AI Memecoins & Lessons for Narrative Investing

Bankless Station, we got a bankless takes episode for you because uh we got something top of mind, and I think it should be top of mind for you as well. David, you wanted to come talk about AI tokens, and I think that will unleash uh a few other topics as well that are adjacent, including narrative investing and how to make that uh you know 20 to 100x gain, whether that's even possible, this bull cycle. But first, what are we gonna cover today? Tell tell me why you're excited about AI tokens.

I think today, Ryan, I want to plant a flag and say that the next crypto cycle, the 2024-2025 bull market cycle, is going to be the AI token cycle. In 2017, we had the ICO cycle. In 2021, we had NFTs. In 2025, we will have AI tokens. And I'm seeing these like just two parallel technologies just converge together in spectacular fashion. So I want to talk about like some of the experiences I had at ETH Denver, call it AI Denver, the world of decentralized compute, the road to AGI that crypto can facilitate, and why crypto and AI is such a match made in heaven, and why I think it's going to define the whole entire market cycle that is this incoming extremely frothy, extremely speculative bull market.

I think that's a good grounding for what I want to talk about too, David, which which relates to sort of AI tokens. The question of is this a narrative trade, a narrative investment? I think your answer to that in comparing it to ICOs would be yes, it is. Maybe there's something fundamental here, but it almost is certainly a narrative investment. And so I think we want to broaden that and talk a little bit about um narrative investing in crypto in general and maybe some of the lessons or frameworks that apply. Because I I saw a fantastic piece by a guy named Victor that I want to share with you too, David. And then maybe we could conclude with talking about what the next big narratives this cycle might be. I think you came back from ETH Denver with a list of three that we want to get into.

One very big one, and then no second place, and then a third and fourth place.

Alright guys, we're gonna get into all of that, but before we do, we want to thank the sponsors that made this episode possible, including Alright, David, tell me about AI tokens. Why are AI tokens uh a thing? Why how is that intersecting with with crypto right now?

Both of these industries are being built in parallel, come like completely separately and adjacent from each other. And I think the space that these two industries occupy and overlap at is going to be the thing that really captures a lot of speculative fervor, mainly from the people that like, hey, I want to invest and have exposure to AI, but I don't want to invest in NVIDIA because it is in the larger than the entire Chinese stock market. There are these two storms that I'm seeing slowly converge together as they learn how to like relate with each other. I'm calling it like a particle collision. Like, what happens when you smash the AI particle with the crypto particle?

You got on the screen you're making me show the uh Andromeda galaxy and the Milky Way galaxy. I think that's on track to like intersect with one another in like five billion years or so. These are galaxy sized collisions, is what you're talking about.

Well, yeah, we have the uh three to a two and a half trillion dollar crypto economy colliding with the, you know, probably a couple trillion dollar AI economy. And I think we're going to produce a few trillion dollars as a result of that collision uh from across the space. We're and I think it's mostly gonna be AI meme coins. And why is it gonna be mostly meme coins and not like AI fundamental coins? Well, what do we do in crypto? We just print tokens out of thin air and we name them things that are like related to AI and then they pump. Uh and so there's gonna be a few ounces of like AI fundamentals that I think are going to be very real that's going to produce like the massive amount of leverage that it has that we have to fuel into the world of AI meme coins. I that's what that's what I think is coming.

So you think this is uh a bit more of a narrative trade? Actually I wanted to show you this. I I was wondering if you if you kind of agree with this. So this is uh a graphic that I saw in uh a a Twitter thread uh about atten attention captured, right? And we have like the serious projects on the top and this is kind of like a graph, it's a very small slip sli uh sliver.

Or ten percent of the bucket.

Right, five to ten percent. And then we have meme coins that comprise the bulk of this, being like, you know, 85 to 90%. Is that what you're saying? Is that how you are thinking about um AI uh and crypto as we get into this? Is it like ninety percent meme, but like ten percent substance?

That's that's exactly right, along with an order of operations there. First, it's the 10% substance which creates the space, the room, the motivation for the 90% meme. I think there is actually a ton of substance between the relationship between crypto and AI. And it's it's that is go as a result of that very real intersection. We're going to attract, you know, for better or for worse, mainly for worse, like the scammers, the grifters. Like this is what it's gonna speed the signal of the top of the market, is that like people realize that they can make a quick buck by like throwing crypto and AI together and like you know, doing some like ICO, if you will, or like some some startup race. We've already seen froth in the AI VC market six to nine months ago post ChatGPT launch. And like, what happens when you cut uh collide that particle with the ability to mint an ERC20 token permissionlessly?

Two frothy world changing technologies that no one understands, and they're gonna smash together and create like what, a super bubble? Is that what

A legitimate crypto super bubble. A super bubble. Super cycle, call it. Like I also want to leave room. I was talking to Mike Ibulito about this. I also want to leave room for the potential of like the whole AI world create something like approaching like the dot-com bubble slowly over the next like you know two to five years. Something stupid like this. Uh AI is gonna change the world. I think everyone has like now realized that. But now the hate whole AI industry has access to quick liquidity tokenization events, meme power, like content generation power. Like I think wouldn't yeah, I think it's gonna get very, very far.

Yeah, the context here then is that this is definitely a narrative investment, it's definitely a narrative trade this cycle. And also there might be some fundamentals here, but it's going to be a small sliver, a small portion of the value here. It's going to bubble and froth and get way over its skis. This is a uh a graphic from Vitalik, a uh recent post that he did at the end of January where he talked about the intersection between AI and blockchain, and they're they're kind of two separate technologies, but there are some synergies data ownership, transparency, monetization, cost cutting, competition, innovation, inclusive, but I'm guessing we'll also create a whole bunch of like AI meme coins that really like have are nothing more than narratives, right? Are not actually making use of blockchain fundamentals and and kind of this intersection in the Venn diagram.

I I think we can also like talk about the blurring of the line between fundamentals and memes. Like WorldCoin, for example, like has a ton of just like very real building going behind it. That is a very real project, has extra uh uh attracted very real VC dollars, and it has done like a very like good job. But like what no matter what you believe about the project, they are doing this very hard thing about building this infrastructure to help identify humans on the internet in the age of AI. It is also, Ryan, trading at almost a hundred billion dollar market cap, and that is a meme valuation. Real project, meme valuation. And so, like, there's gonna be this blurring of the line between like what's real and what's like hot air. And right now, like we have a little bit of hot air. I think we're gonna have a lot of hot air. First, I want to talk about like why there will be hot air in the first place. Because if there's hot air somewhere, that implies that there is a real source of energy producing that hot air. I'm talking about fundamentals here. Like, what are the reasons why there's a fundamental truth about the intersection between crypto and AI? First, I want to talk about decentralized AI models.

Second, I want to talk about decentralized compute platforms.

Third, in order to really strike our imaginations, the road to AGI using crypto as a railway to get there.

And then also talk about after that the shared properties that crypto and AI have. And then lastly, we'll end up where we're ultimately going to converge no matter what, which is AI meme coins. Are you ready?

Totally. Let's go. What

All right. First, decentralized AI models. This is what happens when you have a lot of the properties that we have in the crypto space: open permissionless innovation, decentralization, moving complexity innovation towards the margins with a world of AI. I was at this AGI summit day and I was listening to Casey Caruso, who's sitting on the left there, who gave this absolutely banger take, which is that we have this desire to have as many AI models as possible. You know, your favorite Ryan uh line, Ryan, which is like we'll have as as many chains as there are websites, or we'll have as many tokens as there are our websites. I I think we also want as many AI models as we have, like blogs or websites, or any like large Cambrian explosion of stuff that we've seen on the internet. AI models can like do almost infinity things, and the more of them that we have, the more flourishing that there can be. The more

do you mean by an AI model, David? Like is ChatGPT is that sort of a an AI model and that might be different than um you know the the model behind X? What's it called? Uh the one Elon's developing? Yeah, Grok. Uh is that a model? What's a model?

Yeah, that's exactly right. So like ChatGPT is one model. Grok is one model. And I will call these very generalized models. There can be models for anything, and you can put any sort of utility into a model. Think about a model that has been custom trained, Ryan, as it to be your personal assistant, like your Siri or your like Amazon Alexa. Like we can talk about like the blue chip fat tail models that are very generalizable, that that are kind of one size fits all. And then we can also go down the line of just like, what about like an NPC inside of a game engine? Or something like some model about like the weather. Any sort of AI model that can be either very broad or very narrow. But the idea is that with the proliferation of AI, we have more models and we have a lot of them, or at least we can. That is the world that I would enjoy, especially when like you can start to associate like um AI with the growth of AI models, as in the growth of like human cultural uh proliferation, expressivity. Like, how can we leverage the power of AI in more total cases? And so this is the world that I would want. This is the world that a lot of the people in the open AI community also want. That's open source AI, not open AI, the company. But here's the line from Casey Caruso, who like uh brought us, draw everyone in the audience back down to reality. We want open source AI, we want a wide proliferation of models, custom fit for various use cases, but the tailwinds behind the economies of scale for AI models are very strong. Data monopolies and compute monopolies restrict open AI model development. And so basically, who has all the data, you know.

Facebook, like why is Reddit going public? Well, because they've got all the data for for AI. There's these data silos, these web two data silos that is all of the that has all the data that all of these AI models need to run on. So whoever has all the data has economies of scale for producing very strong AI models. If you, as your you know, regular uh homegrown AI tinkerer, if you don't have access to data, you're not gonna be able to produce an AI model. And so that is one very strong centralizing force that AI has. And then the other one is access to compute. Who's got all the access to the GPUs? It's going to have this massive economies of scale that the regular like long tail end consumer experimenter hobbyists won't be able to access. So I think if you like understand crypto principles, you can kind of see where I'm going with this. Like, we want AI models, but we can't have them because of siloed centralized resources.

Ready to go into uh decentralized compute platforms?

Yeah, so okay, so we want decentralized AI because it's just like uh richer, more organic, more competitive than kind of data monopolies, but that doesn't it doesn't follow that that's what we'll get just because we want it. You know what I mean? Like you you were just saying that these are centralizing, you know, like monopoly forces. So part of me is like, okay, well, game over. Like, you know, we can't we can't just like you know make them decentralized just because we want it. Is there some sort of market force that will propel decentralized AI and decentralized compute? Is is this uh part of the theme here?

Well, so this is getting to some of the actual like already existing tried and true crypto platforms that can actually pull that uh market force down towards the decentralized end of the spectrum. So first you're seeing like centralized AI, call that the Google, the open AI versions of the world, but we are looking for decentralized AI. And I think this like decentralized AI narrative is something that's like very ripe to be taken up by society when they learn about how much um returns on capital, on like monopoly that uh open AI and others will have over us, there's going to be just an interest in decentralized AI as a narrative. Crypto actually has meaningful answers to this. Like what some of the biggest tokens that have pumped, and this is when we'll start to get into like the narrative and meme side of the world. Some of the biggest tokens that have pumped in the last like three months are all these decentralized compute platforms. Filecoin is up a hundred percent in the last 30 days, Rweave is up 200% in the last 30 days, render is up 300% in the last three months, BitTensor is up 120, 1,250 in the last five months.

My God.

These are all various like decentralized compute platforms that have caught like the AI bug, the AI narrative, like have caught the AI bid. And so these in Filecoin, you put data on it and it's allowing for permissionless compute. And so the idea is actually very similar, Ryan, to your and my protocol sync thesis. Wherever you can have the most credibly neutral uh repository of data, it will collect all of the data. There's this thing called a data gravity in the web 2 world.

Whereas like if there are stockpiles of data that exist on the internet somewhere or like in servers, apps will actually migrate towards that data. So it's like thinking of like data as like a black hole. If there's enough of it, the data won't move. Apps will actually move towards the data. And Filecoin, the Filecoin thesis is that so much data will end up on Filecoin because it's the permissionless compute platform for data that just the apps will go around it. And so you know, Filecoin is up a hundred percent in the last 30 days.

I'm starting to see the c why the contours of we we um introduced this as a narrative because that what you just said is absolutely a narrative. It doesn't necessarily have like substance backing it, but like you could for foresee a world where everything you just said uh becomes true like comes true, but that's not necessarily going to happen. You could just sort of foresee it. So it's a narrative. Here's a a tweet you had us uh put in this episode today from from Anna. Today models are primarily trained on publicly scraped internet. What if a hundred million users contributed their private data from siloed platforms to create a user owned foundation model? So this is basically the part of the narrative is how do we compete against these centralized data banks like data monopolies? Well, we get a hundred million users to sort of combine their force behind a credibly neutral compute platform, maybe incentivized by a token. Is this sort of where we're going with the story?

That's exactly right. I love how you you brought up that tweet. Do you think that person's a crypto person?

Um, I've never seen this person, but I I was assuming yes, 'cause it's in a bankless show, but you're gonna tell me no?

The answer is no. This is not a crypto person. I have two overlapping followers. This is an AI person.

Oh wow.

If the tweet is talking about what if 100 million users coordinated their resources together to get them out of siloed centralized platforms to create a user-owned foundational model, that sounds like a crypto tweet. That is a crypto ethos. And so what the AI world is missing is what crypto has. Like the pushing power and authority to the margins, right? Pushing power to the individual. Whereas there's so much power contained in centralized Wall Street, aka Silicon Valley, aka Web2 resources, how they have all the power. Decentralized AI is about taking all of that power and like pushing it down toward the margins. But you need crypto to do it.

And so, like, you're talking about like, yeah, that's that's a narrative now. Well, markets price in the future. And so if people just think that this might be a thing, and also we have just like the markets going up in price, and we have an interest in going out towards the long tail, we have the Bitcoin ETF funneling hundreds of millions of dollars into Bitcoin each week, which is acting like a hydraulic pump.

down for the rest of the risk on market. Like this is the storm that is coming. We have this one ounce of very real narrative and it's gonna create a pound of like meme coinery.

Yeah, okay, so that's what's going on. I s I see the narrative. Uh it remains to be seen whether it actually becomes in put you in point of fact uh a reality here, but I see the narrative around decentralized models and AI and decentralized compute. Tell me about the this next piece though, the road to uh AGI, which is artificial general intelligence, right? Uh by the way, I'm not sure that we want to go down that road, but you included it here. What you know, where does this fit into the story arc?

This is where I think this is a little bit of a um a less assured take, but I would I think it would be very, very fun if we actually got there. Um when I went to got into ETH Denver and went to the AGI summit, which is where actually Vitalik was talking. Uh Casey Crusoe from that tweet that we were just showing was talking. Um Sri Ram from Eigenlayer was also talking. He was the first person I saw uh giving a talk uh at this AGI summit. And his talk I thought was actually the most interesting. He was talking about uh the like this road road to AGI and the three different ingredients that we need to get there. And I felt it was super compelling. I am getting on board on this narrative. Um, one is open source AI models, is one ingredient. Two is permissionless innovation to uh innovate on these AI models, and three, value accrual mechanisms to these AI models. So, like open source AI models, which is like the opposite of what ChatGBT is, the opposite of what open AI ironically is, the opposite of what Google Bard is. Uh open source models. So we can all as humanity coordinate and tinker around these open AI models. That's the first ingredient. Second, permissionless innovation on these AI AI models. So something that Eigenlayer is working on is this uh derivative license so that you, if you are making an AI model and then other people find it use it useful, but then you also make a derivative of it, there is a value chain of payments for developers who are making these AI models. And then this last ingredient is basically the marketplace, where can we link uh the value that these models are creating for users to actually the developer who is birthing them into the marketplace? And so once you have like um derivations or derivatives of these AI models, you get like replication and reproduction, and you add in the market incentive forces, you get a survival of the fittest. And so this is just one slice of one possible narrative that's coming out of this AI crypto space of just like we have these ingredients, we can make like expressivity happen, we can make a Cambrian explosion of innovation happen. Maybe, Ryan, we can even make AGI happen, but it's it's using a crypto foundation for coordination and resource management to produce a proliferation of AI, a proliferation of human culture and creativity. That's kind of like that's kind of the the part that I see.

It also strikes me that those three things are the same thing, like an uh effective accelerationist, like Beth Jesus would say. And that's the thing that's going to prevent us from regulatory capture by like authoritarian regimes and monopolies who will like, you know, use AI against the people, basically. So it it has that shared uh, I guess, narrative, let's say, with crypto as well. Okay, so the projects you mentioned, Filecoin, R Weave, uh, you know, like render, I don't know as much and bit tensor, but R Weave and Filecoin for sure, these are real projects that people are actually using today. But um we're you were also talking about AI meme coins, right? So are you talking about these projects? Like there's some obviously mimetic narrative element to even legitimate projects that are doing something real. Or are you talking about like a new kind of like uh shitcoin waterfall, like down market types of uh meme coins that are just taking this narrative and not producing much of substance and just running with it? Like what what what's that piece of the equation?

I'm talking about both. And like all things in every single crypto cycle cycle, we will start with very real fundamentals and we will end with shitcoinery, mania, froth that ends in tears and ruins. And so we are somewhere towards the beginning of that cycle. And I'm claiming that we are heading towards the memory froth end of the spectrum. And there's like just a lot of properties that the crypto industry has that the AI industry does not have and will probably want. The AI industry is like rapidly innovating. It's moving at breakneck speeds. It's comprised of futurist frontier technologists who like don't want to be under the ire of nation-state regulation because it will slow them down. What does crypto offer them? They offer them permissionless liquidity events with the ERC20 token. They offer them like a way to actually have open access to data, all the fundamentals I was talking about. And we actually are also an industry that moves at breaknet speed. I think there's like this cultural cut from the same cloth between the AI industry and the crypto community that's going to make

A match made in heaven, I don't want to like because there's gonna be a lot of like bullshit out there that is gonna be very bad and it's gonna like sour the crypto industry, but at least we're like 18 months away from that. It's a match made in heaven in the sense that like there's a lot of there there. There's a lot of like clay for both industries to work with, especially just like if an AI company that is gonna start up as some like wrapper around a chat GBT.

They're never gonna IPO, but they could launch a token. And I think just the permissionless access to innovation that we both want and both have is going to one is going to fuel the other. And especially the fact that we have ERC20 tokens that these AI companies can just like print, or AI founders, or like, you know, shitcoin startup founders, they can just uh ha start printing the tokens. You we've seen this, we've seen this movie before. This happens every single time.

Cool, so you're telling me every data model is gonna get a token here, David. That's what that's where we're gonna go. Ma world of many data models w and that each model has a token, maybe.

The rule of if it can be a token, it will be a token has played out very, very well in crap.

Yes. Yes. All right. Well, uh that that that's a great overview. And I think there's so much more content to explore on Bankless around this AI plus crypto theme, some of which will be more fundamentals driven. But David, when we come back, I want to talk about this in the context of a fantastic narrative investing article that I read this week and throw out some guidance for bankless listeners who are trying to get off of the fundamentals train a little bit and looking over their shoulder at some of these attractive crypto narratives out there because I think there's some some useful insights we have for them. So we'll be back with that. But before we do, we want to thank the sponsors that made this episode possible. All right, David, you ready to talk about uh investing in crypto narratives? I I hate using like I don't know if I should even use the word investing. Uh it's more like uh speculating than narratives or something else.

Well, I think that's probably one of the warnings you're about to give us. People think they're investing when they should be very aware that they are speculating.

Yeah, and that can be very difficult to sort of confuse. Um, I guess our model for this and our framework for this is there are three ways to make money in crypto. Uh one of the ways is you be it you become a trader. So you're looking at prices, you're looking at charts, TA, you're very obsessed. Your time horizon is daily or weekly, or sometimes like hourly, right? Things can change uh a lot uh to make a trade real. That's the first type. Uh second, you can be a narrative trader, narrative investor, narrative speculator, let's call it. And uh this is the recognition that crypto, at least in the short run, so we're talking about a time horizon of not days, but maybe weeks and months, is very much an attention economy. Fundamentals don't really matter for narrative investors. That's a second legitimate way to do well and make money in crypto. And the third is more where bankless skews, the fundamentals investor. It's where you sort of forget the noise, you denominate in crypto money assets like Bitcoin, like Ether, you buy the assets with fundamentals, you try every single cycle, like four-year cycles, to increase the amount of Bitcoin and ETH that you hold. And you measure this in terms of years and like even decades. By the way, that third category, basically, everyone I know who has been maximally successful in crypto has at least been in part a fundamentals investor. Those sub have dabbled with um narrative investing. There are very few traders, I think, that actually uh make money. So that's that's the framework I think coming into this thing.

I will say uh traders definitely can make money, but they take their profits and they become an investor. So like the tiers you are in the pyramid can also kind of represent your portfolio. So like, you know, 10% trade. You know, don't mess too much with that, you know, 20 to 30% narrative speculate over a longer term cycle, and then the rest you just set and forget and don't touch. And that's your Bitcoin, your ETH, your monies.

I I know that's very much been your strategy, right? And uh has previously been my strategy in previous cycles. I think this cycle I'm being a little less active on the narrative investing side. I'm just kind of letting it rise because like I'm tired, dude. I'm like tired. But

You're you're going down the pyramid towards the f fundamentals, 'cause that's good, 'cause I'm going up the pyramid towards the narratives.

Well, basically, right? But like I know your strategy is is pretty much what you you've said, which is your fundamentals investing with like the bulk of your your stack, let's call it. I don't know if that's 80% for you or something like that. But you know, the other, you know, 20, 15 to 20%, you're you're you're you're not afraid to put on some narrative trades. And indeed you do, like why not why not ride the cycles out? Um so I I guess I I will say before we get in, for for many, narrative investing is going to be an optional side quest, right? So like I don't do very much of it. David does a little bit more. It's kind of like you don't have to do it to go on your bankless journey to do well in crypto, um, but you can. And I I do want to caveat this by saying it's it's it's also kind of dangerous uh because, as we said, narrative investing is a type of speculation. You're speculating on what's going to receive the most attention in crypto. And like I will say, many of you listening to this probably shouldn't be doing it. Okay, because probably the limiting factor for you is you don't have time. You don't have time to keep up with the all of these narratives, right?

Maybe on the game.

You need to listen to every single bankless podcast, okay? Like you need to like consume it Twitter. Like it's a fire hose and cheer.

Like more niche corners of the internet than bang.

Bankless like will sound

That's not what we do.

100%. And the the other caution I would give is um some of you who go down the narrative path will will be tempted to overallocate. Like you'll be addicted to it because this is shorter term time horizons. And I'm talking about like big risk, but also big return. Like the truth is, narrative investing is the way you do a 20 to 100x in a cycle. You're not gonna do a 20 to 100x on your Bitcoin and ETS stack. You're just not, okay? You might be able to do a 5 to 10x in five to 10 years, okay? But like when we're talking about massive returns, that's why people are so attracted to this, but they they they can become addicted. So though those are the caveats that I would say uh as as we get into this. Um but now let's get into it. And and I think the first rule to understand about narrative investing is you basically you take all the fundamentals you learned and you just kind of set them aside. Because for narrative investing, fundamentals really have no place. So, David, when you were talking about Filecoin, R Weave, and Render, and uh what else did you mention? Uh BitTensor.

Yeah, but ten sure like but there's like ten others I could have mentioned.