Transcription
Do you ever wonder how crypto venture capital investors think? We have on the podcast Multicoin Capital. Although our guest is a managing partner of a registered investment advisor, nothing in this podcast should be considered an offer of Multicoin's investment advisor services or should otherwise be confused for investment, tax, legal, or other financial advice.
We've long admired Multicoin Capital, one of the earliest investors in the space, as thought leaders in the future of cryptocurrency, and it constantly has a track record of developing interesting theses when it comes to the world of internet and digital money and blockchain technology in general. You're going to want to listen to this podcast since he shares some really great insights.
[Music]
Hey Jeff. Um, thanks for being an early investor with us and uh, thanks for inviting me on the show.
Pleasure to be here. Um, hi everyone. My name is Kyle Samani. I'm a co-founder and managing partner at Multicoin Capital. Uh, we are an investment firm founded in 2017, uh, focused on the crypto space. Today we're one of the oldest and largest crypto-focused investment firms in the world managing a few billion in assets. Uh, we have two major strategies we run, a hedge fund and a series of hedged funds. Uh, we like to say our our firm only has one um asset selection strategy, which is buy things we like and do not sell them. Um, but we implement that mechanically across two different uh vehicles, a closed-ended venture structure and then an open-ended hedge fund structure. Um, one investment team that runs across both um and I'd say generally the same line of thinking across both.
Nice. And so being that early on in crypto, 2017, you know, thinking about the future, even where it was at back then, can you explain where it's at today in relation—you published a great paper on internet capital markets—and so almost like a history to ask for today?
Yeah. Um, in 2017, we saw kind of uh wild wild west capital formation in the form of these pretty shitty ICOs. um, you wrote a white paper. It didn't really have anything to do with crypto, but you just said, "Hey, I'm going to raise money for it using Ethereum. Uh, I'm going to launch my own token." Say that the currency that you need to use to uh um you use my service is is some new token. That is a very—that was very dumb—and that whole kind of thing fell apart. And so that was the uh context in which we launched Multicoin was kind of like peak ICO mania. Uh, in the time that's past, the market I'd say has evolved in a few important ways. The most important is that we have figured out DeFi. Um, and in DeFi, for those who aren't familiar, is the ability to program uh money movements on chain. Uh, the best example of that is just like a borrow lend protocol. So today the the largest borrow protocol on Solana is called Kamino. The largest one on Ethereum is called Aave. Both these systems have a few billion dollars in outstanding borrows and even more collateral than that backing them. Uh, but but in all of these systems, u there is an open open contract. The contract lives on the Ethereum blockchain, Solana blockchain, open source, fully audited, every everything is 100% transparent. And in any of these systems, if you want to lend um and earn yield on an asset, whether that asset is dollars in the form of USDC or USDT, that could be Bitcoin, it could be ETH, it could be SOL, it could be anything else, you can deposit that collateral into the contract. Um, and you will earn an interest rate. That interest rate is floating. Obviously, the more supply that comes in, the interest rate goes down. The more demand there is, the interest rate goes up. Uh, but anyone can deposit uh any asset and then once you have deposited collateral, you can obviously then withdraw uh you can take out debt denominated in any other asset. So you can you can put up dollars as collateral and borrow SOL, for example, if you wanted to short SOL, or you can do the inverse—you can put in SOL as collateral and borrow dollars. Um, and uh, the system knows what the prices of all of the assets are, updating roughly every few seconds, uh, 24/7 365. Uh, there are no holidays, there are no weekends. Um, and uh, the system runs and obviously there's like liquidation thresholds, right? So if you borrow, let's say put up $100 worth of SOL, you borrow $70. If SOL comes down 20, you know, there's some liquidation threshold, right? that like as you—the value of the SOL comes down—at some point you'll get liquidated. Um, what's amazing about this system is that there are zero humans involved in running it. Um, the only cost of using this—I mean I guess there's two costs. One is the—I mean the interest rate you're paying to the lender, but like I don't consider that a cost of the system. That's just a that's just a cost of capital markets. Um, there's a net interest margin you are paying above and beyond that. That net interest margin is again calculated and visible in real time and is flowing into a public treasury. Uh, that net interest margin, of course, is accumulating over time and acts as an equity buffer in the event anything goes wrong. Um, and again, it's fully transparent, all on chain. Uh, and then the only other cost in the system is the cost of gas, which on Solana is like a tenth of a penny per transaction or less. So effectively zero per transaction. And so like if you look at a system like this, you have something that it kind of represents the—you know, before the internet—the cost of propagating information was greater than zero. You had to print something out on a book or a newspaper and mail it somewhere, and like that came with gas costs and oil cost and a truck cost and a guy driving the car and all these things. Like we had radio and TV, but like you know, still even those were not free in the way that the internet has made the propagation of information free. And what I just described in the context of this borrow protocol, such as Camino, is uh taking the the ethos of the internet and applying it to financial markets. Um, the system is globally accessible. Anyone in the world can access the system anytime 24/7, and you can post any collateral the system supports. Um, there is a whole series of of mechanisms around risk management. Not all forms of collateral are supported. If you launched a memecoin two hours ago, that will not be eligible collateral in the system. Um, but there's all—there's four supported collateral. Anyone can deposit at any time, and anyone can withdraw any debt at any time. Um, and the cost of using the system are gas costs, which are very close to zero, net interest margin, which is healthy for the functioning of the system because you need an equity buffer, uh, and then of course the interest rate you pay to lenders. It's it's really like the the perfect epitome of of finance and capitalism.
So, what do you think is holding back traditional finance from adopting this new system?
Oh, um, I mean lots of things, but the most obvious of which is just regulatory—like crypto lives in this regulatory gray area. Um, I'm pretty optimistic that the Trump administration and that the current Congress will fix that. Um, the Trump administration so far has been very—in the first couple weeks—super effective. They have—he signed an EO that undid a lot of pretty dumb Biden-era decisions. Um, the SEC has done a tremendous amount of things already to to further accelerate that work. But the big items we really need—like if we're going to get Wall Street to fully embrace these systems—we need blessing from Congress. Um, and and there's really two major pieces of legislation that we are expecting this year. Um, the first of which uh is making its way through Congress now. It's stablecoin legislation. There's a version in the House called um the Stable Act, and there's a version in the Senate called the Genius Act. Um, we've reviewed both bills. We are pretty happy with both bills, and there seems to be bipartisan support on getting getting this through. So that that should happen this year. And when that passes, that will—I won't get into the mechanics of of of the bills—that's beyond this podcast, but it will explicitly make it kosher for regulated financial institutions in the United States to hold stablecoins, to issue stablecoins, um, and to interface with stablecoins on public blockchains.
Yeah, I love it. Uh, and on the side of stablecoins, we found this to be the game-changing use case. And just for me, myself as as a business owner, I prefer stablecoins well over wire transfers—faster, cheaper, more secure, transparency and whatnot. So yeah, can you explain the stablecoin landscape and where do you see things just in terms of adoption and this opportunity for crypto even just—yeah—to be used retail or even banking uh side of things?
Yeah, the majority of stablecoin adoption is going to be driven by people who live outside of the United States. Um, most people in the US are pretty content with their, you know, combination of banks, Venmo, PayPal, Square Cash, Apple Pay. Um, uh, you can gripe about some of the problems in those systems. Um, but like 97% of Americans are content. Uh, and the and the primary reason they are content is not the the financial rails, it's the asset. Um, US dollars are a spectacular asset. uh, if you live not outside the United States, then obviously just day-to-day ability to access, hold, and and use US dollars is substantially reduced. Uh, and that's the target customer of stablecoins. Uh, I I have I have a a question I always like to pose people, which is there are eight billion people on the planet earth. If you could go to each of those eight billion people and ask them the following question without fear of prosecu—persecution uh from their local governments—what do you think the answer would be? And the question is: if you could denominate your net worth in any asset—could be dollars, euros, yen, yuan, it could be gold bars, it could be Bitcoin, it could be S&P 500, it could be Apple. Um, you can choose—it can be a basket of whatever. Um, if you could choose to denominate your wealth in any asset, what asset would you choose? And my guess is that's between 70 and 85% of the global population would say US dollars. Um, and that arbitrage is like stablecoins uh fix that problem. Um, it's I think we will look back on this as like one of the easiest and most consequential decisions in the history of US financial policy. Um, was to uh bless stablecoins and encourage the private sector to get them in the hands of people out there. Um, this explicitly serves US interests. Uh, the easiest way to attack Russia, China, North Korea, Iran, pick pick bad actor of choice, is not via guns and tariffs and sanctions. It's to encourage their their capital base, their tax base, their support base. Make it as easy as humanly possible for those people to exit their local regimes and enter US dollars—like that. It's just—and and doing so requires zero guns. Doesn't even require coordinating with European allies. It requires, you know, signing a couple documents in Congress and the presidency and then calling tech companies and find and saying, "Hey guys, do do capitalism, do free markets, go make money, like have fun." And that that's all the government has to do.
Very cool. I follow you on Twitter or X, and they—I retweeted—you know, Stripe uh their openness to this programmability of money. So how do you—like what's your thesis on integrating with these providers and also you know starting and creating one from scratch that's built for uh you know just uh stablecoins and whatnot. But yeah, would love to hear your insight on that.
Uh, sorry Jeff, I don't follow the question. Uh, just uh uh you retweeted just uh Stripe's openness to integrating—what—like stablecoins—the programmability of money. Do you see the—like your thesis as a capital allocator—into like partnering with these existing systems or starting it from scratch and just see—like overall—what's your uh your insight on?
Yeah. Um, you know, Stripe has made it pretty clear they're going to lean on uh stablecoin payments, specifically on permissionless public blockchains. And that stands in pretty stark contrast to uh a fair bit of discourse about permissioned or private blockchains that has taken place over the last decade. A lot of folks, in particular on Wall Street, have talked about permissioned blockchains. Um, we don't really think permissioned blockchains solve a problem. Um, again, just going back to my little rant a minute ago about, you know, this the five or six or seven billion people who want dollars who who can't really access them um as cleanly and as easily as they would like. Those people are—it's obvious that those people are going to access dollars not on JP Morgan chain um but on Solana or Ethereum um and the reason is is simple—those public chains allow developers to build applications freely and easily—and for example, we have a portfolio company called uh P2P—we have another one—in—this is a group of guys—they were based in India—they've moved to Dubai—uh, and they have figured out a way to uh turn rupees into USDC or USDT um and in a way that they think that the Indian government can't stop it. I won't get into the mechanics of of how it works uh other than to to say that we we underwrote it—that we think it works. There's a bunch of kind of cool advanced cryptography they're using and a few other clever tricks and um like India's got 1.3 billion people or you know plus or minus something like that—um, and like JP Morgan chain is not going to endorse and support that kind of an application—but that developer—if if in fact they are solving a real problem—and hey maybe none of the Indians want dollars—maybe they all want rupees—I don't know—but assuming they want dollars—like you're going to need entrepreneurs like that who are going to go make it happen. Um, and as guys like that um have success around the world, they're going to build up liquidity. Um, and and there's kind of a few ways to think about that liquidity. One is what is the asset pair—meaning? Is it USDC or USDT or something else? Uh, and then the second is what are the the technical parameters of adoption and usage? Um, unfortunately, like making Solana and Ethereum perfectly interoperate—I I don't think will ever be a thing. Um, and so the system where those people adopt um is likely to be uh like is likely to gain additional network effects and and modes. And so um coming back to your kind of original question was like I framed it as kind of like permission versus permissionless. The ship has already sailed on permissionless blockchains. Um, yeah, JP Morgan chain sounds like a big deal, and I'm sure a lot of people may get scared, but in reality, JP Morgan chain is just not going to be competitive for the people who want access to dollars and for the developers who are building the applications on top of these systems to make it possible to get these dollars. There are different versions of of P2P all over the world. Um, and and not all of them are uh uh there's different in variations of them. A lot—some of them are very focused on cross-border transfers. We have another portfolio company called Eldorado focused on on Colombia, Venezuela, a few other Latam countries um does something similar. And so uh the thing we have learned is the go-to-market for a lot of these things is going to be very regional. Um, and you need to understand the intricacies of those markets. Um, in particular, in order to build your your iPhone or your Android app, you know, to onboard these users, um, you obviously need to interface with the local banking system and local fiat system in some form or fashion and then obviously interface with with the crypto systems publicly.
So, when it comes down to that go-to-market, I did speak with your team about how important it is for these crypto projects to really understand who their core target customer is. And so how do you see that just—you know—as an capital allocator when you kind of underwrite these deals and even talking to the teams on just go-to-market strategy?
Yeah. Um, one of the biggest things I've learned as being a a VC is there's a lot of ways to make money, and there's not only one way to be right, and and you shouldn't try to force everything into a single frame. Not to say that uh—I mean to be fair, the flip side of that is—hey, focus—and like know where your strike zone is and and you know stay in your strike zone. So—two sides to every coin. Um, what we've learned as as VCs—and we we've we've done—we've made more money in certain types of investing than others. We've made our most money has been um focused on pretty structured and and rigorous analysis about like technical trade-offs and system design. Um, we've done less well—although still okay—in like consumer go-to-market strategy stuff. Uh, but I'd say the biggest thing is like whenever you're talking to an entrepreneur in which you know the core of the business is they have to go out and acquire a lot of retail customers. Um, the big thing—the big things we want to hear from them are clarity of thought around the problem that the customer faces. Ideally, that clarity of thought is captured in the memo or the the deck that they send me on day one. Uh, and then even more importantly—as in our you know phone call and conversation—it's just very clear that this person—the entrepreneur that we're speaking to—has really deep insight about the core customer they're going after. Um, is—I like to kind of say—um, the people who are going to build Eldorado cannot live in San Francisco—like like that that outright disqualifies you from building that product. The same is true for the P2P product in India. Um, you have to live in those places. Um, and uh, yeah, I think that deep empathy for the customer and then ability to communicate and reason about product prioritization around that—that—that's probably the biggest thing we look for as VCs.
Yeah. Thing. How about the strategic relationship side of things and like mass adoption through platforms itself or is it more like a bottoms-up uh approach that you're you're seeing for go-to-market or—
Yeah, we're we're very focused on on bottoms-up here. I mean, I want to know mechanically how is the next incremental customer going to hear about your service? Um, and why are they going to you know adopt it? What is going to be their psychological state of mind in which they learn about the service and and is that going to be conducive or um a barrier to getting them to adopt? Obviously, the best services are ones that are inherently viral, and not all services are inherently viral.
Nice. Um, any other like strategies that you've seen for certain projects kind of break out to the masses and like what they did specifically?
Um, yeah, I mean the most important probably like adoption mechanism that ever happened was Binance was Tron subsidizing USDT transfers out of Binance. This happened in the 20 18 to 2020 era. Uh, and it really is what led to the rise of USDT um around around the world. And uh, at the time, most folks in crypto, myself included, did not realize exactly what was happening there. But um, the majority of the USDT issued in the world is issued on Tron. Um, and Tron is—for those who are unfamiliar—I would call it a third-rate blockchain. Um, to be generous, I'll leave it at that. I'd say that's a consensus opinion among everyone in the space. Uh, but they were first, and they got it right, and the users are not unhappy enough to leave. So uh that was probably the single biggest thing that happened. Binance became the preeminent exchange. Um, people wanted to get USDT out. Ethereum gas fees were higher than zero. Uh, Tron subsidized gas fees to be zero. Um, Tron paid Binance a substantial amount of money to enable free transfers of USCT out of Binance onto Tron, and the rest is history.
Yeah. So, Tron uh Binance, they all had these famous ICOs back in 2017, and I know you uh hinted at it earlier, ICO mania. Do you think we'll get back to ICO mania? Right. 2020 was about like DeFi, yield farming, 2021 was your NFTs, and then you had private sales 2022, then you had memecoins, right, and now uh like what is the new funding mechanism—they possibly could—could see out of all this?
You know, there's a lot of discussion happening in the crypto ecosystem today about making ICOs great again. Um, and I would love to see ICOs come back. Not in the form that they took in
2017. That was a very naive and bad form. But I mean, actually, the core idea here goes back beyond before ICOs. Um, the Obama administration and Congress passed the Jobs Act. I believe it was 2012 if I'm not mistaken. Uh, and they uh, the Jobs Act had a provision to create um, I believe they called them what was it? A1 IPOs or um, they created like lower-tiered IPOs in which you could raise capital. Um, and uh, they were trying to democratize access to capital markets. Um, they did not remove the accredited investor barrier thing. Um, and all of those, um, non-S1 IPOs, I again, I forget the technical term. Uh, a handful of them did make it, but generally speaking, it did not work. The fact that I can't even remember the acronym tells you everything you need to know. Uh, and so this idea has been here for a while.
ICOs went to the extreme. They said, "No rules, no regulations. Issue a white paper, launch token, go." And that's swung too far. Um, I think the I'd say most mature industry observers in the space, folks like myself and other, you know, serious investors, I think would all pretty reasonably agree on like a set of norms and criteria that are somewhere better than the Jobs Act and somewhere more protective than the Jobs Act and somewhere more protective than 2017 ICOs. And uh, I am optimistic that the uh, current Congress and Trump administration will create a framework to make ICOs great again.
Um, I I think the look and shape is not that different from 2017 ICOs, meaning a person has an idea, they want to raise money, they create some materials, they put them in the public domain, um, you know, there's a mechanism to contribute cash and then some sort of kind of follow-up, um, you know, like accountability and ultimately like product and token launch and stuff. I think that paradigm is approximately correct. Um, the primary thing we just need is just like some investor protections and some like basic disclosure requirements and controls and stuff. Pretty pretty straightforward. So basic KYC and just um, I I'm not even sure you need I don't think KYC is the problem. Um, you need the problem is is uh, I mean I'm not opposed to KYC here. You don't want people laundering money and stuff, but like the core of the problem really is uh, that like there was no accountability or requirements from the issuers, like that that was the big problem. Yeah. Dada.
Yeah. I I know in the early days of the FBI, the regulation CF, Reg CF, and then you have the Kickstarter, Indiegogo, StartEngine, WeFunder, and then you also look at the I know it's like professional, you have yourself, you have Y Combinator, you have some of the best like incubators, and then you look at what's boring and and you know, scale these successful projects. So it's really interesting just to hear the lens on like what is that structure where you know that successful companies usually come out and how it's shaped and formed. Yeah.
And to be clear, I don't think anything I'm proposing here is going to like change the path for uh, like Stripe or SpaceX. There's the best VCs are not going to get disrupted here because the best VCs do add value and the best entrepreneurs know that. Uh, so that's fine, but that doesn't mean that there isn't a place for public sales. Um, that also doesn't mean and again that that's not a mutually exclusive decision. My hope is that the future configuration is we we take on the most risk as sophisticated VCs. We have the longer time horizon and duration. Our uh capital structure of both of our vehicles accommodates for that. And uh, we invest early when it's, you know, three guys in a garage or not much more than that. And as they get polished, as they get more presentable, as they have a product to show, as they have a story to tell, then they ultimately can engage public markets and then public markets can participate.
Um, there's a lot of discussion today among amongst crypto participants about um pricing structures for you know, new asset issuance and I think today that the ecosystem is doing a pretty bad job of that. Um, I won't get into all the the mechanical stuff here, but like the simple answer is an auction. This is not rocket science. Um, and again, there's plenty of auction mechanisms to choose from. Uh, and, uh, I'm pretty optimistic we'll get to a, and it'll take a few years, but I think we'll get to a common set of norms around kind of different auction designs, auction structures, and which kinds of products and teams should choose, you know, which auction structures for for which reasons. Yeah.
Well, Dud, I think I was at BlockCon in in 2018 with Solana. Was that three guys in the garage and nothing boring than that? And and the fact that you were able to see that with foresight. You know, I always look back and uh, hindsight's always 20/20, but do you have any like advice or insights for those guys that are building right now and the best way to kind of work with uh your ecosystem or just generally build a a really good product kind of going forward right now?
You know, um, the answer is not really. Um, if you're interested in building on Solana, there's a ton of documentation on the Solana website. um, go there. There's tons of Telegram groups you can find from there, Discord groups. There's plenty of community supporters out there um that will will do their best to answer questions. So, um, you know, if you're interested in building stuff, jump in. Um, and if you're, you know, trying to to raise money for your your product, go ahead and email me. My email is kyle@multicoin.com. Hi.
Uh, in terms of these Solana builders, how important is it for them to attend these like hackathons and these like in-real-life events and just to surround themselves and get up to speed with what other projects are are building? Yeah.
Um, so the the Solana Foundation kind of formally endorses a uh a group called Colosseum um which runs an incubator and a fund. And disclosure, I'm an LP in that fund. Um uh and they run a hackathon a few times a year and actually I think they just a couple days ago announced like a perpetual hackathon or something. I I don't recall all of the details of how it works, but they are kind of like the the canonical Solana ecosystem hackathon. Uh, would strongly encourage anyone who's interested to to go get involved and check it out. The hackathons really do work. Um, Multicoin is invested in, I think, at least 10 different teams that like were born out of a hackathon, and the Coliseum guys will give you the official stats, but I I I would assume no less than 500 teams have been funded coming out of out of those hackathons. So, they they work, they they concentrate attention um from capital allocators and uh they are a forcing function for entrepreneurs to ship. So um I'm a big fan of it. Yeah.
And and last thing um, you know, AI and so even with with AI startups like there's a very high density in Silicon Valley in these cerebral valley, I think the neighborhood is it is it's called where it's just AI everywhere. Uh, yeah, what's your thoughts on on AI and just integration with crypto and what are you excited about with the um, very excited about AI. I do think it is one of the most transformative technologies of my lifetime. Um, Sam Altman likes to say the cost of intelligence is dropping, and I love that framing. If you think about what is the cost of making a human intelligent, the answer is 18 years and a lot of school. Uh, and that is obviously extraordinarily expensive. And um, AI is reducing the cost of intelligence. Um, it it is it will be looked back upon as as one of the most consequential inventions in human history. Um, love to see it will will change society in a big way. Um, have some personal investments there. Multicoin has a number of investments at the intersection of crypto and AI. I wrote a blog post about two years ago. It's on the Multicoin blog called crypto and AI for intersections or something like that. Um, and uh, you know, two years later, I am actually pretty happy with the post. Um, I haven't reread it recently, but what I can recall, I got most things right.
Um, the post specifically said, um, the most interesting area to focus on is GPU marketplaces. Um, one thing I've believed fairly strongly now for a couple years is that, um, there's been a paradigm change in the type of computation that is running through silicon. Um, the vast majority of computation running through consumer silicon historically was deterministic computation. Uh, meaning just 2 plus 2 equals 4 as opposed to like running through these LLMs and these models which are inherently nondeterministic. And those non-determination systems um require a lot more memory bandwidth compute. And uh, so my theory was, well, if we're going to inject transformer-based systems into all of software effectively, then um, probably the entire silicon supply chain um is not equipped to deal with that. And building out the silicon supply chain is um very laggy because like the time from deciding to build a fab to a fab making chips is on the order of 5 years and like 20 billion. And so uh uh if demand is accelerating and supply cannot easily step up then that means you're going to need to repurpose existing hardware in the world um in a more efficient fashion. And um the GPU marketplaces we've invested in um one is called IO. Their token is live. It's been live for maybe nine months now. Um, another one's called Cusco. They haven't have a token yet. um, these teams are hyperfocused on uh understanding there's different kinds of people who own GPUs that are that are underutilized and then figuring out um all of the DevOps infrastructure, container infrastructure and then the orchestration across hundreds, thousands, tens of thousands of GPU deployments across the world to provide a single API and a single endpoint. So the developers just say, "Hey, run this model, you know, Llama 3, Mistral, Deepseek, Rwan, whatever it is. Uh, run this model with I I need an answer within this amount of time and I want this many tokens per second. G, I don't care where the GPUs are. Give me an answer." Wow.
Um, and so all of these guys are are, you know, that's the abstraction that they provide for developers. So the developers don't have to think about the underlying complexity and orchestration of of all of the the supply underneath. Um, we think it's a it's a really big vision, really big market. Have made a few bets there. Um, I'd say the other stuff at the intersection of crypto and AI. Uh, not really seeing it yet. Um, I identified a few areas in the blog post that I basically said, you know, intellectually cool, not really investable. Two years later. I I I stand by that those claims. Um, yeah, I think the other area that like the crypto AI strikes me as is I guess there's two two that strike me as super interesting. One um is the idea of what we call a zero-employee company. And I I I wrote about this in a blog post in at the beginning of this this year. Um, and then my partner Cheyion wrote a follow-up post about it a few weeks later. Um, but basically we have a theory that um there is going to be in the next couple of years zero-employee companies uh where the company is an AI u and there will be humans above the AI and humans below the AI. And what I mean by that is the humans above the AI will act as if what I call the board of directors. um, they need to set a a mission and objective for the company, maybe some parameters maybe around ethics, norms, business practices, whatever, and then the AI goes and runs the business uh and then to the extent there are things that the AI cannot do itself, let's say it requires going to a fish market in you know the Philippines and picking up fish for whatever reason, AI is probably not going to do that, so the AI is going to need to hire you know some humans to go do that work um and then it'll come you know deliver some output and the AI will go from there. Um, we we think that this has an interesting intersection with crypto. Um, because uh we we think that the the demand for capital formation around this is going to lend itself very naturally to crypto rails. Um, in the extreme case, AI does have massive job displacements. Um, and so like if that's going to be the case, then you want people to be able to go long AI businesses. um in in a very democratic democra democratic way. Um, it's a real shame that Stripe is at 90 billion and no American retail investor has had a chance to invest. Obviously, it's the calls company. They get to run it how they want. But like I wish that Americans had the chance to go long Stripe if they wanted to. And I I I look at AI as like a a even stronger instantiation of that idea, which is uh if you think AI is going to disrupt whether it's your job or your neighbor's job or your mom or whatever, um wouldn't it be nice to be able to go along um that that business in a very tangible way. Um, so I think that's one pretty compelling intersection of crypto and AI. Again, my partner Shyon has a fairly in-depth post on it written a few weeks ago. Um, and then the other one is, um, you know, I can't remember the other one that was on my mind, but I'll leave it at that. I've been ranting for a while.
Yeah, no worries. So, how do you think about power then for AI? Just, you know, the obvious constraints, right, and via uh and now we're at like a power uh issue, but you know, is there anything on that side of things?
Yeah, I mean um, look, power is beyond so my understanding is that most of the power constraints are more around very large data centers for large training clusters. Um, inference you don't need as much power in a single location. Um, obviously you just don't need lots of power, but like this lends itself exactly to the GPU marketplace thesis that I had just identified. Um, and teams like IO and Cusco are pioneering this. Um, IO and Cusco in no way will will kill Amazon or or Microsoft or or Azure or Google Cloud or whatever. Um, but I believe that uh these decentralized permissionless GPU markets will coexist alongside the hyperscalers. [Music] Cool. I love it. I'm excited for the future. Is is there any project that you wish could be built right now that doesn't exist, you know, just from like a macro that that's obvious out there?
Um, yeah. I really want to see uh a zero-employee company. I don't even care what it does. I just like it. It It will it will be a a big like probably that'll probably be like the next moment after ChatGPT that like everyone's brain kind of like implodes. So similar to like what Operator is right now or like the Zapier that hooks into different things and just brings it all together and processes it or Yeah. But but it it in and of itself chooses what to do next. That that would be the key difference. Zapier, awesome service, but it's it's it's deterministic. It's just if this then that. Yeah. Um, it doesn't think. Wow. Um, and I think if we have a a zero-employee company that's doing like 20 million in revenue, growing very fast, I think that's a the world kind of brain explodes moment. Yeah. I love this. Kyle, uh, best way for people to follow you and just kind of support and just be part of the ecosystem.
Yeah. Um, if you want to follow me on Twitter, I'm fairly active on Twitter. Um, known for lots of fiery takes. Uh, my Twitter handle is my name. So, Kyle Samani @KyleSamani. Uh, and then if you want to email me, you can just email me. My email is kyle@multicoin.com. I love it. I love it. Well, thanks, Scott. Thanks, Jeff. Pleasure to be on. [Music]