# Crypto's Pragmatic Shift: Infrastructure, AI Convergence, and Institutional Adoption

**Podcast:** a16z Podcast
**Published:** 2026-05-07

## Transcript

The most successful founders in this next era are going to be the ones that are much more product focused, much more go-to-market focused, and also more pragmatic rather than ideological.
There's a strong sense that in order for crypto to succeed, it has to work with the system as opposed to trying to overthrow it.
What we found in crypto is that finances, for a bunch of reasons, are kind of a little hanging fruit.
One kind of mental model that I have is we should try to get a billion people into sort of being almost daily users of blockchains through stocks and bonds and stable coins and payments and remit.
What does it take for a new technology to move from ideology to infrastructure?
For years, crypto was framed as a revolution, a way to rebuild financial systems from the ground up.
But today, the trajectory looks different.
The most meaningful progress is happening where crypto meets the real world, payments, markets, and financial infrastructure that people will actually use.
At the same time, new forces are emerging.
Regulatory clarity is opening the door for builders, while AI is reshaping how software and even economic activity itself gets created and executed.
The question now is not whether crypto can replace the system, but how it integrates with it and what that unlocks.
Here, the A16Z crypto partners discuss the launch of Fun5 and what comes next.
All right.
So, hey, everybody.
I'm here with all of the GPs at A16Z Crypto.
I think this is our first all-GP episode.
We've got some familiar faces, founder and managing partner Chris Dixon, Ali Yaya, general partner who's been here since the start, as well as Eddie Lazarin, chief technology officer, recently promoted to general partner.
And Guy Ouellette, also elevated to the position of general partner.
So we've got all the GPs here, and we are announcing a new fundraise, Crypto Fund 5.
That is the premise for today's talk, and I'd love to get into it.
Maybe we could start at why we are raising now.
Chris, you want to take this one?
Yeah, I mean, so as folks may know, we've had our crypto fund now since 2018.
I've been involved since, I guess, formally since 2013 with the investment in Coinbase.
We're at an interesting point now in the evolution of crypto where on the one hand, on the negative side, market prices are down.
Some of the sentiment is negative.
Some categories that are sort of non-financial have not kind of panned out the way that we've hoped and some have hoped.
But on the positive side, there's a lot of good things happening.
So most notably, we're seeing real kind of mainstream traction.
Stable coins have been growing.
There's about 300 billion in issued stable coins, the volume of transactions, rivals, large payment networks like Visa.
Importantly, that growth is not correlated with trading volume.
If you look at the kind of growth curve, it looks much more like the growth of a computing network or internet network.
So it seems like a really healthy trend.
That was catalyzed in large part by legislation passed by Congress last year, the Genius Act, which provided a regulatory framework around stablecoins, which I think a regulation is doing two things.
One, providing a pathway for good actors, for builders, so they know what the rules are, which is very important, having that kind of clarity.
But secondly, providing protections and guardrails for consumers, so they know that when you buy a...
genius certified stable coin that if you put a dollar in, there is a dollar in the bank corresponding to that stable coin and that the issuer is audited and has no appropriate safeguards around it, which is extremely important for both protecting consumers, but also building trust in the market.
For those who recall a few years ago when FTX crashed, there was another big scandal called Terra Luna, which was a stable coin that crashed.
So, you know, we have evidence historically of the danger in unregulated markets.
And the word stablecoin there doing a lot of work.
Yeah, the stablecoin was in quotes in the past.
Now it's a, you know, government certified term.
The bill was passed last year and we immediately saw an increase in entrepreneurial energy and, you know, new founders coming in with new ideas around it.
Because if you're a founder...
Do you want to risk entering an uncertain regulatory market and all of the risks involved with that?
You don't.
You go to do AI or you do something else.
So stablecoins provides this well-understood, well-regulated kind of area with real use cases.
And what are those use cases, you might ask?
So, for example, firms like Stripe have leaned heavily into stablecoins and it's taken their coverage from dozens of countries to over 100 countries overnight by enabling stablecoins.
Stablecoins are lower fee, you know, so you're moving bits around, right?
If you think about sending an email and just all the other things you do on the internet, it doesn't cost that much money to move bits around.
Amends in the U.S., for example, are 2.5% typically to make a payment.
Remittances can be significantly higher.
Really, we don't have a global financial network today.
We have a patchwork of lots of small networks broken down by country, banks.
It's a patchwork system glued together by humans and legacy processes.
And before you had WhatsApp, you had SMS and you had a patchwork of these SMS networks and you had to pay all these high fees and interoperability was sometimes a challenge.
And you had new services like WhatsApp that came over the top and built a new kind of modern digital network on top.
And that's kind of how you think about stable points.
It's a global network from day one built on top.
So an obvious thing that's happening now is people are building lending markets on top of stablecoins.
Once you have money, you can have verbs, things you do with money.
So lending is an obvious one.
Yeah, the second area that we're seeing a lot of mainstream adoption is in financial markets.
So people may have heard of perps.
It was originally a crypto kind of concept.
It's become very popular as a way to get exposure to various equities and other kinds of assets.
What we're also seeing is a lot of interest from traditional financial institutions, Wall Street.
Almost on a daily or weekly basis, you'll see announcements where firms are talking about tokenizing stocks and bonds and basically taking these financial instruments, which are, you know, huge markets.
And we're seeing a lot of interest and a lot of traction of these financial firms moving these onto blockchains and modernizing that infrastructure.
So that's on the positive side.
And then the other positive thing I mentioned earlier, Genius, is I think of stablecoins as roughly 10% of the crypto world.
And now we have regulation there.
And I think we're close to getting regulation for the other 90%.
There's a bill pending in Congress and the Senate called the Clarity Act, which would be a comprehensive framework for that 90% that may or may not pass this year.
If it doesn't pass, I think we'll see regulation come out of agencies like the SEC and CFTC, which will do similar things.
And I think the hope and our expectation is that that will...
do for the rest of the crypto market and all the other kinds of use cases that we've talked about, that I wrote a book about, that we've blogged about for years.
For example, network tokens, which are tokens like Bitcoin and Ethereum and DeFi tokens and so forth.
Clarity would provide a regulatory framework for those and provide a clear path for builders.
And I think one of the reasons that crypto kind of has a bad name in the world.
today in the broader world, and some people have soured on it, is that there have been a lot of scams and bad actors involved.
And another benefit of regulation is it will mostly eliminate that.
Our experience is new technologies like crypto go through cycles, and we're clearly in sort of a down part of the cycle from a price point of view and a sentiment point of view, but we think a lot of the fundamentals are strong and position the market to come back strongly over the next few years.
The way we look at venture capital is the combination of strong fundamentals and a lot of other investors having moved over to other areas, in my experience, has been a good environment for investing in a new fund.
So we're excited to have dry powder now.
So you're making the case that both the environment is ripe for it.
We have regulatory certainty and clarity like we've never had before.
On the one hand, you also have interest from major institutions and Wall Street.
And then on the other hand, from a product perspective, the things are actually working.
We're seeing traction for things like perps, bond chain finance, stable coins, and other such products.
And then to your other point.
The distraction of other tech trends, perhaps like AI, which you referenced earlier, like why would a founder not go work in AI?
Why should they come work in crypto?
You know, that's something that probably a lot of entrepreneurs out there are thinking about.
Where should I make my mark?
Where can I make a lasting impact?
By the way, I don't think it's either or.
I think there's a lot of overlap in crypto and AI and we've made investments there.
And we're also very excited about AI and generally and think it's a really important technology in the firm's, you know.
heavily investing in it.
But we, you know, we think that there's opportunity here too and that some entrepreneurs, you know, who might today be thinking that crypto is not an option should take another look now that we're getting regulatory clarity.
So Ali, you were the first full-time hire on the crypto investing team back in 2017.
You've been here since the start of things, since the launch of the first crypto fund in 2018.
How have things changed since then?
And by the way, your background also is in AI.
You worked at Google Brain.
So maybe you can add some color to this intersection and convergence of the two big tech trends right now.
Yeah, happy to.
So I think there have been many changes, but probably the biggest change has been cultural.
I actually like to remind people that 2017 was still extremely early in the space.
2017 was when I joined the firm and shortly after in 2018, we raised CryptoFundOne.
formally started the crypto vertical.
People think that crypto started in 2009 with the launch of Bitcoin, but the real start of crypto as a developer ecosystem and as a startup ecosystem was really 2015 with the launch of Ethereum, which, because of the fact that it was programmable, really opened the aperture of the kinds of things that you could really build in the space.
But Ethereum, given that it was a direct descendant of Bitcoin, shared a lot of the cultural values from the kind of the Bitcoin era.
So the vibe back then, back when I joined, was that crypto was going to be a revolutionary movement, primarily.
It was this very cypherpunk and anarchic movement.
And the vibe was that code as law was better than government law, that crypto systems are better categorically than trad systems, and that we would ultimately build some kind of parallel financial system that would wholly replace the existing one.
I mean, shortly after I joined, we had the ICO bubble in 2017.
And then since then, a lot has changed.
On the infrastructure side, we went from the 14 transactions per second of Ethereum to tens of thousands of transactions in modern blockchains today.
And the ability to send any amount of money from anywhere to anywhere in under a second for less than a penny.
And then we've also seen great innovations like on-chain marketplaces, things like lending protocols, and very importantly, stablecoins.
And finally, we've had the regulatory clarity that Chris was talking about.
So since then, the vibe has shifted dramatically.
And standing here right now, a decade later, there's a strong sense that in order for crypto to succeed, it has to work with the system as opposed to trying to overthrow it and working against it.
And it's also, I think, much more fundamentals based with a more significant emphasis on solving real.
world problems as opposed to building infrastructure or solving technical problems in a vacuum.
And we, by the way, we see this with the founders that we meet every day.
Our sense is that the most successful founders in this next era are going to be the ones that are much more product focused, much more go-to-market focused, and also more pragmatic rather than ideological.
So it was once a revolution, but now it seems like there are compromises being made or maybe reality has set in and people are realizing that, you know, you can't just completely do away with the old and in with the new.
There has to be some negotiation here.
And one way to frame it is that it's not either or, it's and.
Both of these things can coexist and you don't have to overthrow the entire pre-existing system in order for the new system to succeed.
Yeah, Guy, I've heard you talk about this and you describe it as crypto entering its collared shirt era.
I'll note that there are two collared shirts on this call.
Still some tees.
I do see Eddie in a blazer, which is pretty unusual.
I don't know that I've seen that before.
So kudos to you, Eddie, stepping it up.
But Guy, maybe you can tell us about this collared shirt era.
What do you mean by that?
Yeah, I mean, I really like Ali's revolution analogy, too.
And that's something we've talked a bit about.
You can win the revolution and then you have to figure out how to govern.
And I feel like crypto has won the revolution and had a couple of years of trying to do the Articles of Confederation and realized maybe that that wasn't the perfect system and is kind of going through the process of maybe trying to write the Constitution and have something more enduring.
I will say personally, I was definitely one of the cypherpunks that Ali mentioned, who was very ideologically motivated and inspired by the beginnings of crypto.
And I think both seeing that that...
does not lead to the sort of commercial success that we strive for and our founders strive for.
And also probably doesn't lead to the results that you want at like a societal or an ideological level.
That has certainly been beaten out of me.
And I think the space as a whole.
has gone from, you know, we're writing smart contracts in mom's basement with our hoodie and our fuck you flip-flops to we're now putting on a college shirt and a suit and tie and taking meetings with major banks who are seriously considering replacing their, you know, backends and core ledgers with blockchains.
And I treat that as an incredible improvement and a sign that, you know, what we've been working for for many years is finally working, not a capitulation the way I think some other people in this space maybe would view it as.
Yeah, what would you say to people who see this with a tinge of wistfulness, who kind of look upon this as that the movement has lost its spirit, its ethos, its founding core principles?
Well, I think you can let perfect be the enemy of good.
Certainly, it's fun to strive against some great challenge.
And there's kind of a narrative component to that where...
The struggle is maybe more fun than winning.
It's usually the process that's more fulfilling than the ultimate end goal.
And so I think maybe now we are looking for a new goal, having really won the process.
I would also, I think, compare this to open source, which started similarly as an ideological movement.
And I think some people were upset that that may be calcified in the form of GitHub owned by Microsoft.
And you would have said that maybe that didn't succeed on all of the original ethos.
On the other hand, code is by default open today.
And I think as Chris has said many times, the composability of open source software is one of the reasons that software companies and the industry as a whole has been so successful.
And so I think there's a big difference between the theoretical or the intellectual and the practical.
And it's a wonderful time to be a pragmatist building on chain.
Eddie, I'd love to get your view too, as somebody who's been involved in this since the very earliest days, even before you joined this firm, what you think about the evolution of the industry.
Yeah, I don't know.
I get and I respect the tone here a lot and I agree deeply, but I would reframe it a little bit as an extension of what's possible as opposed to a capitulation.
Like in the last couple of weeks, I had AI write like a command line tool so I could control my Zcash wallet and I literally just zipped Zcash straight into my Coinbase account, right?
Like in literally one shot on a CLI.
That is simultaneously the most cypherpunk thing I've ever seen in my life and also compatible with the trad system.
Literally money, like anonymous, programmable money that was, you know, directly under my control.
And then boom, right, like basically connected to my bank account.
That's what I think the future looks a little bit more like.
Obviously, you need to round out the edges a little bit to make different parts of the system compatible with each other.
But I really believe that the set of...
What is most helpful to people directly is the thing that is getting the most attention.
I think a lot of the pressure on the space has been to deliver concrete value immediately for as many people and institutions as possible.
And that obviously requires rounding out the edges.
So I see it more like that.
I see it as a practical turn.
Things that we believed were possible have not become impossible.
Just the focus has shifted.
What's personally getting you most excited?
today.
The truth is, I mean, so many things.
I think everybody in tech is pretty euphoric.
You know, I've been enjoying my weekend AI psychosis like everyone else is.
I love the feeling of it and letting the insanity wash over me.
I mean, I just gave you that example.
Like, it's kind of uncanny to be able to sit down and say like, OK, here's like four different really high quality crypto APIs that require calling smart contracts and signing some transactions, like stuff that would have given me pause in the past.
And now, With the kind of, I babysit a terminal for a couple hours and kind of have a conversation.
And now I have like working code.
Things that would have required me to like stop working and like really like dedicate myself to like recreational coding.
How it just happens in the background.
And it's kind of like a light managerial conversation.
It's incredible, right?
That has me excited.
While we talk about like the transformation of themes, like what was a big theme?
Over the last five years in crypto, programmable money, right?
Programmable money.
Are we giving up on programmable money?
Honestly, I don't think so.
Like now we have programs that you can write with just a couple of words and a conversation with a bot.
If you connect that with programmable money, now we have money that moves as quickly as you can talk to it.
You know, that's insane.
So that has me incredibly excited.
I see their range as we make money.
more amenable to direct software control, which is basically what crypto is doing.
We make it more amenable to human control because AI gives us direct control over software.
So, I mean, that's it in a nutshell.
I'm incredibly excited.
That's a great answer.
Guy, you've been spending a lot of time in the on-chain finance world.
Maybe you could tell us what you're seeing, what is getting you most excited about all the new projects that are popping up.
Yes.
If you look at the incredible growth in the outstanding issuance of stablecoins.
I think this, as Chris was alluding to, requires that you have a whole new ecosystem for capital formation.
You have stablecoins looking for new higher yield opportunities.
You have stablecoins looking for productive working capital.
And then I think you also have a lot of more traditional players in the lending and credit space seeing the opportunities for efficiency that the blockchains give you.
We also are at a very interesting point in time in how lending is happening generally.
After the great financial crisis, a lot of lending shifted and changed from primarily bank lending to non-bank lending, where banks would lend to private credit funds, would have an equity buffer and be the first lost capital, and then those private credit funds would ultimately lend to businesses, and those businesses lend to end consumers.
And private credit is...
You know, both this change in intermediation, but also a change in duration where they are intentionally making long term loans.
And that has had some interesting consequences the last year.
I don't know if you've seen any of the double pledging questions or some of the redemption questions.
But I think both because there are so many stable coins on chain looking for interesting credit opportunities and because of some of the shifts in the credit market outside of crypto, it's become a really.
convincing and interesting time to try to build new lending products on chain.
For people who don't know what's going on there, these double pledging incidents, maybe you can give some explication.
Yes.
So if you want to lend against an asset as a lender, there's a whole process where you go through UCC filings and they call it perfection, where you can say, I own this asset and I'm pledging this asset as collateral against a specific loan.
It's obviously a challenge to make sure that a specific asset has not been pledged multiple times.
And there are a number of, you know.
I would say, good ways that that happens in the existing world.
And then I think more as a result of duration mismatch than, you know, these double pledging instances, there have been some significant redemptions from or requested redemptions, let's say, from private credit funds.
And, you know, that's a...
A different topic more broadly, you know, we're very focused on on chain and in the crypto products.
But I think there are a number of secular trends that are pushing people towards a real interest in credit on chain.
And that has had a lot of positive knock on effects just in terms of the quality of talent and founder that we see coming to crypto and the willingness of traditional players to adopt on chain solutions.
So that that has given me a lot of hope and is something I'm very optimistic about.
I think the other.
component of on-chain finance that we've seen really start to work in the last year or two has been about building new forms of markets.
Elise said many times, and I think it's a great comment, that crypto is really a coordination technology.
And I think blockchains have proven to be incredibly good at building new markets.
So Chris was alluding a little bit to how popular perps have become as a product.
We've seen perps move from being primarily focused on crypto tokens specifically to applying across all sorts of traditional assets like equities and commodities and FX.
And I think it's very exciting for me personally that the products we as an industry have built over the last five years that I think are genuinely better products do not have an asset question anymore.
They can operate on top of high quality traditional assets as opposed to being specifically applicable for network tokens.
And then we've also started to see new markets that are not well served in the traditional financial world.
I'm specifically thinking about compute in the forms of things like GPUs or data center build out.
And then about things like electricity, either in the form of solar panels and batteries, or even seeing, you know, oil price discovery happen on hyperliquid, arguably a couple of weeks ago, you know, starting to see new markets being built by default on chain.
I think it will take obviously time for traditional assets and markets to move on chain, but much in the way we kind of hit an inflection point with open source where if you were building a new project, you would just make it open source by default.
I think we're starting to hit an inflection point where if you're trying to build a new market or exchange, your default is to build it on chain.
And I think that's.
underestimated in how powerful it will be moving forward, not just for, you know, those specific products, but the ripple effect across blockchains generally.
And just succinctly, you know, to give people a sense of what is the value of some of this on-chain activity when it comes to traditional finance, for instance, you know, what value are people seeing that suddenly the light bulb has gone off and they're like, oh, we should use this technology.
It can actually help us and make things better.
In what ways does it provide value?
I would say the specific things that traditional financial players key in on would be latency, capital mobility, the fact that these markets are in almost every case open 24-7, as opposed to having specific trading timelines.
And then, you know, we in crypto have called it decentralization for many years.
But if you rephrase what decentralization actually is.
specifying in a very concrete way the trust assumptions you have when you send a transaction, that would just be counterparty risk.
And I think, you know, to our point a few minutes ago about double pledging and credit, I think people in traditional financial markets have an incredible sense for how important counterparty risk is.
And, you know, counterparty risk would be the...
Financial terminology for platform risk that I think is salient and top of mind for any founder in tech.
When you think about how important that was in Web2 and certainly when most of the world will likely be rebuilt on top of AI models, how important platform risk will be there.
So I think this is not just a financial question, but a broad societal one that crypto has built, I think, the best answer to.
Counterparty risk management doesn't have quite the same ring to it as decentralization.
But that doesn't diminish its significance.
So we've been talking a lot about these financial uses and the value that you can get from crypto in the sphere of finance in particular.
Chris, in your book, Read Right Own, you laid out this vision of blockchains, a grand vision of how they could provide a foundation for a more open, fair, and accessible internet.
What do you think about the way things have shaped up?
currently, where much of what we're seeing tends to be highly financial or financialized.
Yeah, I think, you know, my experience with these kind of technology rollouts is they roll out in different use cases over time.
So, for example, in AI now, you know, AI is general purpose, but we're seeing sort of the killer app is encoding.
And, you know, I expect with AI that we'll see sort of many other things, many other killer apps over the years.
I think what we found in crypto is that finance is, for a bunch of reasons, is sort of, I think, kind of a low-hanging fruit.
One reason is, you know, in the U.S., we have a strong financial system, but around the world, it can be much weaker, including just basic things like, you know, savings and payments.
And so, you know, in some ways, the bar is much lower in finance, whereas you take social networking or other kinds of things, you have very sophisticated global systems already.
I think it could take time.
You know, one kind of mental model that I have is, You know, we should try to get a billion people into sort of being daily or, you know, almost daily users of blockchains through stocks and bonds and stable coins and payments and remittances and so forth.
You know, once you have people onboarded, they've used the infrastructure, the wallets and so forth.
It's natural at that point to offer adjacent services.
So, you know, that's kind of how I think things are evolving and how I expect them to evolve.
So I think this next kind of era, next few years.
We're open-minded and, you know, we invest in people and great entrepreneurs.
So it'll be up to the entrepreneurs.
But my expectation would be over the next few years, most of the use cases will be our financial.
And in your post, The Long Game for Crypto, you've written that finance is not separate from the larger vision.
It is the foundation.
So that goes to your point about how maybe you need to get everybody on chain first.
So we touched on this a little bit, but obviously one of the most interesting trends right now is the convergence between AI and crypto.
Ali, given your long tenure and experience in AI as an AI researcher and now investing at this intersection, what are you finding most interesting?
What are the most productive areas where these two trends are overlapping?
Well, actually, let me give you a story first from the time that I was working at Google.
I first worked at Google X and then moved over to Google Brain.
When I was at X, I had already been following crypto for a very long time.
And I actually pitched someone very high up at X to start exploring crypto ideas as a part of X.
And this is X, the quote unquote moonshot factory, which is supposed to be maximally open to new ideas, maximally open to new potential technologies.
And this was 2016, 2017, at a point where it was already clear that there's a lot of excitement around crypto as a new technology.
And I was basically laughed out of the room.
So that was kind of my first experience trying to get a company of the size of Google to touch crypto, even with a thousand football.
But then I moved over to Google Brain and I was working on robotics and AI.
And at the point when I was moving over from Google Brain to join the firm, everyone on my team essentially tried to talk me out of doing this.
There was like this simmering revulsion towards the space.
There was this sense that I was throwing my career away and that it was going to be kind of a disaster.
And actually one person said one thing to me that really, like I remember it vividly because the stakes felt really high.
He said, you are joining a group of people who want to do nothing but trading turds.
And he was actually quoting Charlie Munger because Munger had said something along those lines.
And that was kind of the vibe.
And I think that this kind of resembles the vibe and the relationship between AI and crypto, at least back then.
I think that these are communities that for a long time have not really talked to each other.
And in many ways they are.
culturally diametrically opposed with AI being about centralizing top-down control of over say like compute data machine learning talent to build a kind of AI panopticon that can see everything it can learn everything and then can reason about everything whereas crypto is much more about empowering individuals on the fringes it's about breaking down power structures it's about enabling radically free markets that are global from day one and putting control on the hands of individuals as opposed to large big tech corporations.
And so these two things have been like oil and water for a long time.
But that is actually beginning to change because the two spaces are beginning to converge.
It turns out that the financial system that we have built so far outside of crypto was not built for AI agents.
And I strongly believe that in the near future, the majority of transactions that happen in the world are actually going to be done by AI agents as opposed to humans.
That number may quickly become something like 99% or 99.9%.
And it is very, very hard to imagine that all of that happens on ACH and Swift or even credit cards because stable coins are virtually free.
They are, as Eddie was saying, they're fully programmable, they're internet native, and they are the perfect way to convert AI agents from tools that humans use into economic actors that are first-class members of the financial system.
And one last point I'll make about that specifically is I was having this debate with someone on our team about Visa who was arguing that Visa is too entrenched and is going to be very difficult to disintermediate and compete against because everybody already has credit cards and it's too hard to get everybody to switch to a different kind of payment mechanism.
But agents just don't have that preference.
They actually don't have any preferences at all.
And Visa charges something like 16 bps on every single transaction that flows through their network.
So there's a very strong incentive for agents and merchants alike to completely cut out Visa whenever the transaction happens in a way that does not involve a human stepping into a store and providing something like a piece of plastic to make a payment.
Yeah, I mean, I couldn't agree more with Ali saying is like people are going to be a little bit surprised by how agents are going to be not very nice.
OK, like when I say not nice, I don't mean they're going to like do anything bad.
I just mean like if you're going to say, hey, save me a bunch of money on my monthly spending.
They're just going to do it and they're not going to care what they have to destroy in terms of software to achieve that goal.
That's a beautiful thing for consumers.
People are going to save a lot of money.
They're going to make things that are much more efficient.
And that sounds like a cute little consequence.
It sounds like I'm saying that your credit card bill or your monthly stable coin balance has to be smaller.
Right.
But it's actually much deeper than that because it ripples across the stack.
The agents are going to go and they're going to rewire everything for you.
If they're really empowered to do so, they're going to try to save every penny.
And they're also going to want to pay per use because signing up for like a monthly subscription or paying annualized in advance is a big commitment.
All these things kind of point at crypto systems.
Yeah, the other way in which these two worlds are intersecting is that crypto can actually be the technology that helps fight against deep fakes and sort of the overabundance of media that's completely AI generated and difficult to...
tell from actual sort of human or real media.
There's a company I think many people are familiar with called World that creates the ability to generate a proof of humanity so that on the internet you're able to know whether the people you're interacting with are actually humans or if they're agents.
And that's just one example of the ways that cryptography can stem some of the problems that AI will intrinsically create by virtue of making it virtually free to generate media that looks human and looks very real.
Ali, you've mentioned this idea that like maybe most of the transactions, like upwards of 90, 99% could be being done by agents in the near future.
Project for us out, you know, five years, like five, 10 years, I don't know how long is it, one year, some indeterminate amount in the future, what the internet looks like after it's undergone this transformation.
One very futuristic vision of the way that these two worlds collide is a world where You have agents that are fully empowered with crypto wallets such that they can pay for things, they can get paid for things, they could potentially even raise money.
They're also able to generate value by creating services, by writing software, by creating content, by essentially becoming an employee for another agent or for a human and doing all of this in a way that is almost entirely autonomous.
And so, for example, one interesting project is working precisely on this problem of building a harness that gives an agent the ability to pay for its own compute or whatever resources it needs to continue to exist.
And then on the other side, generate value as a way of continuing to exist as a self-sustaining entity.
So there is this somewhat sci-fi future available where many entities that are existing out in the world are fully autonomous and they're entirely run by AI agents.
It sounds insane.
It sounds like it's not something that could happen anytime soon.
But given the exponential curve on which AI is currently on, it is not a stretch to believe that in five years, the models will be so smart and so capable that actually going out and creating value in a capitalist society is something that they can very well do.
Are they going to be working for us or are we going to be working for them?
Maybe a combination of the two.
That's fair.
Guys, so you were out on the road for a while raising this fund, talking to LPs, hearing from people out in the field about...
you know, what they're interested in.
When you had your ear to the ground, what were you hearing from people?
What are they most interested in?
What resonated amid this fundraise?
I think to the topic we were just discussing, everyone is very focused on AI these days.
And the question is how to think about software other than AI in a world of AI.
I think a really interesting point or discussion there is that almost everything in crypto is building a network effects business.
And you cannot vibe code USDC.
You cannot.
have a weekend project and end up building something like a Hyperliquid.
And so I do think there's an interesting set of ways you would evaluate how to build a crypto project now or whether it's worth building a crypto project now.
And I feel like there's kind of a bifurcation of crypto as a very overlooked place to build that, you know, if you are a really talented founder would be the right place to look.
And I think certainly.
A number of the folks we talked with from the LP base mentioned this.
I also think we as venture capitalists talk a lot about getting to work with founders.
And I think that's absolutely one of the most fun or valuable parts of our job.
As an industry, we don't talk very much about, you know, the returns that we generate and how those go towards kind of the broader society at large.
And, you know, specifically thinking about crypto as a technology for markets.
we should be the users of our own crypto products.
And so, you know, I thought it was very heartening that a couple of folks had asked about, for example, funding capital calls and stable coins or starting to transition some of their own existing business on chain.
And, you know, I suspect that with each successive fund, the questions will change a little bit and will become more sophisticated.
And that certainly brought a smile to my face and brought me hope.
That's very cool.
Yeah, so you're actually seeing the products gain traction among the LP base.
You mentioned something which is kind of funny.
It's a little paradoxical, the fact that, you know, you can't vibe code a, you know, major financial application like a really highly regulated stablecoin or perps exchange, DEX.
And it's funny because, as Chris was saying earlier, some of these financial applications are in some sense lower hanging fruit just because crypto is kind of financial.
by the nature of it.
And yet if you really want to create a lasting product, you need to have these deep integrations and comply to all of the existing regulations, whatever.
So it's kind of funny that it's both easy to get up and started with financial stuff, but also it's extremely difficult and there's some moat around that for people who are able to do it effectively.
Yeah, I mean, I think if you wind the clock back five or six years, a lot of the questions or challenges in crypto were highly technical.
like the position of highest status in crypto might have been the researcher.
I think it's interesting to see that, you know, be the case today for AI.
I think what we most need in crypto is really strong go-to-market and is not a more intelligent or clever protocol or mechanism design, but is instead kind of the shoe leather of going and talking with all of the different participants in a network and convincing them to see the value, whether that's through incentive mechanisms or just the hard work of doing good go-to-market and BD.
And I suspect that that only becomes more true in a world where, much like Ali was talking about, agents are taking more actions on our behalf and, you know, intelligence becomes more and more of a commodity and less specific to any human individual.
I think, you know, the sorts of network effects businesses and coordination networks that crypto has specialized in building will be even more valuable on a relative basis and probably not something that you can build purely with intelligence, but that requires kind of agency.
I think it's interesting that You know, we call them AI agents because they inherently require a human to tell them what to do as opposed to, you know, taking actions on their own behalf.
I think the important point here is that the things we're building on chain today, I believe will be long term enduring, you know, networks that have broad societal impact.
Whereas if you're interested in building a short term project or pure software, I think there's the obvious question of why does that not get eclipsed by one of the large model companies in a couple of years?
All right.
So.
Guy, with what you've been laying out about the value of some of these technologies, you've kind of been getting at the point that a lot of it has to do with coordination, something that Ali brought up earlier.
Ali, you've also been interested in some other trends like privacy.
You recently, you know, put out a thesis saying that privacy will be not just a moat in crypto, but perhaps the only moat.
Tell us about that, because that sounds pretty important the way you phrased it there.
to say is that privacy is the most important feature that really has been an afterthought for most of the space up until very recently.
The reason for that is likely that people were solving some of the other harder technical problems or the initial technical problems that needed to be solved in order for us to just have blockchains that scale and privacy is the next thing to go after.
But the state of the world today is that most blockchains are almost entirely public and transparent, which means that every transaction Everything that happens on a blockchain is entirely visible to anyone who cares to look.
And the reality is that there is no way that blockchains and crypto will become mainstream if that is what the technology requires.
It's inconceivable to imagine that you would have your salary be entirely visible to everyone.
And if that's the case, then you wouldn't ever be paid in crypto because you wouldn't ever want that to be the case.
It's also impossible to imagine that a company's...
balance sheet and transactions would also be entirely public and transparent.
So privacy is essential in order for crypto to really break out and become mainstream, especially as it makes inroads into institutional use cases.
And I think that there's a second piece to this, which is that as interoperability between blockchains has become very seamless, it's starting to feel like block space and blockchains as businesses, as value capture mechanisms, aren't really...
all that defensible because it's very easy to move and migrate whatever state you have on one blockchain over to another.
And the reason for that is, again, because there is full transparency into every blockchain.
You can see the data, you can easily copy it, you can move it from one blockchain to another.
Privacy begins to change that because as soon as the data for an application or for a set of users who are using a particular system is encrypted, it is much harder to then migrate all of that state from one blockchain to another.
which creates a much stronger network effect around the blockchains that have privacy, simply because the switching costs goes up significantly and it becomes harder for people to migrate elsewhere.
So maybe the case in a world where it's very easy to fork a blockchain, it's very easy to migrate from one blockchain to another, where we have potentially a lot of fragmentation and lots of people starting infrastructure projects that compete in the realm of block space, that privacy ends up becoming the one thing that really preserves strong network effects.
It's still very early to see whether this holds, but I suspect that because of the fact that it dramatically increases switching costs, that'll be the case.
Yeah, and your point here is really interesting.
You're saying that infrastructure is getting, to a certain extent, commoditized, at least block spaces, because there's so much of it and it can handle tons of throughput.
It wasn't so long ago that that was totally not the case.
The pace of progress has actually been kind of bewildering.
This stuff has been rapidly advancing and progressing over just a few short years to the point where now you're saying that it's almost beside the point that there's enough chain space.
It's all of these other features where value shifts.
Well, it may be that in the long term, it doesn't get fully commoditized because of Jevon's paradox and induced demand.
As you were talking about before, if the number of transactions that happen in the world scales by 100x, 1000x, a millionx because of AI agents.
It could be that even though BlockSpace is as cheap as it is today, that you still need capacity for millions of transactions per second.
And there are still strong network effects around the blockchains that are the most secure, the ones where you have most of the users, the ones that have privacy, such that BlockSpace actually ends up not being commoditized and actually does trade at some premium and does offer opportunities for value capture for the underlying blockchains and tokens.
How far are we off from a point when Those problems are solved when suddenly we don't have to worry about our salary being exposed or whatever bills being transparent for anybody who wants to look.
I would say we're already there.
I mean, there are numerous different technologies for approaching the problem of privacy, which certain chains like Tempo, Arc, Canton are taking, where the way that privacy is provided is...
essentially by trusting a central party to keep transactions private and to only post commitments to the underlying chain.
And so that already works.
It's easy to build.
It has some trade-offs in that now the underlying infrastructure is less credibly neutral and it's less verifiable from the perspective of outsiders, but it fits nicely at one point of the trade-off curve that I think many institutions out there will need and it'll work for them.
And there's also, More in the middle of the spectrum, the use of trusted hardware, so trusted enclaves.
These are features on processors that allow you to execute transactions in a way that even the person who controls the physical machine can't really tamper with or access.
And then I think finally, at the end of the spectrum, there's the cryptographic approaches.
And zero-knowledge cryptography, which is the core technology that enables building a blockchain that is private entirely from a cryptographic standpoint, has improved.
something like 10 to 100x in the past 10 years.
And actually one of the major projects from our research team led by Justin Taylor, who's one of the research partners on our team, is called Jolt and is entirely focused on building efficient zero-knowledge cryptography to allow for both very scalable systems but also private systems.
Yeah, so you brought up the Jolt team internally led by Justin Taylor.
Zero-knowledge proofs have been around for decades.
Why are they suddenly so interesting?
One angle to this is that zero-knowledge cryptography has allowed us to solve the scalability trilemma, which historically has been the tension between decentralization, security, and scalability.
And the way that zero-knowledge, like ZK, proofs help us break out of that tension is that now it is possible for one node, one computer in a network to do a lot of work and to have everyone else in the network be able to verify that work without having to redo the work.
which was the state of the world in blockchains from before.
Up until now, most blockchains still have this property that every node in the system has to redo the work that every node is doing, which is one of the major bottlenecks to allowing blockchains to scale.
So with efficient zero-knowledge cryptography, you can break out of that and then have every node do work that is additive to the network, which in principle allows a blockchain to scale horizontally, such that the upper limit of how many transactions per second it can process is unbounded.
And instead of being capped as something like 14 transactions per second for Ethereum, you can go up literally to millions.
You know, we started this conversation talking about crypto evolving from the cypherpunk days to the present day.
But I'll say that sounds pretty cypherpunk to me.
You know, that opening up of access to anybody, the total democratization of participation, enabling cheap compute, and also the privacy guarantees that go along with it, all very cypherpunk to me.
Yeah, I totally agree.
I mean, I think those things are going to happen.
As Chris was saying, like the pace that things develop is unpredictable.
Different sectors of technology find product market fit among different sectors at different paces.
It's always hard to say.
I don't think anybody's ever thought that crypto was not going to be a financial.
That's right.
That's ridiculous.
It was from first days, I think.
The question is what happens first and because of what other technological developments, right?
So we're looking very much at AI because it obviously alters how software is made.
It alters much of the cost structure of software.
It alters the interoperability between different types of software.
And a lot of what we've been talking about, like potentially the commodification of block space, the increase in capacity.
different facets of privacy and the consequences on defensibility.
Like these are all consequences of changes in the technology and how they will affect the development of technology going forward.
This is the fundamentally the hard thing.
This is the thing that you need to keep your eye on the ball because it's changing all the time.
And this is the thing that we keep in our minds when we invest out of like our new fund.
I think the idea of cypherpunk is using cryptography to make social change and people individually.
saying that like I would like to have more privacy in my life or in my world.
I find it fascinating that actually the draw or the pull for privacy today is certainly, you know, the use case Eddie described with something like Zcash of wanting individual privacy.
But at scale is more coming from institutions who are saying, you know, we need strong privacy.
We need zero knowledge proofs.
Thinking about things like zero from layer zero or Providium from ZK Sync, like literally working with banks and large hedge funds who are the folks.
that are most loud about wanting privacy.
And so I think there's like a wonderful full circle aspect to that, that, you know, the interest in these areas maybe came out of an individual desire, but they're solving a real problem for institutions at scale.
Well, in that respect, it's very similar to earlier movements in computing like the open source movement, which began as an ideological attempt at social change.
And as you were talking about earlier, the fact that GitHub eventually got acquired by Microsoft Open source kind of made it.
It made it differently than anybody might have imagined that it was going to, but it has become pervasive and ubiquitous.
Yeah, I mean, this may be a more general comment outside of just crypto, but I think technology is the greatest driver of social change, like increase in living standards, increase in productivity, increase in time for leisure, increase in health.
We specifically as the crypto team, but A16Z as a whole could not be more.
optimistic and positive about technology as, you know, being the primary driver for human flourishing.
And so, you know, I think the specific way in which that manifests is very hard to predict decades in the future.
But that has always been true throughout human history.
And I am pretty confident in making that bet moving forward as well.
Indeed, we are all techno optimists here.
And yet there are problems that need solving.
One, Chris, that you've talked about is about the centralization of control and power and consolidation in the AI industry.
Maybe you could speak to some of the challenges that the world faces there and what crypto might be able to do about it.
For those who know the history of the internet, one of the exciting things about it in the beginning was that it was a decentralized network where, you know, anyone could put up a website, anyone could start a business.
Anyone could, you know, create a new software product and distribute it.
Over time, the internet has gotten far more consolidated.
There's a relatively small number of large services now that control the vast majority of the money and the traffic that flows through the internet.
You know, all signs point to AI sort of furthering that consolidation.
AI is very capital intensive.
There's, you know, maybe four or five U.S.
companies that are leading labs right now.
Probably very hard for new entrants to compete with that.
You know, the internet data continues to show that further consolidation, you know, even going to websites, you know, Stack Overflow was an investment I made years ago.
I was on the board for a long time.
They ended up getting acquired, but a lot of that data that was Stack Overflow was used for training data in AI, and now Stack Overflow traffic has dropped dramatically.
I think we're going to see that's sort of the canary in the coal mine.
We're going to see that across many other categories of the internet.
You know, I think that's a negative trend.
I think that we want to kind of keep the original vision of the internet, that we want to have a level playing field where anyone can, you know, two people in the garage can start an internet service that competes on a level playing field with bigger companies, that small businesses can flourish.
You know, consumers benefit from the fact that...
They can go direct to a business or another consumer and make a payment or do some other kind of activity, economic activity or other activity.
And, you know, I think the only kind of credible technology around right now that can counterbalance that consolidation is crypto.
I think we're starting to see that in financial services.
And, you know, and I hope that over the next decade or two that we'll see that in many other areas.
And you're speaking not just abstractly, but...
concretely and personally as an individual, you know, having founded several companies that were based on websites that you could just throw up and being a blogger and being able to have a global audience.
That's right.
Yeah, no, I've spent my whole career on the internet.
First as, you know, an internet entrepreneur and then as an investor and was attracted to it partly for this.
I came from sort of the open source.
and was attracted to that kind of ethos of decentralization on the internet.
I think a lot of people shared that enthusiasm that I had and I think still do.
And I think that at some point the pendulum will swing back and we're still relatively early in the development of the internet in the grand scheme of things.
And I think it's important to have tools that allow people to help nudge it back that way.
So I hope to see that over the next, the coming years.
Technology and business, it's all a process of bundling and unbundling.
And we've seen a lot of bundling lately.
Guy, maybe you could speak a little bit to some of the ways that crypto could address some of the centralization and consolidation challenges faced by the AI industry.
Sure.
I think Alou was talking about this a little bit earlier.
One of the most salient ones is just being able to individually, uniquely identify a human being.
which is deceptively hard.
And so, you know, building a proof of personhood so that we can all exist on the internet in a proper and identifiable way, I think is really interesting.
Probably the biggest bottleneck today for some of the AI companies is just the access to compute and data.
I'm very fond of the bitter lesson that...
The path to building a better model is just throwing more compute and data at something.
And the existing large model companies have a huge advantage in terms of the ability to raise money, in terms of their ability to form capital and form data.
And I think crypto is the only technology that has shown itself to be a good mechanism for coordination, for fundraising or crowdfunding that could rival the strength and power of some of the large labs today.
And we've seen a number of...
AI Project started to do exactly this, trying to crowdsource GPUs either for training or fine tuning.
A number of companies that are trying to allow people to submit their own data and then have an ownership in the eventual model that is trained.
I think there are a lot of technical challenges with this.
And probably the most successful attempts at building open source or decentralized AI models have come from distillation of the large existing models.
But I remain pretty hopeful about the idea that crypto can help be a coordination layer for individual participants to train or fine tune or inference their own models in the future.
And you see it as a way to overcome this bottleneck for compute capacity as well.
Is that right?
Yeah, I think like capital markets for compute.
will be built on chain.
I think it's, you know, always a challenge to take an existing network that, you know, sits offline and put it on the internet or put it on chain.
And I would be much more optimistic about the idea that we will have new, more important, more relevant markets that are, you know, built natively on chain.
I think the compute market is probably the most important market on earth.
Certainly if the quality of AI models continues to improve at the current pace and societal impacts of that are as broad as we think they could be.
then the GPU, the piece of thinking sand, is probably the most important asset in human history.
And the market for that today, I think, is just highly immature.
And the ability to, you know, own, access, or finance that as an individual participant, as opposed to one of a very small number of companies, I think is one of the most important forms of freedom.
And, you know, one of the most important markets to ensure is open into the future.
Hear, hear.
Eddie, what does Fund5 need to have funded for it to feel like it has worked out?
What would you like to see come into existence as a result of this financial vehicle?
I mean, so many things.
I think everybody in crypto that's been in crypto long enough has a stream of mainstream adoption.
I think something that just has very concrete mainstream benefits that many, many people can use.
I think the space has shifted around to thinking that...
This is something that's achieved indirectly, right, through institutions as infrastructure for other types of companies.
That may be a way to do it, you know, so it may be one of those kinds of mixed victories.
But I think in any case, what I really want to see is I want to see an intensification of the competitive dynamics in the market.
I want to see.
new ways to allow software to enable people to own things, to enable machines to own things, right?
Those are kind of the key themes I've always thought.
So if we can get to that destination with mainstream acceptance, that would be a clear victory to me.
Good answer.
Same question to all of you.
Ali, I'll hit you first.
What does Fund5 need to have backed for you to feel like it has worked out?
10 years from now, what I would really like to see is a billion or more people interacting with blockchains, either directly or indirectly, every day.
And in order for that to happen, we have to back collectively, both us and other people, need to back the protocols, the services, the tools that enable that.
And also to move the regulatory environment into a position where that's possible and that's legal.
So that's one.
I think the other one is just moving the majority of the world's finance on chain.
And maybe the third one would be succeeding at transforming AI agents from tools that we use to first-class economic actors.
That's a great set.
Guy, you're bringing a fresher perspective to this as a new GP.
What's your answer?
What'll make this fund a success in your review?
I think if crypto does nothing else, we will give every human being on Earth a dollar-denominated, stablecoin-powered neobanking account.
And what we take for granted.
The ability to save our dollars and to invest them as people living in the U.S.
and in the first world.
There are still billions of people that don't have basic savings infrastructure.
And I think that itself will be a huge improvement.
And there will be very many successful companies built either as a first or second order effect of that.
Then I also think if crypto can.
accelerate the timeline in which we as a species produce more energy and more compute and build more efficient and open markets for energy and compute, that will be an incredibly positive knock-on effect.
And so I think we've already made a number of investments in compute markets and energy markets, but I'm very hopeful that those will be not just interesting from a financial or product perspective, but have broad knock-on ripple effects that kind of accelerate technical progress as a whole.
And then so we got the fresh take, the new view.
Let's get the OG take from you, Chris.
You've been here for every fund, even predating all the crypto funds with your Coinbase investment back in 2013.
What would make Fund 5 a success for you?
Not financially, but perhaps philosophically?
I mean, I think Eddie kind of hit on it, which is we want to see real mainstream use cases.
Oh, you mentioned a billion users.
Like that should be the target, I think, for any significant internet technology.
I think most likely the next couple of years, as mentioned, will be in financial use cases.
So I think, you know, getting the regulatory clarity that I think we're getting soon and having, you know, really kind of world-class entrepreneurs come into the space and build financial services that really get us on a nice path to that first billion users would be the goal over the next, call it two, two and a half years.
And just to add on to that, you know, it's only been a couple of years since Read Write Own came out, but a lot has happened since.
If you were writing a last chapter for that book now, how would it end?
I guess I was writing it three or four years ago.
You know, ChatGPT had come out and I do talk a lot about AI in there.
I talk about how it will lead to consolidation, capital intensive, big companies.
I think that was mostly right.
You know, what I try to do in the book is really distill the essence of the technology.
And I think that is, you know, mostly a fixed thing that has to change.
So honestly, there isn't a whole lot that would update today.
And I think...
My kind of core belief is that every technology has kind of an essence, and you boil it down, and that technologies arrive in kind of deceptive ways.
So, you know, social networking started off people in San Francisco sharing what they have for lunch, and you could have, if you just looked at it superficially, you would have thought, okay, this is kind of a, you know, toy.
It's not a deep technology, but if you really kind of looked through the trappings and looked at the essence of it, what it really was was a way for, you know, anyone on earth to communicate with anybody else.
And why wouldn't that be used for culture, business, politics, and so forth?
I kind of think of it as you can boil down the technology of its essence.
You can, in some ways, predict this long-term trajectory.
So that's what I try and do and read right on.
I think I basically stand by it today as sort of, you know, boiling down the essence.
And so I think over time, things will bend towards that, towards, you know, the key features, the technology, the strengths, the technology, the problems that they can potentially address.
Those things stay constant.
Obviously, the particular apps and people and interactions with other technologies, that changes and it is virtually impossible to predict.
But that core of the essence is, you know, it took me a long time to develop that and I think it's still today.
Fantastic.
All right, everybody, all the GPs, thank you so much for all your time.
Chris, Eddie, Ali, Guy, this was a great conversation.
Thank you, Robert.
Thank you, Robert.
Thank you.
You guys.
Thanks for listening to this episode of the A16Z podcast.
If you liked this episode, be sure to like, comment, subscribe, leave us a rating or review and share it with your friends and family.
For more episodes, go to YouTube, Apple Podcasts and Spotify.
Follow us on X at A16Z and subscribe to our sub stack at A16Z.substack.com.
Thanks again for listening and I'll see you in the next episode.
As a reminder, the content here is for informational purposes only.
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.