Has AI Revived the DePIN Sector? | Dmitriy Berenzon

Transcript

I think we've seen that with compute, you know, on on the virtual resources side where I do think the why now is AI demand, you know, and and and the demand for both uh inference and and and training. And um in that context, compute is like electricity in the sense that no one ever says, "Hey, we have enough electricity." It's like the more you have, the more you can do with it. And I think that's been a structural tailwind that I think is making viable a lot of these decentralized compete plays today where 2, three, four, five years ago, you know, pre-transformer, it just wasn't that significant of a need. Quick shout out

to today's sponsor, Echodin. We'll hear more about their decentralized club on AppOs later today, but for now, let's get into the episode. What's up, guys? This is Bokio, the bad host, and the better host, Danny. Uh, today we have a very, very important announcement for you. We are hosting Permissionless in Brooklyn. Um, what dates? It's going to be in June, June 24th through the 26th. Very good. Um, you can use code 0x10 to buy your tickets. Uh, we're going to do lots and lots of stuff. um dinners, games, you'll get to meet the whole crew, the OX research crew.

Hopefully a bunch of the Block Works research and other Blockworks crew and uh maybe a dunk tank. We're going to have fun. Uh whatever we can smooch off of Yano and Meo, we're going to do it. So, if you are a Zerx Research listener, make sure to use code 0x10 to buy your tickets. Danny, any parting words? And we'll see you there. Nothing said on Zerox research is a recommendation to buy or sell securities or tokens. This podcast is forformational purposes only and any views expressed by anyone on the show are solely our opinions, not financial advice. Pokio Ryan and our guests may hold positions in the company's funds or projects discussed. What's up guys?

Welcome back to another episode of Xerox Research. Today I'm joined by Dimmitri who's a partner at Archetype. Um, thanks for joining us Dimmitri. Yeah, for sure. Thanks for having me. Um before we get started, do you wanna I remember you were at uh somewhere else before. Maybe do you want to give a short introduction of what you've been up to over the past few months um and your background? Yeah, I mean investing in crypto for for a while. Um since 2018 across a number of funds, Coin Fund, crypto family office called Ballinger Investment Group, 1KX, been with Archetype since last January.

work has been the same throughout the years generally uh investment due diligence research and portfolio support. I cover things like DeFi, payments, deep pin, interop and scalability. And I work with teams around go to market token design and follow on fundraising. Perfect. I um I actually remember reading a few of your Medium posts. Um, I think I may have copied a few of your thoughts and key takeaways back in my like original when I first got started in crypto. So, I'm excited to be able to actually chat to you and then maybe also going into the future copy some more of your takes. Um, but uh I mainly want to talk to you

about deepen. Um, I think that's deepen payments and then maybe some AI today. Um but primarily I am very I'm semi- ignorant about deepen. Um and I always bring this up to VCs. Um because VCs are typically involved with deepen. I think it's more of a something that's coming up and not something that's very linked right now. Um do you want to maybe share your thoughts on why deepen hasn't succeeded? Um or I guess what would you say would be success for Deepen and why it hasn't achieved that yet? Yeah, I think there's two components you know. I think the first is I would argue certain projects have been successful. It took a while. You know, Helium today I think they have around 700,000

um uh subscribers and 23 terabytes of data um uh uh that is being offloaded by their network. It took a few years and I think you know they they they admitted saying the IoT approach probably wasn't the the thing that gave um organic demand. Um Geodet has a few million dollars of of revenue. They committed to burning um 80% of that into their token.

Um, storej has a bit over 10 million in ARR, but that is in their off-chain uh devco upco. So, the it took a while. I think the the issues have have generally been twofold. One is that you needed to get to par to your centralized counterparts on the demand side. And I think that's having very basic SLAs's. It's abstracting anything blockchain and it's communicating it the right way because I think while while we might have a

certain value here really in the let's call it the real world nobody buys decentralization for the sake of decentralization. So they buy lower costs, they buy resilience and I think it's been a learning process from a lot of these projects. The other issue has been this disconnect as you mentioned between almost private investor interest um value acral on the equity side and the reflection of that on the liquid markets and I think that is still a work in progress. Um, a lot of the patterns with these deepen

projects is they have a protocol and that can be a foundation and then they have a opco or a devco that pays the network but then they receive in in fiat payments from their customers whoever they are. And with a lot of these the the value flow is not really fully there yet between like how do you actually convert that and you know look a lot of it is concerns around securities laws and and and and whatnot but I think a credible commitment to value acral in the token I think is something that the liquid markets are looking for. I think

that makes sense. I feel like a lot of other sectors had um a second iteration of protocols where tokconomics were adjusted to reflect some form of value acral. Um and deepen it seems like it never really had that like the demand side for the token never really came um just because there was no value acrruel. And one thing you brought up is the idea of um the back end is kind of complicated in terms of the token is there's like equity side and then there's a lab side and then someone's paying in. Um do you think um with I guess better regulation that will change and is that something that you guys focus on and that we want

to make sure that like if we're doing a deal I want to make sure that there's a proper setup in place such that equity and token it's very clear we're not like doing something that shouldn't be done. Um is that something you guys think about? Yeah absolutely. I mean, I think the the the issue is really the credible commitment to it and you know, it's very much like founder DNA. Um uh and and you know, I like I do think there's probably, you know, a bit more risk uh in in involved. Um but I think that is part of you know accessing let's call it like the the the token capital markets

you know and I think there has been this return to fundamentals over the last few months you know and I think you know AR is the new TVL and and I think like like we've been looking very deeply into actually what what revenue is is there because I think for I I definitely for deepen the idea that you can use token incentives to bootstrap a supply side that's no longer an argument. You know, of course it works really well and crypto is really good at it incentives and now it's how do you build a good product on the demand side and and and

how do you have you know it goes back to startup 101 like who is your user persona? what are you building that actually solves a shark bite and not a mosquito bite for them? And how do you have a willingness to pay? Um, and then how do you have a structure where you clearly outline how that value acrews to the token? What a lot of founders have done and I understand why they did it. They basically say, "Hey, we don't have a lot of uh dollars in the treasury left. our token FTV is low. We are going to raise again um from a new a new equity

coco of which owns a small percentage of the token supply and then we're going to argue that this separate equity co acrru value and imagine that that opco holds 5% of the token supply and then they sell 10% of the opco So what the investor is getting um is is is usually a tiny subset of actual um exposure to the token network. You know, you're getting 10% of 5% effectively and you balloon the FTV to the point where then it

almost doesn't make sense as an investor. If you're excited about the network, you just buy the token, you know, and then a lot of uh founders are stuck between a rock and a hard place. And and I think the the right structure is to have a credible commitment over again the flow through between the revenue acrewed by the OPCO and uh the uh the token network. And if you are actually raising funds into the opco to be sure to structure it where investors are getting roughly a onetoone um flow through um across both instruments um and and that is really just also optionality as well um uh and you know

it it's but it's a conversation where there's still friction today. Yeah, I think that makes sense. And one other thing that I've noticed with deepen, I mean this can be said of a lot a lot of projects, but I feel like deepend projects typically suffer from this is on I guess defy coins or any non deepen coin there's kind of a focus on making sure that the markets are liquid um versus I feel like most of the time when I'm looking for a deepen token like on the liquid um for our subscribers for example for blockworks research I feel like there isn't stuff that's particularly liquid. Would you say that the deepen um I guess in general that side doesn't particularly focus on making a token investable for liquid

funds and more so is like we need to figure out a way to incentivize users which is clearly the first thing that they need to do like they need to bootstrap growth and make sure that people want to use the network. But do you think that they often avoid it or is that something that I'm just kind of missing? I think yeah I actually think it's not intentional by by a lot of these founders you know I think the the interesting part about deepen is that it attracts a different kind of founder persona which is almost like the electrical engineering PhD that that found you know issues in their own market structures across you know telecom and energy and mobility and and they say okay like we can use tokens to

to solve this market dislocate but they're often not super cryptonative and and I actually think it's it's not something that they intended and you know I've gone to to certain founders saying we you know we we'd like to do a treasury uh diversification for you and they said please buy on the open market but I don't think they thought about the the mechanism to to to structurally have you know at least 8 to 12% float um uh I do think it's it's it's top of mind and it's solvable over time as you are giving more rewards to users. Um well both the supply and demand side and then that becomes more

liquid. Um but I think you know for for funds like us uh depth actually continues to remain in an an issue and and I do think there's the intention from uh from founders to to to want to address it. Today's episode is brought to you by Echodin. Echodin is a decentralized orderbook based exchange built on Apdos with a mission to create a scalable institutional-grade trading platform. To learn more about Echodin, follow the link in today's show notes. As always, investments in blockchain technology involve risk. Terms and conditions apply. Do your own research.

I want to also I want to take a step back and speak a bit about um what a deepen network like how you guys would approach a deepen network investment. Um when let's say we're it's a telecom company um or like a telecom uh deep end um what is what would you say they need to focus on? Is it like underserved communities or is it just like where there's a gap in the market in like I guess a bigger market? Is it um what should they be focusing on in that sense? And is there anything that was happening I guess two years ago, three years ago that is now you're like we definitely should not have been doing this back then like this didn't make any sense um in the sense of like how they

were offering the product um how they were approaching the market etc. That's interesting. I think for a lot of these projects what what we think about and one of the most important questions we ask is why now? And it's really looking at what are the inflection points in the industry that the founders can almost leverage as as tools or weapons to shift the future into their vision. And I think that applies to things like I mean across telecom, energy, mobility. Um, this can be new

technology advance ad uh advancements like um having more wireless capabilities and and and being able to to to transmit um uh data without needing to actually um have wires uh installed. Um uh it can be um economics around the actual hardware. So today, for example, it is cheaper to uh install and buy a new solar panel than it is to maintain an old one, you know, and and I think like the unit economics there make a lot of sense. And so that is, you know, a trajectory in in solar that I I think is very exciting. And it's asking

the question of what is naturally going to become decentralized and and where can you start to apply uh crypto incentives to to actually um uh create a functioning network. So so I think we we we pay a lot of attention to um market structure and for for a lot of projects you know it was simply too early. You know, I think about the the web van example where where that like that's basically Amazon fresh today. Um, but that was 20 years too early and you know investors burned a lot of money on it and you know at the time there wasn't mobile and

there wasn't a lot of internet penetration. Um, and uh I I I think about that a lot. Um I think we've seen that with compute you know on on the virtual resources side where I do think the why now is AI demand you know and and and the demand for both uh inference and and and training and um in that context compute is like electricity in the sense that no one ever says hey we have enough electricity it's like the more you have the more you can do with it and I think that's been a structure cultural tailwind that I think is making viable a lot of these decentralized

compete plays today where two, three, four, five years ago, you know, pre-transformer it just wasn't that significant of a need. Yeah. I I actually wanted to ask about decentralized computers. Well, that's another thing that was on my mind that um do you think that's just a like it's overcrowded? there's way too many decentralized computes marketplaces or do you think it's just the not the right time yet and we need to wait a few years where the demand really goes up a lot more? Yeah, I I think it's crowded, but it's like I think it's a good thing, you know, and and and I think there there's a lot to innovate around the uh the approach where, you know, I like I don't

want to call out certain projects, but the way that orchestration and routing is done today is quite elementary. Um and I I think there is still this uh uh this vision around you know how do you coordinate heterogeneous resources to actually provide a coherent uh a project to a demand side you know so so from 4090s to A100s how do you find a way to um split a workload and distribute a workload where you're not bounded by the weakest link

which is the weakest machine on your network. And and that's a very difficult problem. And and there's been very there's been interesting approaches um recently around for example, how do you um split the model itself and then route different layers um to different machines depending on what you have to do. Uh we recently spoke with a senior exec at OpenAI who is leaving and starting his own decentralized compute project which is doing exactly this and he said that I don't see myself as a crypto or blockchain company you know I just see blockchains as the right tool

for the job and and I think that was you know inspiring I suppose you know we're uh because we are seeing very impressive practitioners now coming into the space. Um, and I think we'll we'll have a lot more sophisticated approaches and actually like getting this um performant um because it's still a huge issue. You know, we uh we spoke with this one company the other day that's doing um deep fake uh AI video uh production. So I can, you know, in a prompt say uh like generate Dimmitri as the pope, you know, and and it shows a video of me as a pope and it cost them a $1.40 per inference.

And and the like the unit economics on that just don't work out. You know, if you have any meaningful attraction and and Sora probably has, you know, millions of of monthly active users. If you get beyond 10K, that that that doesn't work, you know. So, so I think that's where uh these decentralized compute uh projects can be quite impactful. Um because I do think we need to get those costs down to you know cents per per inference. I want to ask a bit about that cost because um and please treat me like I don't know anything about this which is the case. I don't know anything about it. um if I am a decentralized compute um I guess marketplace, how am I

competing with um regular compute marketplaces? Like what is it just at a steep discount um or is it how does that work exactly on that end? Yeah, it's often the uh the discount and the um the supply of machines like what is your supply? Are you aggregating, you know, individuals with 4090s or are you aggregating tidy a tiny data centers with A100s or H100s? So the the the supply side matters a lot. A lot of these uh uh decentralized compute projects say they

have some data center, but do they actually own the supply and how sticky is that supply? Um a related component is the actual um model layer. So are you serving um uh and hosting open-source models as well or or or performing open source models? I think that's another um component. Um so I think those are generally the the two main vectors. That makes sense. And um are they verifiable in terms of like if they're saying this you're using an A100 is it typically

that or is there a risk that it's not? I think there's a risk. I I I don't have a lot of visibility on how verifiability is done today. I think it remains an an issue. I do think it's trust me bro. Um uh but also you you can't have this fully on chain you know um because the moment you do you lose any real performance and cost benefits you know and and and again I don't want to name any specific project but we we've seen the issue um and I do think there's a spectrum of verifiability and and I I I do think you know some spectrum of trust trust is

probably okay in the short term. Um but yeah, I I I would be surprised if we had any um trustless mechanisms to to say your your inference is coming from this specific machine type. You know, I think you probably see it indirectly in in the um I don't know, the time spent training or or or the time uh to to produce an inference. Yeah. Only if something is wonky would you realize, okay, well, this is probably not right.

I'm being lied to. Yeah. Like this is taking too long or or or this is way more expensive than I thought. Engineered to meet the demands of highfrequency traders and regular traders alike, Echodin redefineses onchain trading with lightning fast performance and decentralized security. Echodin's building a scalable institutional-grade trading platform with execution speeds as low as 5 to 15 milliseconds, deep liquidity, and zero maker fees that leverage Apostos' cutting edge technology for both spot and perpetual futures markets. To learn more, follow the link in today's show notes. And thanks again to Echodin for sponsoring today's episode. As always, investments in blockchain technology involve risk. Terms and conditions apply. Do your own research. I'm going to switch over to payments and stable

coins. I we have a conference called DS, which was like two months ago now, and ever since then, this is all I can hear about and talk about is stable coins. Um, I want your thoughts on first stable coins. Do you see there being a single winner? Um or I guess like USDT USDC I would say is like a single winner at this point um together even though that doesn't sound right. Um or do you see um new stable coins popping up bank stable coins they're just like all of web two coming in and saying we're going to launch our own version. Um how do you see that space shape up? Yeah, I mean I think categorically there there will be more than one winner because there's

more than one flavor of stable coin and in the onetoone backed realm yes I I I would argue today it is USDC USDT we will also see both um fintex well fintex larger financial institutions and large tech companies come in and that's going to be really interesting fireworks to see. You know, having PMF is a double-edged sword in that regard. But there's also, you know, the cryptobacked stable coins. There are the yieldbearing stable coins. Some have yield as the product where it's baked

in. Some have yield um as a go to market where it's not baked in directly into the asset but the company uses that as a go to market where they pass on the yield. There are non US dollar stable coins um and you know like not highly inflationary ones um but that's been popping up across different countries around the world. There's stable coins that pass on non-treasury yield. You know, this can be high-grade sovereign debt as well. So, I think there's going to be actually a few winners in in a lot of those. And, you know, maybe it'll

take longer for non US dollar to to to have a a a use case at scale. There is around $600 million of non US dollar stable coins issued today compared with roughly $230 billion of stable coins in in general. That's from Artemis. Um uh so let's see how long it takes for uh for that to to to catch up. And then another framing is you know close loop verse open loop. Um this terminology is from payment networks. Um so so a lot of like what we talk are talking about is open loop where you know it is freely transferable

um uh and and it's uh meant to be you know a medium of exchange. There's the other use case where you have um a fintech who mints a stable coin and keeps user deposits uh into that stable coin to earn treasury yield on it. And Venmo already does this today with your user deposits and and this really enables any company to have a business model that is effectively what Venmo does today. Yeah, I feel like in my mind most of the web 2 that comes in will end up doing that. Um just capture whatever they can offchain and then allow for

some form of quicker settlement. Um you mentioned non USD stable coins. I feel like this is I don't know what you got you like obviously you've spent more time thinking about it and you probably understand this better than me but in my mind um so I'm originally from a country that has very high inflation so everybody just wants to or hold the US dollar like we've always seen people want the dollar sometimes the euro but not as much um I feel like the like euros I guess the Swiss Frank I feel like that'll always be like maybe sub 10% % and the majority will always be US dollar. Is that correct in thinking or do you think we see a shift?

I think so. I mean, you know, I don't I don't want to sound too American, but the the US dollar rel relatively is arguably the strongest currency. So I think whether it's a business type use case of holding your treasury in an asset to you know a family in an emerging market. I I I think the ideal pattern is that they are able to hold funds in digital dollars but pay in local fiat and so that the off-ramp and the FX hap well the FX happens at the off-ramp I suppose um and and I think

that is probably the the pattern that most want assuming that you can do that without much friction the the I've been think you know thinking about the case of why tokenize non non US dollar I think there is an angle around FX markets you know bringing FX markets on chain um uh I think that's very interesting um where the gap is in my head is what is the demand for and it you know it could be a market maker it can be an exchange to hold a fiat in a tokenized form rather than just on

balance sheet. Yeah. You know, and and and I don't quite have that answer in in in my head yet. Um I I do think there will be a pattern where more local exchanges are tokenizing the um uh local currencies in uh uh where where they operate. Um and and uh it might be a play for them to participate in in FX markets more let's call it like B2B FX markets because they already play in FX today um just by nature of the liquidity where you um you can effectively use Bitcoin as a cross rate for um between

two different currencies. So, US dollar to Bitcoin, Bitcoin to Mexican peso, uh you could actually get very tight spreads. And so, a lot of payment companies are already using um the the the byproduct of speculation effectively as as um uh quite an attractive uh business prop for themselves. And and and the question is okay like they already are being used for that. what is the benefit of those assets actually being tokenized? I I haven't yet gotten there. Um, but it's an area I've been diving into. Interesting. Surely it's to I don't know if like if anybody has ever

dealt with converting currencies. I feel like that process definitely needs to become a lot more efficient. So, I think that's like a reason to at least do it. I don't know what it's like in the US to be honest with you. And I I know web two companies take a spread not as bad as what you would get if you just tried to exchange fiat for fiat. But um that kind of brings us to payments. Um and I feel like I can ask this question about every sector in crypto right now, but what's the we've seen a decent number of crypto banks coming in. Um Epheri is the is the big one that I would say is like more cryptonative. Now we

have Stripe working with ramp and bridge um is it bridge? I can't remember the name. Um to build out something as well. I feel like that'll actually be useful. But the main question here is do you think somebody just comes in and swoops in like a web two company and offers the best UX that they can just by having stable coins on the back end or do you think there's a chance for um cryptonative companies to win out there? And the followup to this is how do you see that um space evolve over the next few years? Because I really want actual crypto banks. Um yeah. Yeah, for sure. I mean, I think you can take a step back and and look at

this as an evolution across several decades where you have payments 1.0 which was effectively correspondent banking powered by Swift. And this was a network that, you know, it's it's global, it's secure, but it's very slow and it takes days to weeks. I've heard from founders over a month to have a payment land. Uh, and then you have, you know, payments 2.0, know which was effectively these fintexs like wise which used um uh preunded accounts local rails for payins and and and payouts to effectively simulate real-time payments.

So they effectively have liquidity pools in each jurisdiction and while that effectively is faster, it came with a lot of other issues. Um notably it's still quite opex expensive for them because they need to maintain u a complex treasury and compliance infrastructure. They have to have multiple local correspondent banking relationships. They need to have risk engines to manage float and effects volatility. And I think payments through 3 3.0 is really using blockchains and having this internet native settlement layer. And now we can do transactions in

the order of seconds that cost in the order of cents. And there's no evening or weekend or holiday cuto off times. And it's fully transparent. So you always know where your uh your payment is. And I think you know a similar question to a lot of the deepen um uh advancements is the why now because this is a topic that's not new you know um but I think the the why now is we have stable coins that are mainstream. we have better on and off ramps uh uh with more depth because those local OTC desks and uh market makers they do want to hold nonvolatile digital dollars on

balance sheet along with their local fiat. We have uh embedded wallets like privy that actually give us a web 2 like UX. Um we have pending clear uh regulation around uh stable act and the genius act and I think we're we're really opening the floodgates for for for both uh fintex and larger financial institutions to come on chain. Um, but this it it's it's a double-edged sword for the let's call it the crypto incumbents because what crypto teams really suck at is solving for for distribution, you know, and Revolute alone has 60 million

MAUs and and and I think the conversation today is very different than it was three four years ago. I I remember probably 2020 or or 2021 I was I met the the the head of BD Chime and and um I asked them you know like Chime's users are middle inome Americans in um like central US states and and I said you know like wouldn't it be nice if you gave them access to DeFi yield and and the guy was like I don't want anything to do with crypto you know like I I'm so busy trying to not die competing against uh uh other big banks that I don't want to take any more risk.

And today the conversation is flipped where Revolute has their own exchange. Um they have some really interesting other products in the pipeline. Um you see it with Robin Hood and uh obviously Stripe. Um, and a lot of the DeFi projects like the crypto neo banks, they are good products, but they do not solve for distribution. And if you can figure that out, then I think the DeFi mullet is probably the best way to go, but it's going to be very teamspecific. But I've I I've spoken to many let's call it high pedigree founders that are building a crypto neo

bank that cannot answer how they will get to I don't know like a million users. Yeah. It's especially kind of disheartening because Revolute, if you have a paid Revolute account, they give you maybe $30 in incentives per month versus in crypto the like to acquire a new user. You're a giving up equity um and b you're also just spending millions of dollars potentially across like six months. Um and people don't stick around. So yeah, it's distribution is one. Um, I want to challenge you on the

privy being good UX. I don't I don't think we've solved, this is not a hot take. I don't think we've solved UX at all. I still feel like it is absolutely horrible. And the main reason I think so is that we don't build mobile first. we still build web first which I feel like is ridiculous for even a like for a payments app it should be mobile first. Um, so I guess first what do you think about that and second if you do talk to teams do you tell them hey guys like you have to build a mobile app it's not it's non-negotiable yeah yeah so I agree that we haven't had this before but I do think we are seeing some really interesting ones um I mean

even something like moonshot in uh in the US you know that that was you know outside of the New York issues um that's an Apple pay on-ramp. You know, the the the the holy grail of the on-ramp is um Gmail login, privy wallet, Apple Pay on-ramp. Yeah. And and I think we are actually getting there today. Um uh either you know in the app store or or a PWA. Um, but there are some some really nice experiences. Um, uh, Farcade, like I don't know if you play with that one, um, uh, in the Farcaster app, um, you just play like mini games on there. Um,

surprisingly smooth. I do think the next wave of at least on the consumer side um uh is going to be mobile first and and uh it leans on you know on on the payment side to remittance apps peer-to-peer payments on the DeFi side where um you know it kind of blends DeFi and and and and consumer um but yeah I would argue it's actually happening today. Um, and it's an area we've particularly been excited about like mobile first cryptonative um, uh, w with a really good UX I I I think is is is pretty magical. Yeah,

agreed. I, um, I'm going to give my own example. Uh, Slingshot UX on mobile is fantastic. It's a trading um, and you can trade on whatever chain you want and you don't have to deal with gas, etc. just allows you to do it. I don't know if it has Apple Pay. I need to check. I hope it has Apple Pay and if they don't, they should get it on um I guess with recent Apple um changes, we're going to see more and more applications on the App Store and less PWAs because I feel like PWAs are always less than ideal. Agree. Agree. Yeah. the the the holy grail here is to be accepted into the app store and you know until we have enough of the population using the the

uh the Salonophone or whatever um yeah I think that's probably the the uh the path of least resistance. Yeah. I um if you are talking to a company or a new project and they say we want to launch on chain X do you tell them actually I think it makes more sense for you to go like for deepen I imagine a lot of people's first thought is Salana just because there's so much deepen on Solana um is that something that you guys also iterate to them and say hey like it makes sense if you go to this chain actually because it's doing better in terms of attention or there's like more support or the tooling we've heard is much better or is that something that you don't really talk about with them?

Yeah. Um, there there's nuances here. I I would say two things. One is that as a startup, your general goal is to not die, you know, and and the more risk you take because you already have a 95% chance of dying, right? So, so why bump that up to 98 um if you can avoid it, you know? So, so I think if it is, you know, consumer the base is probably the the the the least risky. Salana is probably the the least risky. Um uh if it is DeFi, Arbitrum is not very risky.

There's a lot of DeFi on there. Um I hold no bags in all of these by the way. Um it's just like observations. Um the the the the nuance is that if you are building something novel and you get distribution, you can choose whatever chain you want because the end users probably don't know. And there's a lot of very performant chains now that can do block times subsecond and that's like good enough UX provided there there there's a a good enough on and off ramp.

Um but you know like Pump is a good example of saying we we own the end user. We commoditize everything underneath and we've seen the the success with Pump Swap. Um, I have not spoken to him about this, but I do assume there's going to be a pump chain and and you just like vertically integrate all the way down. Um, Pudgy with abstract, you know. So, so, so I do think um going top down after you have distribution is very doable and then you're much less I don't know um be beholden to to any specific L1 or L2.

That makes sense. Um feel like polygon poly market is somewhat of a good example of that where like it gets users. Polygon is not like a fantastically active chain, but it still works well. Um, I um I want to ask a bit about consumer crypto. What are you guys What are you excited about? Um, is there anything that you think people should specifically focus on? I feel like consumer crypto sometimes gets um kind of muddy because we do a lot of trading like retail trading goes into consumer crypto. Social buy goes into consumer crypto. So either of those is fine but which one do you see um I guess like what do you see coming up over the next

year two years where do you see it going? Yeah, and I think that's the point in that um we are seeing a lot of genres blend you know I' I I've seen teams tried to take the approach of you know Robin Hood democratized options trading effectively and and uh if you've used the app the the uh the UI is really intuitive and and it um really nicely lays out the uh the payout functions um for you. And I think we're um we're seeing that with these Sociali apps. Um I do not uh turn away from speculation.

I'm not one to judge what PMF looks like. Um and I think there's still a lot of um demand there. Um, and the moonshots of the world h have have shown that good product and good UX can get you distribution. Um, I think the um the rise of generative content is very interesting. Um, and I think there's um a lot of unique value that uh crypto can offer around attribution and monetization. And there's a spectrum

where it can be something super simple like it like a meme or an image to something more complex like a game from a miniame to more complex ones. I do believe in a few years we will likely have um a new foundational model or some fine-tuned model that is specifically focusing on uh game generation. And so if if creating games is as simple as a prompt, um then you get into this um market structure of um ephemererality, I suppose. um and um just like a like um over supply of content. Um, and I think

there's a lot of interesting areas for for for crypto to be used as the mechanism for attention and as the um uh the the the carrot I suppose to actually bring um I guess like developers or content creators onto a platform if they know, you know, if the game that I made or the image that I made is being remixed and I have some kind of attribution or or creator reward there. Um I think that would be quite interesting. Yeah, that makes sense. We just saw um Pump Fund implement 50% fees creator fees. I know that's on the memecoin side, but I imagine we'll get more and more and

more. Um I have a this is a non-crypto-reated question and a non-investment related question. I just like kind of want to hear your thoughts on it. You mentioned that in a few years a foundational model might just be able to spit out a video game based on a prompt. Um, how do you see that change the way that you and I value something? Like, if you can create a game just off a prompt, do you think that's worth the same as somebody who spends three years building out a game, an indie game?

Because I feel like this is something that we're going to get more and more of with image creation, art, everything. So, curious to see if you have any thoughts there. Um it's it's a philosophical question and one that we will need to see how the market values it. Um to okay I think that in in five years um only 5% of the content online will be human generated and and I and I think that cuts across all of the use cases.

Um, and I do think there will be a premium in in in certain uh use cases. Let's call it the humanity premium. Um, I think for things like games, I don't think that's something that to to speculate, I suppose, what an end user cares about. you know, I think they they primarily care about the utility and the game mechanics and uh and the fun that they're getting from it. Yeah. So, so um I don't know, bearish game devs, bearish musicians, you know, I don't know if you've used any of those. Um

uh I think it's called. Yeah. Very good. Yeah. It makes some great music. Yeah. You know, and and um I think it'll there's going to be a lot of very deep implications of um how we um value things and and um it has deep implications on what the nature of work looks like. Um I think that's going to be a fascinating thing and and and and we're seeing it today. you know, it it it does have um you know, education as well, you know, like like do people value that a professor said something or you know what I mean? Like like they

just want a a piece of paper saying that they learned something. Yeah. From a a university. you can learn everything just like skill maxing on chat GBT for uh for you know I did that for like four hours last Friday on just like energy markets you know um and I think it'll redefine um a lot of you know business models and and and really like bring to light um what what people actually value. Yeah, completely agree. Um, related to this, do you use I mean you just mentioned that you you scale max energy markets last Friday, but how I guess how has

your investment change like the way that you guys do investments and the way that you do DD and like how you work on your side? How has AI changed it? Has it cut down your time significantly or is it We We have not yet. We've been thinking about it. Um, one of our port codes is uh one called Exo. Um, they do edge compute. Um, so you can run like a mini like like a Mac mini data center in your office. Um, for us a lot of the concern is, you know, we have a lot of internal due diligence notes and memos and we don't feel comfortable with that going

in the cloud. So, can we create, you know, a tiny exo cluster and then fine-tune like a llama model on all of our DD and have uh just like a AI analyst. Um, and I'm sure funds are are doing that. I I've spoken to like a DAO that's trying to do exactly this. Um, I think that'll happen. Um, and and you know, it's interesting because there were um approaches to this even pre-LM. I believe uh there's a fund called Signal Fire that uh that was trying to use AI to do better deal sourcing and diligence. Um fact check me on that. Um but but but there's always been the

question of like how do you extend? Um it's been uh an idea but but it's not something that like we've actually done. Um but probably a competitive advantage for for for someone who's productized this. Yeah, I um that makes sense. I'm only asking because we on our end I mean you can imagine um I think all of research is kind of in a spot where it's because it's changed drastically like deep research is fantastic. Um so we've been thinking more and more on our end like how to utilize it and it does it saves you a lot of time and we don't really have the internal notes that we have to be really really careful about.

So I think we we get off a little bit easy. Um, it's gonna be interesting. Let's see how everything shapes up. Um, I want to ask one last thing somewhat tangential. Um, AI agents being mostly onchain. Um, I guess AI agent commerce being mostly on transactions uh on um AI agent transactions being mostly on chain. I think that makes sense. Um I I want to ask why won't they just be able to use I guess a traditional banking system? Um is it just because it's too slow, settlement takes too long, or is it just it's easier for them to do um everything on? Yeah, I think

it's as simple as like how does you know to to open up a bank account you have to you know walk in a branch or or go go online and provide pieces of your identity. Um, and you know, pieces of code don't have an identity. So, uh, uh, at least in meat space. So, how do they get bank accounts? And I think that's really a it's a great fit, you know, co because you can have, you know, a public private key pair um uh that is uh designated to an agent and then they can have a wallet and they can um send and receive funds to and from other agents

to and from other people. Um, you know, we'll put this in the bucket of uh Apologies always right, you know, with um what what what he was doing with 21 co um but it's less of like your your your fridge paying your toaster with Bitcoin. Um uh but it's more so like using stable coins and having that be agents around the world um paying each other um uh in microtransactions for um specialized jobs that that that uh that that each agent can can do. Um, and yeah, blockchains and and crypto rails are are really the perfect medium for that, you know, because you can't

actually have a cost structure where you can do transactions in in in fractions of a cent. Um, so we we we of course have not seen many of those use cases today um play out. Um but I think it's also because we are waiting for I suppose like more performant agents with uh more utility and better orchestration frameworks to actually coordinate um you know like the the the meme of like agent swarms but that's on a one to threeear time horizon. Yeah I think that makes sense. I have used one ajet style protocol uh which was on

Ethereum which I felt like was the wrong choice for it cuz a it's very expensive and then um I will say I'm kind of surprised that agents haven't become more like in real not real life I guess like non-crypto agents haven't become more popular. I I feel like it would be it would have been here by now but I guess it takes quite some time for it to figure out. I think the only use case right now is coding maybe. Have you seen anything else? I mean that like that alone is I think really interesting. Um I think with operator which is open's um uh uh product. I think that will give agents almost like arms into the real

world. Um I think we're on the cusp of it. you know, the the the the pace of innovation is um pretty mind-blowing. Um so it's always like, you know, things are um closer than what they may appear. Um I put that on, you know, a three to six month time horizon probably. Interesting. I'm hoping so because I I have so much stuff that I just want to delegate off some like agent and just leave me to do my whatever I want to do instead.

Um I think I think that was perfect. I think that's it. Um thank you so much for coming on. This was fantastic. Yeah, this was fun. Thanks for having me. [Music]