invest. And also is the primary point of contact with the portfolio company thereafter. though the whole team again

I just basically went through the whole team that ultimately the portfolio company will work with and the general partner ends up being more of a router we're like it's like oh you need help with recruiting well let me let me connect you to Craig like research Tim and Tim like you should talk to Tim or there's an engineering problem

you can talk to edit so it's really a team effort like we all we all work very closely together and we're different from many other venture firms in that we really do work as a team and it's not as if every investor has its own, his or her own mandate and is like sort of it's like survival of the fittest, which tends to be the case in other places

where there's just less teamwork and each person is more of a solo actor. Here, we do have a holistic general thesis for the space. We all believe it. We've all kind of crafted it together. We debated constantly and we work closely together to make sure that all of our companies are supportive. Let's maybe talk about the thesis a little bit, or to the extent, to whatever extent we want, we could probably talk about the thesis for a long time. So what is the, what is the ACXZ thesis when it comes to crypto? Yeah, well, the heart of it starts with the insight, which I think by now is now better understood

that a blockchain is a computer. It's a, the best way to think of it is a, is a computer and it's not, it's not some sort of database or ledger or whatever yet. You know, there are many, there are many other terms that people like to use to describe blockchains. And that the right model to think about this is a new paradigm for computation. It is a computer that has special properties. If you, if you look back at the history of computing, every, every paradigm along the way

has introduced something that's qualitatively different. And that's what that's what kind of demarcates the different, the different phases of competition. And that's also what allows for the kinds of applications that we can build to expand and to kind of lead into more aspects of our lives. And so we started with things like mini computers and mainframes, and then we moved on to personal computers. And then we had cloud and mobile in the whole kind of internet chapter. And now we contend we have, we have blockchins. And if you look at previous waves of computation, new kind of computing paradigms, like, for example, take mobile, tend to actually be worse than previous paradigms in many ways.

Maybe in most ways. And for example, a mobile phone is a slower computer than a personal computer. And it doesn't have a keyboard. And it has limited to that battery life. And it has a small screen, right? And it's kind of maybe clunky to use it first. But it has one key feature that personal computers don't have. And that's that it fits in your pocket.

and that it has a GPS and that it has like this, all of these mobile features that actually unlock new kinds of applications that you couldn't have possibly built on a personal computer, which include things like Uber, for example, right? Or include all of the, you know, the things that we love about like the mobile revolution, the kinds of things that the people have built over the past couple of decades. So our contention is that blockchains, yes, like there are criticisms of them that are true, like, for example, that they are very slow. Like a blockchain is a computer.

It's a very slow computer. It's an expensive computer to run. But there are features that blockchains have that traditional computers don't have. And in particular, blockchains are a kind of computer that allow developers to write programs that have a life of their own, in a sense. These are programs that can make strong commitments about their own behavior in the future. These are programs that run without interference from anyone, from including the people who originally wrote them, including the people who run the actual physical machines that execute the logic of those programs, and including the people who interact with those programs as they run, these programs can run free of interference from all of those players. And that's a very powerful thing.

and in particular is powerful because it can create a kind of trust that a normal program that runs in a computer that someone else controls cannot engender. One way to putting this, this is credit to Chris, is that blockchains invert the power relationship between software and hardware, where traditionally hardware has power over software because whomever controls the hardware can turn the software off or can change it into whatever they want. Whereas in crypto, on a blockchain, is the reverse. The software actually has power over the hardware.

The hardware is actually just provided as a commodity by various different participants who have an economic incentive to participate, but who do not have any kind of control over what the software does. And that's a very powerful shift. That has never happened in the history of computer science. Having a network and the software have power over the hardware, that's what's unique. That's what's different about blockchains. And as I mentioned before, the powerful thing about this is that it can create applications that

have a different level of trust. The first application of that was Bitcoin, right? The fact that you can trust that Bitcoin will only ever have 21 million Bitcoins and that transactions will be processed in a way that is well defined and that is valid and that no one can really subvert the logic of how Bitcoin works. And that's the first example, but it's a very simple example. And we can And there's so many different things that you can do. Like now that you have this primitive, you have this kind of program that can make commitments, what can you build with that? And some of the applications that you could build might be financial in nature.

Finance lends itself well to this space because of the fact that it requires trust. So you can build maybe more sophisticated financial primitives. That may be the whole, like, defy world. You can build exchanges that are decentralized that have a somewhat property of trust. Or you can build a lending platform. You can build stable coins. You can build some of the interesting defyprivatives of people I've experimented with.

But you can also build things that have nothing to do with finance. And I think one of my favorite examples, which is fairly topical these days, is the whole world of social media, because you can conceivably build something like a decentralized social network that is truly decentralized and doesn't have a monopolistic $44 billion tech giant in the middle that can control who you get to follow,

that can control who gets to follow you, that controls what you see and has like the one given algorithm, the one given curation mechanism that controls what your feed contains. And instead, you can build something that is decentralized where you as a user control your own identity, you as a user control who you get to follow and who get to follow you.

You maybe even control the algorithm that's used to curate content and you have control over your feed. and there could be a whole marketplace for different kinds of recommendation algorithms that is competitive in people participate in it. And then you can do that with crypto in the same way that you can build Bitcoin, and it has the same property that you can trust that the network will maintain and will commit to certain rules that will be upheld throughout time that cannot be changed by some kind of central operator.

So that's basically the heart of the thesis, and there are many implications to that. But that's where it starts. Yeah, that's an interesting way to, you know, I think if you still, if you ask a lot of people today, what is a blockchain? I think like 90% of people will say it's a decentralized database or it's like a decentralized ledger or, you know, this idea of a decentralized letter, ledger still permeates. And I think there is a shift, you know, in narrative in the industry that is that is leading more. towards this idea that it is a computer and you know like it sort of started that way right with

Ethereum as the world computer and then it went to decentralized ledgers like in the whole like mid 2000s with like enterprise trying to build whatever their private blockchain that whole thing failed and blockchain not bitcoin exactly yeah yeah and so i think a lot of that narrative came from that and and now it's it from from when i'm seeing like teams that we're looking at and and the narratives that we see, at least like, I'm fairly well connected in the interchain ecosystem, those narratives are starting to come back again. And I think it's quite positive because it really does, I think, align well, like you said, with the evolution of computing. And, you know, it might very well be that, you know, if you go back to that example of mobile,

mobile enabled new types of applications that are well suited like for for people who are not who are not stationary right it's like there there are applications that are well suited for for mobility they're mobile right um where here you know the types of applications i think we will discover is types of applications and sort of like self-serving the what i'm about to say but types of applications that that require, typically that will require trust in a person or an entity. And what we'll realize, you know, as crypto and decentralized applications become more mainstream, is that this will be the defining factor. And yeah. So, you've started off with the reminds that these blockchains that distributed computers and the key property is that programs running on these computers are able to credibly commit

something about their future behavior to participants and because they're able to credibly commit about future behaviors, they naturally end up getting applied to finance because those sorts of commitments are important in finance. Yes. From this viewpoint, does it, do you see particular verticals

in the crypto area verticals of crypto that are especially good or especially bullish about? So what I mean is you can scan across crypto you have the crypto gaming area then you have the

storage area or like computing services area then you have D5 and you have these exchanges you have this data sharing area data dows etc right are there any particular areas that you have especially bundish about and believe that in 20 years, that those one or two areas are going to be the massive successes, the power law successes? Yes, absolutely.

So I think one of the other ways in which trust really plays an important role in the world of software is this notion of platform risk, which historically, in the world, web 2 world has been a huge source of risk for any startup that is potentially building on top of some other companies APIs. And I think you, I mean, we've all kind of seen this. There have been many startups, for example, that built on top of Twitter APIs a lot of time ago. Remember like Tweety and there's this whole kind of set of companies that were doing interesting things and composing different kinds of Web 2 APIs. But that ultimately, because those APIs are controlled by a centralized company, the central company can change the rules and can decide to move against

the startup that is starting to get big and instead charge more or instead like change something about how the APIs work such that the central company captures more value and then maybe the startup ends up dying. And this has happened time and time again, right? There's certainly kind of like the tweet, Twitter example. There's Zinga Facebook. And it's also, by the way, it's back to kind of the world of operating systems where you have operating systems like Microsoft Windows, also subverting the functionality of some of the applications that are built on top. So Platform Risk has been a theme and technology and software for a very long time. And I think it gives you some insight as to how crypto can be used for things that are no longer

financial in nature. So when you can build a program that has a kind of API that is committed, it to, that the program commits to not change it, then you've neutralized platform risk. And it as a result allows startup companies to build on top of that program in a way that it is no longer possible in the world of Web2. Like these days, if there's a company that comes to pitch us, for example, if they come to pitch our main phone, consumer farm, with an idea to build a product that depends entirely on the APIs of Instagram

or the APIs of Google or whatever, that's a non-starter because it is now very well-known that that won't work. You cannot build a massive company on top of the APIs of a company like that because at some point, sooner or later,

whomever runs that centralized company will change their roles and that they will want to capture more of the value which doesn't work. So now, I think this is, this is by the way, what enables composability in the world of crypto, the fact that you can neutralize

platform risk. The fact that you can actually build on top of Ethereum and not worry that Ethereum will change the rules on you and will try to take a bigger cut of the money that your product makes.

That's huge. And the fact that that continues even above that, right, the fact that you can build a product or a startup on top of, say, uniswap. You can build a startup on top of some of the applications

that are already running on top of Ethereum. because those applications also make commitments that you can actually trust. So I give you one example of a vertical or of sort of an area that's super interesting, which is that of social media, like decentralized social media. The fact that you can build a decentralized social network that has rules and has APIs that can be trusted in perpetuity is very interesting because now you can have a social network where the social graph is kind of like a public good, right, that exists

on chain that is owned collectively by all of the users. You own your node in the graph and other users own their own nodes. And then all sorts of startups can build on top of that same social graph. You can imagine a startup building one particular client that gives you one of you into the social graph and that maybe decides to curate content in one particular way, but someone else can build a client too that's different, that has a different interface that maybe highlights different kinds of content. Maybe one of them is kind of like Instagram and another one is more like medium. I don't know. But it's the same social graph and this is what enables the composability. And none of those startups that are building on top of the social

graph have to worry that the rules will be changed from underneath them and that as a result they'll go out of business. So that's one very interesting reticle where where there's notion of platform risk plays a role where crypto, I think, in crypto and the trust features that crypto provides really make a difference. But there are many others. I think another one, as you mentioned, gaming is another interesting area, where you can look at the history of gaming. You can look at examples like Fortnite, right, the fact that people pay a lot of money

for in-game items that are purely cosmetic, right? The fact that these in-game items don't actually change gameplay at all, They only change the way you look in the game or the kinds of little emotes that your character can make. And it's like a multi-billion dollar run-grade business. That's fascinating because the company at the center of it has full control over that economy. Imagine if you could actually build a crypto economy into a game such that the end-game items that you buy are actually yours. such that you can actually take them out of the game and maybe even take them into another game.

If it's a standard, for example, that multiple games begin to support, then maybe your character doesn't actually isn't just specific to one game. You can actually take it from one game and into another game, and this becomes a much larger economy that spans multiple different kind of digital universes, and this is what people like to call the Metaverse, which should now become a word that has been co-opted and is very overloaded. But it's an interesting idea, and it's, it's, it's born bottom up.

And again, the trust features really matter here, right? Because it's important that you trust that you actually own the things that you own. That's what allows this ecosystem to emerge. So that's another interesting area. Maybe a third one, and then I'll stop, I'll stop branding, is the area of media generally. And maybe I'll lead with an example. So we're investors in a company called sound.xyZ, which is, is a decentralized music streaming platform, which I think is a super interesting application

of crypto. If you look at the music industry historically, the record labels have an inordinate amount of power. And the reason for that is that they control the catalog. So a company like Spotify, for example, cannot really negotiate against the record labels because any one of these record labels controls so much of the historical catalog that if they pull out, Spotify becomes almost unusable because now it's missing, like a third of the music, and its whole value proposition of being able to provide you with every song that you might want to listen to

is no longer held up. So the record labels have enormous leverage. They take a huge cut, and then Spotify has to take a cut, and so therefore there's only a sliver that's left, that it actually goes to the artist. So the artist is royally screwed. And if you actually look at the numbers, there are something like 8 million artists on Spotify and only like 14,000 of them. So 14,000 out of 8 million make more than 50K per year, which is hardly a living. Which is to say, basically, no one makes any money on Spotify.

The only people who make money are like the very head of a distribution, the kind of the Taylor Swift's or the sort of a jZs of the world. And then everybody else makes 50K or less, likely much less. In order to actually make any kind of money on Spotify, you have to have millions, if not tens of millions or hundreds of millions of streams. It's insane how much traction you have to have as an artist to make any money on Spotify. So sound is a little different. So you guys remember SoundCloud from before?

SoundCloud was this nice guy of music servers. You as an artist can upload music. And it had this cool feature, which was that you can, as a user, you can comment on the track as it plays. on the website, such that other users can then see your comment. So as the music plays and it rolls over the time at which you left your comment, they can see what the comment was, which was like a cute idea. It created this kind of social dynamic around a piece of music.

So sound to that idea, and their initial product idea is that you can actually make that comment an NFT, meaning that, say, as an artist, you upload a piece of music, and there's 100 NFTs for that piece of music. whomever buys those NFTs, if you have one of those NFTs, you now have the right to add a comment to the track. And if you sell your NFT, then the comment disappears and then whomever holds the NFT now, the new folder of the NFT can now add a comment as well. So this creates a bit of a little market, like a little market, right, for these NFTs. The beauty of it is that 100% of the money that is recouped from the sale of these NFTs goes to the artist, or maybe like 95% of it and then a small percentage of it goes to sound in the protocol.

So that's great. But the more important piece of this is that as an artist, you now have a direct relationship with your super fans. These are the people who are willing to buy an NFT to add a comment to your track. These are the people who truly believe in your music. I would truly like your music. And it's not only a relationship, it's an economic relationship. You know the crypto wallets.

And that's a community. You can now start doing interesting things. You can, for example, I don't know, do like backstage passes for the people who hold any one of these NFTs. Or you can have like a Discord group that is token-gated and only people have NFTs of this kind can access it. Or any other kinds of benefits where you as an artist now have like a more direct relationship with your fans. And it's connected to this idea. There's this great blog post by David Kelly called A Thousand True Fans, which argues that all you need as an artist or as a creator to make a living online is 1,000 true fans. And his logic is if you define a true fan as someone who pays you $10 a month, then if you do the math, right, a thousand true fans, each paying you $10 a month, that's $120,000 a year.

All you need is $1,000,000 a year. clearly on Spotify that doesn't work but you need millions of fans to actually make any kind of living but if you had a more direct economic relationship with your fans, you can actually make a living and so Sound is basically exploring how you can use crypto to do that

right? As an artist, you can build a community of two fans and have an economic relationship with them. If they pay you at least $10, that's more than you could make on Spotify. So far, by the way, they paid out like $5 million dollars. It's not... As a content creator getting to a thousand

and true fans, even for just five bucks a month, it's not easy. It's not easy to get there. But making it easier for that to happen. And, you know, things like Patreon have definitely shown how, you know, this model works, right? Like, I think Patreon has a similar type of model and one that is very different from, say, like, a Spotify model. But, you know, I think, like, those, there were probably, you know, just as many musicians 50 years ago that were making zero dollars, they just weren't streaming.

But I mean, of course, like the platform allowing this, you know, does allow opportunities for people to achieve those goals. You know, there was one thing you were mentioning earlier about, you know, the platform risk. And recently, in last year or so, I used to have this idea that like, blockchains are immutable, the rules that are embedded in the blockchain and the smart contracts are immutable. and like that's it and you know you could you can sort of like deploy there and be assured that like your your code will be run in perpetuity forever and then and then there was the whole juno whale thing and I don't know if you followed this this particular part of Cosmos drama but essentially

a governance boat on Juno which is a smart contract chain in the Cosmos ecosystem that that runs Cosmosin smart contracts a governance boat decided that funds that were held essentially by an individual who had received air drops for reasons that I'm not going to get into here were to be clawed back essentially and because because you know the rules of the airdrop had not specified certain provisions about people holding other people's funds or whatever it's a there's a whole block quest about it

and and if that's not enough the people who were meant to take those funds and sort of transfer them into a treasury so that they could be effectively managed or whatever made a mistake when transferring the funds and lost like tens of millions of dollars or maybe even hundreds. I mean, it was a massive catastrophe.

But it just showed me that I just, then I realized that, you know what? Like, blockchains do have this immutability aspect and sort of like this, like, this is, this platform. risk assurance when they arrive at such a state that governance is solid enough that if governance and incentives are aligned such that validators are not going to screw the user base or account holders. And I think there's a delicate balance between validators, users, and application developers that really mirrors kind of governance that we know and sort of from from

from our everyday lives and democracy and sort of like human governance. But when it comes down to it, like, blockchains are our mechanism for executing, you know, human decisions. And the power balances that exist in the real world can also exist in blockchains. You know, how do you reason about that when thinking about, you know, this whole idea of platform risk and eliminating platform risk? Yeah, it's a really good. It's a really good point.

I think that the way to think about this is that a blockchain gives you an enormous design space, on top of which you can build all sorts of different things, including different kinds of governance systems, which I think is an exciting, it's a really exciting thing, because now it is possible for a thousand different companies who are building different ecosystems to run a thousand different experiments about how to run, a governance system for their community, which actually gives us insight about governance generally. And it gives us insight that maybe can translate to beyond crypto.

And in a way that was impossible before, where like I think before crypto, if you wanted to test a new kind of sort of political governance system, you had to spin up a whole civilization, institute your new experimental governance system, wait 10,000 years, and check to see if everyone has died, or if it has somehow paired well or not. whereas now it feels like you can run much more isolated experiments with a network that then maybe does one simple thing and has a community that governs it. You can come up with a new governance system that hopefully does better than other governance systems. And some will do well and others won't. And I think over time, we will figure out what the best practices are.

Maybe we conclude that one token, one vote is not the right model because it leads to a kind of plutocratic outcome. maybe we conclude that the governance system has to be restricted in some way so that you can't claw back tokens. Maybe that's not something that's within the purview of the governance system. Maybe we conclude that we need different kinds of stakeholders. Like maybe you have token holders as one camera in the governance chamber. And then there is also like actual applications or drivers of volume or drivers of activity in the network that also have some say and have some ability to. participate in the governance process. It's a very big design space, and I don't think we found all of the answers.

And I think that some of the initial experiments are interesting. I think, for example, Maker is pioneering. What pioneered, I think, one of the initial models for governance with the MKR token and the way that governance works for that role system, which is very complex, but it was interesting. I think we learned a lot from it. And also compound was also initially one of the, one of the, one of the, one of the pioneers on token-based governance with their governance module, I forget actually the name

of the code file governance alpha or something. And I think since then we've seen many other experiments and it's exciting to see, I think, what when you think people may try in it, and along the way, there will be failures, like the one that you described. But I just think that that's part of the learning process. In my crypto journey, one of the things that I found very hard is making sense of crypto token valuations. By that I mean is, you know, pick any random coin. And you often have the sense that, okay, if this coin is trading at 100 million, that's awesome. And if it's trading at 10 billion, I'm not going to buy that coin.

Right. You often have this kind of gut instinct where what you invest in depends on the market. kit cap of the coin you're looking at. Yet there is no good way to quantify or think about it logically. So I'm actually curious how A16Z as professionals that make these decisions stay in, day out, how do you think about valuation and what makes sense and what does it? Yes. It's a great question. Well, and I think this, this, often is a conversation that we tend to have on a case-by-case basis. It very much depends on the kind of network, the kind of protocol,

the kind of product that we are looking at. And there are different models for different kinds of things, especially when it comes to different layers of the stack. So, for example, we can work through the stack, something that exists at the user interface, like client level, an application that's maybe communicating with protocols that run on chain, but that it in and of itself isn't on chain, and that maybe has a more traditional business model,

something like a wallet, where a wallet's business model is to potentially just take fees on all transactions or something like this. That's straightforward. We don't have to spend too much time on that. We use traditional methods of reasoning of evaluation in that case. We move one level down. There may be applications that run on-trial.

chain that are protocols that for example like run on top of Ethereum and this could include some of the some of the defy protocols like uniswap or or compound or like applications like I mentioned like sound that XYZ as like an on chain component there's a protocol that runs on on Ethereum. Oftentimes in those cases there's this is by the way I think of apparently like a misconception about space people think that people think that all cryptocurrency are just like fully speculative, in most cases, there is still some notion or something that resembles revenue, where there's like a fee that's being taken by the protocol on chain, the fee accrues to some on-chain treasury, and the token has some claim on the treasury or has a potential

claim down the line based on governance potentially, and in the same way that maybe the shareholders of a company have a claim to the assets on the property. balance sheet and to future cash flows, token holders similarly have that kind of claim over a protocol's treasury. Of course, it's fully decentralized, right? So there's no company. When we're talking about the protocol, there's no company that controls the treasury, and it's just a broad ecosystem, which is, by the way, it's why the token is not a security, the fact that it's a fully decentralized ecosystem where there aren't any asymmetries of information. It's just all on chain and it's all open source.

So for those networks or those protocols, it's also reasonable to think of it in terms of the traditional methods of valuing the company. What is the current revenue, the current cash flow? And what is our thesis for what that might look like in the long-term future? We believe that this protocol will somehow become like a fundamental piece of the crypto world.

and if crypto is going to ultimately reach billions of people, then you can make some argument for why it will be much bigger than it is today. But the methods are still fairly traditional in those cases. And then there's the question of how you would value a layer one blockchain. So something like Ethereum, for example, the good news is that even a blockchain like Ethereum, which is a layer one blockchain and the token serves many different kinds of purposes, also has a notion of revenue, and that comes in the form of

ETH getting burned with every transaction, thanks to EIP-1559, that is a form of revenue because it is taking heat out of circulation, and that's mathematically equivalent to essentially doing a distribution of that amount of money to all ETH holders. So you can use that, at least,

to reason about the value of ETH, but there are many other drivers of value for the ETH token. So the fact that ETH is used for staking makes it a kind of productive asset such that you can actually earn money by holding it and owning it in the way that validators and the networks do. And that drives some value to the token because it's a productive asset. You can think of it as capital. It also is the currency with which everything gets paid for.

in the network. You need ETH to pay for competition in the network, which means that people have to buy it to be able to use Ethereum, to be able to participate, which is another driver, which gives the token a kind of monetary premium. People begin to think of it as a kind of money that is also maybe a good store of value. And as a result, they probably just hold some amount of E's longer term, which drives value to the token. So there are a number of different, and there are, there are I mean, I think we can keep talking about the drivers of value for a token like Heath. But we did to think of all of those things and try to at the very least get some quantitative sense for how all of these interplay. But it's very hard.

It's very hard to make it a very quantitative analysis for what a layer one blockchain should be worth. And people have tried. I'm sure you guys have heard of the basically the application of the equation of the equation of exchange to two layer one blockchains, which is that equation mp equals to pq which is used to value forex foreign exchange currencies like one currency with respect to another you can think of ethereum as a kind of like a like a sovereign nation state that has its own currency namely ethereum and it has one export and that's decentralized computation and then you can apply the equation of exchange to try to value it in that way

and there's a term there which is the velocity of money like the more eath gets transacted then the less valuable it is because it's not being held So that's the intuition. But if you actually use that model to try to predict the price of need, you're going to be completely out. So we take all of those factors as inputs, and then we think about what the future might look like. If we believe that, and there's a full thesis that we have about how there will likely be a handful of winning smart contract architectures, there won't be 1, there won't be a handful, and there won't be 1, there will probably be a handful in the same way that there's a handful of winning processor architectures. and the winning ones will likely be,

if we're right about crypto and crypto will reach billions of users and will become this major phenomenon in a whole world that's akin to Web 2 will be built on top of Web 3, then we strongly believe that each one of these winning smart contract architectures will be worth many trillion dollars. So that's kind of how we think about it, and we just try to imagine what the future looks like and then backtrack from there

to try to then make a make a bit more. argument about what the token should be worth today and then make investment decisions in that way. So one curiosity is, so it makes a lot of sense, right? If you have a wallet, value as a traditional business, okay, present value of future, discount, present value of future cash knows. Easy. When it's a protocol, when it's a protocol like univiswap, you can still, still view it.

that way the unique token may be available that way and you can have some kind of rational model for it but it comes into each ethereum you can have a rational valuation for it but then there's some kind of premium that needs to be added because the market is attaching some kind of monetary premium and if you ignore it it's going to be hard to justify the values at which each is practically trading in the market. And then when we get to Bitcoin, well,

you really can't see any good revenue stream at all. And then it is entirely some kind of monetary premium which is really hard to approximate. Do I get that right? And then, so my question would be, do you think

other N-1s, L-1 tokens, we also have monetary premium in the future, or is the market kind of already won by Bitcoin in an sir? It's a good question. I think that to touch along this on our thesis for why there will likely be a handful of winning architectures, the thinking is that no one blockchain can fully cover the trade-off space.

there are many different things that you can be optimizing for. You guys may be familiar with the scalability trilama. There are a number of different optimizations that you can try to make, and it's very hard for a single architecture to just be the God blockchain that basically takes the whole thing. And this is similar, by the way, to just traditional computing, where you have CPU architectures that are intended for general purpose computation, like X-64 and ARM and things like that,

which, by the way, themselves have big differences. between them. But you also have different kinds of architectures that are optimized for completely different things like GPUs that are optimized for parallel compute and optimized for gaming. And now they can turn out actually work quite well for West She Learning and AI. And so similarly, we believe that there will be a handful of different kinds of architectures for smart contract platforms that collectively cover the entire trade-off space. But no one architecture will will just win everything. And I think that that's maybe the reason for why there won't be one.

And now I also have to argue, I have to make the argument for why there won't be a thousand of them. And the reason there is that there are very strong in our effects. And so if you do have two blockchains that are fairly similar and not differentiated with respect to one another, I think that the equilibrium state is for one of them to ultimately win. And that's because of the self-reforcing feedback loops involved. The fact that there are network effects at every level, there are network effects at the security level, because as a validator,

you want to validate on the blockchain that's most secure, and the blockchain that will give you the biggest reward, which is the blockchain that has the highest token value. And so that results in a self-reinforcing feedback loop that causes one blockchain to win. There's also network effects at the application level, where because of composability, you want to build your application on the blockchain

that has all of the other applications, because you can use those. other applications is building blocks. This is the whole notion of like Lego building blocks in crypto. The fact that there is no platform risk allows you to leverage other smart contracts that are also running on the same blockchain as inputs to your own application and you don't have to necessarily reinvent the wheel, which is another reason you you're more likely to build on top of Ethereum than you are on top of a random new blockchain that has nothing on it. And then the same applies at the integrations level.

Like if a blockchain is already integrated into all of the exchanges and all of the wallets and all of the interfaces and clients, a new blockchain has a hard time displacing all of that and competing against all of that. So the network effects, I think, reduce the possibility that there may be many different blockchains that are all identical to one another. So that, I think, precludes the possibility that there will be a thousand blockchains. And so as a result, I think my strong belief is we'll probably end up with something like five. I don't know, like some handful of different kinds of architectures, all of which will likely be very big.

It's a small number that support all of the decentralized competition in the world. And I think as a result, it's likely that token for each one of them will have all of the same factors that drive value for eat, right, including the staking, including the fact that there is a monetary premium because people use it to pay for things and those ecosystems. But take all this with a grain of salt, you know, if you'll see, it's very hard to predict the future. This is just one way to look at it. In your experience investing in some of these protocols, I'm sure like A16C comes across the case where this split of revenue may not be clear at the time of investment.

What I might mean is like maybe I can only take a practical example. like the uniswap case. So where you have uniswap labs who build the protocol and for a long while did not actually want to launch a token and then out of the competitive dynamics with Sushi Swap,

they were forced to, their hand was almost forced to go into the direction of launching a token. And when I personally look at uniswap like rational valuation of the uni token is made for me hard by the fact that how does the revenue get split between these two these two entities there are many of these protocols in which that is like rational evaluation is made hard by in fact that there's a foundation and there's a there's a community treasury and the dynamics of that may not be to year up front

Now, for me, as an external market participants, the only thing that is available is the unit. But for A16Z, in such cases, you would have the option to invest in either the public token or the founding team. do you have any kind of general guidelines for when to prefer one over the other or how to think about the state of space? Well, we primarily invest very early stage and at that stage, it's often too hard to know whether the value will ultimately be captured by some token that the team decides to build or whether it will be captured by a company, a traditional company that company builds, the team builds on. top of some protocol that maybe is just open source. And so as a result, the structure that we've landed on is one where everyone involved that includes the team and it also includes investors like us should get both things. They should get both an equity stake in the company.

In the case that the team decides that that's the path that they will take, maybe they decide not to even launch a token. Maybe the token is the wrong thing, and they just want to build a traditional business. And it's too hard to know that, a priori. So in the case that that happens, you want everyone to have equity. Or, and then also, I think in the end, it is actually a token thing. And the token is what captures the most value. You also want everyone to have the token.

And the reason this is important that's just for alignment reasons. It just wants everyone to be on the same boat. No matter what happens, if the company succeeds, everybody succeeds. And this also, by the way, maximizes flexibility. It doesn't constrain the founder to absolutely necessarily launch a token because the investors expect that. Or to just build a revenue earning business, again, because the investors expect that. As investors, we are on board. We're on your team no matter what.

And everyone is aligned. And we found that that is actually the best structure. And we don't try to predict a priori, you know, where the value is. going to be captured because that that can impose constraints on the founding team that you don't want to impose that early. Yeah, that's that's really interesting. And, you know, I think, you know, for us, as we think about how we want to construct our portfolio is also, this also aligns with what, what we look for when investing. And, and certainly, you know, there has become, I think, a sort of standard that most teams raising in crypto will, you know, will raise on a sale of future equity

agreement and and then a token warrant or something like that. So it is quite commonplace to have that sort of agreement with teams that are raising. Yeah, maybe just before we wrap up here, one final thing I wanted to touch on, I mean, taking a step back and looking at the events of the last couple of weeks is, you know,

how has the firm, you know, change or say, how does the firm advise companies and teams in your portfolio to protect themselves against, you know, a potential banking crisis or the shutdown of, you know, perhaps like the bank that they're working with or, or more broadly, you know, an economic crisis that might, you know, have their treasury be wiped out? Or to some extent, you know, even regular. regulatory risks, like how do you approach these sort of, you know, risks that are inherently tied to the system in which crypto operates in order for teams to be better prepared against these types of events?

Yeah, it's a really good question, and it's a harder one. I don't think that there's a truly great answer. And by the way, I think that everything that's been going on with a banking crisis is actually just a reflection of this. failures of centralized finance. So in a way, like, it kind of highlights the way, the reasons that you actually, you actually decryptial. And being able to have the transparency such that you can tell, you know, whether an organization or a protocol is solvent or not is some of the essential, is part of the essential ethos of the space. But given that we are still embedded

in the traditional financial system and our company still leave bank accounts, In our advice, really just comes down to diversification. You want to not be entirely concentrated with any one bank. You want to maybe split up the company's assets across three backs or so. Maybe you could hold some amount of capital on chain in the form of USC as well. It's another way of diversifying. But that's really ultimately the only answer. Barring, like most companies in the space at the moment cannot get bank accounts

with the big like G-Sib organizations, like the banks that are so big that they're too big to fail to fail because those banks don't want to bank crypto companies. And so then as a result, the only other option is to maybe get three bank accounts with other more regional banks

that are more friendly towards crypto as a way of minimizing the risk that any one of them might fail. Yeah, I think that's sound advice and certainly it's what I've encountered so far in the last few weeks as the dust is settled and people are starting to reason about how they want to reduce that risk. And certainly for us, you know, we were, we were about to start working with one of the banks that was shut down and now are thinking, okay, well, I mean,

we probably need to engage with several banks. Exactly. As we set up our structure, so that makes a lot sense. Yeah, this has been really fascinating. I really enjoy this conversation and all the insights you provided here. And so we'd like to leave our listeners with one following work. Where can people find you or, you know, some of the things that you mentioned, the work that A16Z is doing, the research, etc. Yeah, absolutely. So we are A16Z Crypto.com for all of the 86NZ crypto-specific content. And that's where we all are. And that's where you can find us. And then personally, I'm on Twitter at Alive underscore Eith. Great. Thanks. Well, Ali, thanks for coming on the podcast.

Thank you so much, guys. This was so fun. Really appreciate it. Thank you for joining us on this week's episode. We release new episodes every week. You can find and subscribe to the show on iTunes, Spotify, YouTube, SoundCloud, or wherever you listen to podcasts. And if you have a Google Home or Alexa device, you can tell it to listen to the latest episode of the Epicenter podcast. Go to epicenter.tv slash subscribe for a full list of places where you can watch and listen. And while you're there, be sure to sign up for the newsletter, so you get new episodes in your inbox as they're released. If you want to interact with us, guests or other podcast listeners,

you can follow us on Twitter. And please leave us a review on iTunes. It helps people find the show, and we're always happy to read them. So thanks so much, and we look forward to being back next week.