Next-Gen Infrastructure: Decentralized Edge Networks w/ Harrison Hines and Parsa Ghadimi (Fleek)

**Speaker A:** Hello, and welcome back to the Strange Water Podcast. Thank you for tuning in. Today we're going to start a little outside of web 3. CDNs, or Content Delivery Networks. A CDN is a geographically distributed set of servers that deliver web content more efficiently. Let's imagine a world without CDNs. Let's say we have a website, and that website is hosted on a server in Virginia. Every single request needs to be routed back to that server, whether that request. **Speaker B:** Is coming from a few miles away. **Speaker A:** In Washington, D.C. or. Or if it's coming from all the way across the world, let's say Indonesia or Australia. A CDN will take the data from our centralized server in Virginia, and it'll mirror it across this geographically distributed network of servers. With a cdn, when a user requests content from a website, the request is routed to the nearest CDN server instead of the website's origin server. The CDN then delivers the content to the user, taking advantage of its physical proximity to reduce the time it takes for the content to reach the user's device. Now, the CDN model is not new. It became the standard way the Internet works in the early 2000s. Today, CDNs have evolved into what we call edge networks, signifying the edge connections between centralized servers and the individual users of the Internet. Edge networks are centralized companies that offer many services across their geographically dispersed computation network. Well, centralized for now. Today we're here to talk about Fleek Network, a protocol that is taking all of the technologies and paradigms developed in crypto and applying them to core Internet infrastructure. Edge Networks co founder and CEO Harrison Hines and CTO Parsa Gahami are here to help us understand what an edge network actually is, the implications of a decentralized edge network, and how fleek is building a solution to aggressively challenge the UX and the pricing of incumbent edge networks. One more thing before we begin. Please do not take financial advice from this or any podcast. Ethereum will change the world one day, but you can easily lose all of. **Speaker B:** Your money between now and then. **Speaker A:** All right, let's bring out Harrison and Parsa. **Speaker B:** Parsa, Harrison, thank you so much for joining us on the Strange Water podcast. **Speaker C:** Thanks for having us. We're excited to be here. **Speaker D:** Yeah, thank you. **Speaker B:** Of course. So I'm a huge believer that the most important part of every conversation are the people in it. So with that as like a little bit of a frame, can you guys just tell me a little bit about who you are, how you found, like, this weird space of web3 and ultimately, I guess, like, why instead of running away. Did you decide this is the place for you? So either one of you, like, please feel free to start us off. **Speaker C:** So my name's Harrison, co founder and CEO at Fleek. I joined the industry back in 2016. Basically, I was involved in equity crowdfunding with a project that was called Seed Invest in New York Equity. Crowdfunding never really took off the way it should have, largely due to the JOBS act and regulation that kind of, you know, prevented it from, I think, being what most thought it would be. And then when I saw Ethereum launch and the first use case that really took off were these sort of crowd sales, that's when I had reached out to Joe Lubin and Consensys and had the idea for Token Foundry, which was the first project I started in the space, which was kind of like the tokenization arm of consensus back in the ICO days. Did that for a few years and then left with a few people to start Fleek. Mainly because I think we just sort of had a personal interest on maybe not so much the decentralizing Wall street narrative inside of the ecosystem, but we're a lot more interested on, like, the decentralizing the web and Internet infrastructure side of things. And it felt like there weren't that many people focused on that side of. Of the ecosystem. And so that's really where the idea for Fleek came for and sort of why we started working on it. But I'll. I'll pause there. **Speaker B:** Yeah, awesome. Can't wait. But, Parsa, can you cue us in a little bit to your journey? **Speaker D:** So, yeah, I'm also a software engineer. I been kind of been involved with software for kind of as long as I can remember. Like, it's been more than half of my life at this point. And I kind of started to know about Flik almost three years ago, and, like, I started working with the company and we saw an opportunity at the Flick Network stage for making a decentralized CDN at first. And we kind of thought, okay, we could do way more here. And kind of. We took on it and I've been leading the technical progress we've had there and kind of excited to talk about it. **Speaker B:** Awesome. Very cool. So I want to, like, jump back to something Harrison said, which was like, this moment where you're realizing that so much of the energy is around this, like, super financial aspect of what we're building, and then, like, getting really inspired to, like, build the rest of the, like, the vision that has, like, kind of nothing to do with, like, finance or money or Whatever. And I think like, a real problem that we have in this industry is that like, to hear cryptocurrency automatically puts you in this like, mindset of like digital gold or like a new financial system, or like all of these things that have to do with money. And I'm a big believer that what we're building here is a decentralized computer, right? It is like so much more than money. Just in the same way that like, yes, you can use money, sorry, you can use servers to do high frequency trading, but they also run like the rest of the world in ways that are like, much more dynamic than that. And so like Harrison, can you talk a little bit about like that moment where you kind of right before you found fleek, where you're realizing that like, huh, there's more here than just, you know, a new way to crowdfund projects. And like, what was the gap that you were seeing? Like, why was fleek the right product to build in that moment? **Speaker C:** It was definitely not something where we understood completely what and how we were going to go about making an impact on that side of the ecosystem right when we started. Because what led it is, you know, we saw the early sort of iteration of what people were calling dapps, which were essentially, you know, if you look at the stack of a DAP, they have APIs and are communicating with a smart contract on Ethereum. But everything else about the app was pretty much still using traditional web infrastructure and cloud platforms. And so we were interested on, you know, how that side of things was going to play out. But back then, the only real technology on the, like, decentralized web infrastructure side of things that really seemed to have any sort of staying power was ipfs. So that's kind of where we started with things. But I would say as we started to build and focus on what were the use cases people were really using IPFS for. One of the main first ones was hosting front ends. And so that's sort of the pain point we started with, is just making that an easier and better experience for people. And then things like using IPFS for NFT metadata and things like that started to also appear. But in building that infrastructure to support those use cases, what we quickly realized is that Most of the Web3 ecosystem and the infrastructure powering it, even if it is sort of positioned as decentralized and censorship resistant and all these things, IPFs, for example, everyone is just running IPFS nodes on AWS infrastructure. And so it doesn't actually solve any of the problems building directly on it does A little bit in terms of, you know, reduces vendor lock in and gives you a little bit more freedom and flexibility. But for the most part, you know, when you talk about censorship resistance and deep platforming and a lot of these sort of like qualities that are driving, you know, the Web3 ecosystem to more decentralized alternatives, it doesn't really solve any of that because someone could just go to any IPFS pinning provider and tell you to stop pinning or serving these files. And if you look around the Web3 ecosystem, it's pretty much the same situation at any part of the stack. Every single Web3 protocol, middleware or app is using some sort of Web2 cloud infrastructure, either entirely or to augment themselves to kind of reach the performance level needed to sort of appease modern developers and the end users of their applications. But to start, we really first explored other options. You know, we did a lot of work in the filecoin ecosystem, we did some work in the Internet computer ecosystem. It wasn't our initial intention to build our own protocol. But what we realized after building on a lot of these, I would say, like first generation attempts at decentralized cloud infrastructure, web infrastructure protocols, what we quickly realized is that a lot of them were built on like 2016 blockchain scalability and architecture ideas, very monolithic, doing everything kind of in the same place. But if you look at the modern web and Even the modern Web3 industry, everything is modular, you know, serverless, microservice based. And so we just started to realize, even just with surfacing some of those protocols, features and functionality in the fleek platform, that those solutions just weren't viable alternatives for developers. And the main reason was they just did not perform at the level needed to even be considered an alternative to their current infrastructure setup or, you know, providers. And so that's where we kind of started to get the idea for fleek network. Because we saw a lot of new scalability approaches and blockchain architectures coming out of the sort of financial and smart contract side of the ecosystem, new consensus mechanisms to get around problems like global state or state off chain execution. So, you know, all these things, but no one was really applying those to decentralized web infrastructure use cases. Everyone was sort of trying to apply them to make faster, higher transaction per second smart contract platforms. But given the experience we had building on a lot of these other decentralized web infrastructure protocols, and we kind of had a firsthand view of kind of everything they did wrong in a sense, you know, and it is a big benefit to be able to kind of wait and see how things have went. And, you know, it's much like other more modern blockchains getting to learn from the mistakes of all the smart contract platforms and Ethereum killers that have attempted, you know, to go after that, you know, space previously. So that's kind of where we started to feel that we gained the knowledge, the experience, and had a good idea of how we could architect a protocol that you could potentially meet or potentially even exceed the performance levels that web or traditional cloud platforms are getting leveraging Decentralization, not just from like a security advantage standpoint, but could we actually leverage decentralization in a similar way that Uber does to actually outperform and make a better version of the current status quo for cloud infrastructure and edge computing and this sort of area of the stack. So I'll just pause there. But yeah, that's. **Speaker B:** Yeah, man, you just. This is going to be a tough conversation because you just like Unleashed 18 Things I Want to follow up on. But I guess while we're still in this beginning part of the conversation, I think it's probably worth kind of unpacking the problem space a little bit more. So maybe Parsley, you're the right person to talk to this. Or I'll throw it back to Harrison, you guys tell me. But like, can, can we just walk through the state of being like, if we don't have any decentralized, let's call it like cloud storage providers. Like, first of all, what are we looking for for these cloud storage providers? Second of all, if we don't have decentralized options, like what are the risks that we're up against? And like, why does that of make all the value that we're creating in blockchain a lot less useful? And then I think you made it very clear, but maybe just touch on again, like IPFs to people that are not in the weeds. Sounds like it's already solved this problem. Filecoin sounds like it's already solved this problem. But like, why? And maybe it just boils down to, as you said, the technologies were still too young then to do this right? But why did the. Do the existing systems fail to like, achieve all the properties that you're talking about, which is both security and performance? **Speaker C:** Just taking decentralized storage, for example, I think the main thing that's holding the usage of decentralized storage back is probably a combination of it being new. So, you know, for a traditional just software company, that's probably holding things back, but I think largely it comes down to like Performance and developer experience. It's just, it's not super easy to use these alternatives. And for the most part you still are leveraging Web2CDNs and other things to make it performant. So it sort of kind of like counteracts the, the, the pitch where it, it is. But I will say that those I do think have found a little bit of product market fit arweave especially you know they do something net new with permanent storage that's pretty cool and novel and I think it's shown that that has some, some product market fit with, with filecoin as well. I think on the, on the decentralized storage side, you know, I do think that there, there's merit there and, and that those protocols should exist. What I would say is, is the downside if those protocols don't exist for just like the storage vertical of cloud infrastructure is that like the same inherent risks that you have with banks and centralized financial infrastructure, you basically have the exact same risks on the centralized or corporate controlled web infrastructure side of things. Mainly you know, the risk of censorship. You know, the same way your bank account could get censored or deplatformed or shut down. That's happened with cloud infrastructure already. And, and with the direction things are heading, you know, it seems like that might just heat up even more. There are things like data abuse, you know, tampering, you know, just reliability. When you have, you know these central points of failure, you have thing a lot of corporate nonsense around, you know, not sure what if you know pricing is going to increase on you or different things like that. So I think it's, it's much along the same value prop that like decentralized financial infrastructure like provides as opposed to centralized financial infrastructure is just the more transparent, permissionless, you know, you don't have to sign contracts, you don't have to talk to people, you don't need permission. You don't have to worry about getting sort of like the rug pulled out from under you. Like it has happened with you know, people who have built on top of Facebook's API or things like that. So I think that's really where the value is is, is you know, if, if you're using for any meaningful software, I think by the end of this decade it's going to feel like an unnecessary use corporate controlled cloud infrastructure as opposed to decentralized alternatives assuming they could get the cost, performance and developer experience on par or superior to Web2 alternatives. But where I would say, I think where we noticed other protocols have maybe struggled is everyone is trying to do everything. And so I think, you know, when you look at a lot of these other existing protocols and you know, some of them have started with storage or have storage as an element, and now they're trying to add things like compute or other services additionally within the same network. But they were architected as a single purpose solution to start, kind of. And the way they're architected and the redundancy required to meet those use cases doesn't really match like what you would need in terms of an architecture and a network layout and an incentive system like to meet like a cloudflare use case. You know, that's all about high performance, cdn, edge, compute, all these things when you're built on more of like a monolithic, you know, EC2 like structure and you have to have, you know, multiple nodes coming to consensus. So like for example, on the consensus side alone, a lot of these protocols are just bottlenecked on response times due to that where, you know, you can't return a response. In Ethereum's case, you know, if you're using a chainlink function, it's 12 seconds because you got to wait for a block time. You actually have to wait for two block times. But if it's on something like one of these other decentralized web infrastructure protocols, you know, maybe you have to wait minimum two seconds. So yeah, it's faster. But two seconds is, is a lifetime in terms of the Internet and the importance of performance and speed on the decisions of users and the applications they use. 40% of users won't use an app that takes more than three seconds to load. So like, if your website loads in one second, in three seconds instead of one, your probability of a bounce increases like 30%. And if it's loads in five seconds instead of one, your, your probability of a bounce increases Like 90%. Amazon says for every 100 milliseconds of latency added, they lose 1% of total sales. So that's billions of dollars they're losing. For every 100 milliseconds of latency added, that's 1/10 of a second. So everyone is, is pretty much like trying like every major infrastructure decision like this decade is largely going to be driven by performance and like trying to reduce latency because of how important it is. And so that's really where we saw that a lot of these other existing options just had these unnecessary barriers that were going to prevent them from ever being able to meet the performance that is going to be required based on just what the situation is in terms of performance using traditional cloud platforms and apps today, and they're certainly not going to take a step back. The only reason they would switch infrastructure is if you can give them equal or better performance because it's just, it's not something you can sacrifice on as a software company or. Yeah, any sort of, you know, web application or app. **Speaker B:** Yeah, man, I very much take your point and like, couldn't strongly, couldn't more strongly agree with you that like, people are not going to switch if it doesn't, like, at the very bare minimum, keep their life like, just as easy. And like, that's even a very high hurdle for change if like, you're not even really making improvements on the things they care about. And I want to spend like the bulk of this conversation talking through like, what is fleek? And how are you building something that is both decentralized but not like, held back by a lot of the, the things that, that you're, that you're gesturing to with other protocols. But I guess before we get there and when you're talking to people like potential customers for fleek, like, I just got to ask man, like, what is the appetite for like the credible neutrality and for the decentralized of this? Like, is that something that is kind of like a nice to have? Is it something like when you're buying like your chair, you like to see that this was made with green materials. But like, let's be real, that's not really what people are making their decisions on or whatever. Like when you're going and selling your product and selling decentralization, what are, what are developers actually buying? And like, do you see that like the focus on what this industry is supposed to be about, like, do you feel like that is legitimate and is that like changing over time? **Speaker C:** It's definitely changing for us and it is a great question. But like our home, our, our, our messaging and, and, and like the homepage and, and just what fleek is will look a lot different in a month from now. But we are like, we are not really planning on focusing on the decentralization advantage as much, at least not right now. It is a nice tab and within the web3 and like ecosystem, it is appealing. But I would say the cost, performance and more traditional like merit that we believe we can provide as a, as an infrastructure alternative is much more of our focus going forward. I do think the decentralization and censorship resistance and those values and the permissionlessness and you know, not needing to deal with any of the corporate nonsense and stuff, I do think over time that'll become more important. But I do think if you rely on that now as your main unique selling proposition, that is a very, that's going to be a very tough time. And you might, if that's your only advantage, you might not make it till the end of this decade or whenever that becomes enough of an advantage by itself to, you know, be like attractive to developers. But our, our approach going forward is we just think we can provide a better alternative or a better option for any type of software in terms of the infrastructure needs and where you go for, for those infrastructure needs as opposed to traditional cloud platforms. But our approach is to really focus on solving problems for developers, the large majority of which are related to the cost for using these cloud platforms. 75 to 80% of IT costs are attributed to human related things. And so that shows up in a lot of these cloud service costs like AWS Lambdas are marked up maybe 5x above the actual cost to run them. And it's because of a lot of that human bloat for, so for a lot of the same reasons like Uber was able to make it economically feasible for their model to work by not needing to hire, you know, managers and regional managers and drivers and stuff. By decentralizing that and just letting the, the, the incentive system allow you to kind of operate the network algorithmically. When you could cut a lot of that human bloat, you could get the cost way closer to, to the actual electricity and hardware costs that are the, you know, essentially the inputs to provide these web resource or cloud resource outputs. And then so that's really like your question is spot on. And it's something we've thought about a lot, especially recently. But yeah, you know, I think our stance is similar to Uber. They really focus on the product and just because they came up with a novel new approach to the infrastructure layer that allows them to provide that product and advantages over existing options for, you know, getting a car. But it's like the, the decentralized driver network is very much an afterthought for someone who just is interested in using Uber. And I think that is the approach that it'll take to win. Because what we feel is the issue now with a lot of other Web3 infrastructure projects is if you go to their home pages, if you're a normal web developer, you would feel like you're in the complete wrong place. It's talking about airdrops and nodes and wallets and tokens. There's nothing on them that actually speaks to the problems they're solving or the advantages they offer for things that developers actually care about, which is, you know, is it performant, is it reliable, what's the cost and is it a good developer experience? Like that's, that's sort of where we're going after is you know, like the number of developers in the world doubled in the last five years. So you have this new younger class of developers and they don't want to use, they don't want to have to get a certification just to use a cloud platform. They want to use something like Vercel or Render or netlify or any of these more modern ones where you just link your GitHub, you don't have to worry about configuring anything, you don't have to worry about DevOps, you don't have to worry about infrastructure maintenance. You just link your GitHub, you deploy it, you get all the performance, SEO benefits, everything out of the box. It's as easy as possible. And that's where we think things are going. And so that's really where, yeah, we don't like, we do think the decentralization angle is attractive and with the web3 market, that definitely is something you know, we think is an advantage we have for them. Cause a lot of them for rollups, L2s, all this type of scaling infrastructure, all of it runs on AWS today. And you know, besides the cost concerns, there are definitely like future proofing concerns where it brings up liability concerns. If you're operating this roll up infrastructure in your own AWS account, how do you know you're not gonna, you know, the regulation's not gonna change the, the same way it's going after wallets to say, you know, hey, you're operating this infrastructure and you've got a options exchange and you're not licensed. So there's those concerns, there's censorship, all these things that we think are attractive for fleek Network. But the Web3 space only has 25,000 developers. There are 25 million developers in the world. And largely the Web3 infrastructure options today are not relevant at all to that much larger pool of tradition or just regular developers. And so our approach is we think that we can provide infrastructure that is not only attractive to these decentralization and on chain sort of niches, but we think the infrastructure we're building and the, and the, and the services we're providing should be valuable to all developers. And so we don't want to unnecessarily pigeonhole ourselves with all this decentralization and blockchain terminology, we would rather strip that out, focus on just the developer experience, not scapegoat anyone, not, you know, have this unnecessary bifurcation between web2 devs and web3 devs. We're just building infrastructure and services that are useful to all devs. Everyone needs Hosting, everyone needs CDNs, everyone needs serverless or edge compute. All these things that make up the modern stack do not differentiate between Web2 and Web3. All developers use the same services offered by cloud platforms. And so those are the services we plan to offer that are just, if not more performant at a better price point and with none of the corporate nonsense. All open source, all permissionless and with seamless dev X. And so that's what we hope and we think if we can nail that, that that value prop should speak to a much larger portion of, you know, devs than, you know, just the, the small but growing web3developer market. **Speaker B:** Yeah, man, I really love that frame or that analogy to Uber because, like, it is spot on to say that they built this like, decentralized network of like, nodes that are able to like, it looks a lot like crypto if you put your crypto glasses on, right? But from their perspective, like, that is just like kind of a emergent property of like, the solution that they built. And the solution that they built came first from like, let's create the best customer experience. And like, how do we just like, find a way to deliver what people are asking for? And I think like, for, for really every single product in crypto, but especially one that you're building that like, is really just like, crypto is like one customer that isn't even more attractive than the other customers that might use like a decentralized cdn. It is so smart to like, focus on experience, focus on like, why customers would actually use it. And like, we just all in this space who believe in crypto, like, need to just like, man up and look in the mirror and like, understand that like, people are allergic to the like, words that we put out there because, like, to them they're associated with SBF and do Kwon and Mt. Gox and like, all of this, like, crazy stuff. And like, I do think that when developers look at a website that is about, supposed to be about a cdn, but all it says is trustless, like, that sounds like bullshit, you know, and like, even if it is trustless, even if it is decentralized, like, it just sounds like you're trying to like, trick somebody into something using words that like, don't really mean anything. And so I think that's a lesson for everyone. And this isn't a pitch meeting and if it was, I don't have any money to give you. But like that is the answer you just gave is the right one, which is like, I don't even care about this noise that you're making. Like what we care about is serving like customers and like doing something that other people don't have like the capabilities of doing. And so yeah, kudos. **Speaker C:** Thank you. **Speaker B:** Cool. So let's talk about what fleek is. And I love like that we're like using this opportunity to shift our perspective a little bit. So can you tell us what fleek is and what you're building less from the standpoint of blockchain world and more from the standpoint of like how does a CDN work? Why is it slow? Like where are the opportunities for you guys to do something that is like serving needs that aren't met today? **Speaker C:** If people have heard of us before, they probably know fleek as the current sort of like hosting platform where they use it for IPFS hosting. **Speaker B:** And, and sorry, just to interrupt the. I am a user of fleek and I use my.eth.limo address to have like a little link, like custom link tree website. **Speaker C:** Oh, awesome. That's awesome to hear. So yeah, that is, that is what most people know us for today. But over the last two years we have been building fleek network which is essentially like an edge optimized cloud platform. And so if you look at just like trends in, in the cloud or web infrastructure space, like moving to the cloud was like the big thing to do the last two decades. This decade is all about moving to the edge largely because people are chasing performance gains. So a CDN is one service you could build on like an edge cloud, but serverless compute. Edge compute. If you look at just the modern web, pretty much anything can be built using like a serverless framework. Like pretty much everyone's back end is lambda functions. And so like that's really where fleek as just like really trying to be a very performant cloud platform. And so we, we leverage it's, it's a perfect use case for a decentralized network because all the work is very localized. The whole concept of an edge network similar to a cdn, it's building off the primitives of a CDN which essentially said, you know, in the past maybe you had a website and you hosted it in aws US east, but then everyone all over the world is, is, is, is querying and Pinging and requesting and files are being served from that one location. So everyone with CDNs figured out, hey, we could just locally and regionally cache these files in multiple places, much like physical distribution, and then just serve it from the closest location to the user. So you have a faster round trip time and load times will be faster and users will have a better experience. So then after that was successful, so CDNs serve a very high percentage of all Internet traffic today, especially with the growth in video and you know, like more multimedia, like use cases of the Internet. But then people started to realize, hey, maybe I could put more of this stack on the edge and get those same benefits, but for compute. So that's where serverless compute and these things kind of came about. And so Fleek is really just an edge optimized or just a super fast cloud platform. And similar to AWS or cloudflare, these other platforms, we have core infrastructure like a physical infrastructure layer, not too different than AWS having data centers. The main difference is they're not all corporately controlled data centers. They're just different web infrastructure operators around the world that we call nodes. Then you kind of have these core resources and service layers where these nodes are providing resources to the network, compute bandwidth units, you know, other things that we might add in the future. And then you have this services layer similar to AWS or these other platforms. You package those resources into different products, services, use cases that are useful to developers, CDN, JavaScript, runtimes, different flavors of compute, databases, things like that that you kind of use as this abstraction layer on top of the infrastructure layer. And that's what developers interact with. And, and so we have the network, but then similar to Uber, nobody interacts with the network, they interact with the application. And the interface that sits on top of the network abstracts it into a beautiful user experience. So all you have to do is click a button, say I want a car here, and the rest kind of happens for you behind the scenes. So that's where the Fleek platform is really just viewed the same way. It's just an interface that sits on top of the network. But an important distinction is like fleek does not store data. The network, we just decided to let you use whatever storage layer you want. So if you're a Web2 company and you just want the performance benefits of Fleek, but you don't want to use decentralized storage, you want to use S3, we don't care. That's completely up to you. If you want to use filecoin, if you want to use Arweave, that's completely up to you. But similar to these other protocols and where we feel they made mistakes is we feel it's an unnecessary lock in to force them to use like the native storage of a protocol versus just letting them use whatever they want and keeping options open. So that's why we're able to have such high performance is because we've made some design decisions where we said we don't have to do everything. We can pick and choose what we want to specialize in and be the best at and let these other protocols handle storage or let these customers use Web2 storage or whatever they want. And so that's really where the fleek platform is not too. It's so fleek is a modern developer platform and the idea is as we continue to add new features and functionality to fleek network, we will just surface new features and functionality in the fleek platform to kind of give that full stack developer experience where we're meeting the needs of modern full stack like just infrastructure and app developers and the things they need, they need serverless edge compute, they need storage that'll be powered by other protocols and storage options that we surface. On the compute side, there will be a range of flavors of compute that people could build on fleek network that will surface database options. So now instead of having to use AWS or other cloud platforms, the idea is that someone could just come to fleek and be able to find those same services or alternatives to those services and just make it super easy and as low friction as possible to migrate and get the benefits of this new infrastructure without having to change behavior or do anything differently. Everything on the network is priced in USD stables, so you could still pay with a credit debit card. We can facilitate the payments and stables to the protocol under the hood for you so you don't have to know there's a blockchain, you don't need tokens, you don't need a wallet. It is exactly the experience of any other modern developer platform. But just using this new infrastructure layer under the hood that gives performance cost and you know, other sort of like, you know, no corporate nonsense, grouped benefits. So yeah, that's pretty much how I would describe it. It's just a modern like fleek similar to Uber. Like if you encompass it including the network and the sort of platform similar to how Uber is kind of their app and their network, it's all just part of what you feel when you think Uber. You know, it's a ride sharing app, but that Also includes their decentralized driver network. But you don't really say that when you say Uber. You just say it's a ride sharing app, you know. And so similarly, fleek is just a modern developer platform, but it also has this unique infrastructure layer and fleek network that kind of is what makes it, you know, unique or differentiated. **Speaker B:** So I think, like, just in phrasing my question, I age myself in terms of like, when I stepped out of like that truly technical world just by calling, I thought that CDNs were still basically a synonym for edge networks. So I think like, what is probably worth, like, just like covering a little bit for myself and for the audience is so like, what is an edge network? And like, why I really want to pick that apart is because, like, what you talked about is Fleek is an edge network that allows you to pick your, like under the hood, like let's say your data storage thing, your data storage service. And so like, in my head, edge networks are CDNs. And so like, essentially what these edge networks are doing are like, serving as outpost from like the big centralized, like aws, like Source of Truth, but serving as outposts that are like, physically closer to the people that want to use it. And so like, I'm a little bit having trouble wrapping my head around like, how you're able to introduce modularity into your edge network. But hopefully my question or my confusion makes sense and hopefully you can clear. **Speaker C:** It up for me 100%. I would say an edge computing or an edge network, like, is essentially, it's a few things, but you could think about it like a CDN with additional capabilities. So like, we kind of, we say this sometimes, but if you were how the modern web works today, if you were to build a cloud platform from scratch today, if you were to rebuild AWS from scratch, you would probably start with a cdn because it is the most foundational and common resource or need across all modern web services. You just want fast movement of bits of data, no matter what it's really for. So if you just look at how serverless compute or edge compute works, the way it works is it's stateless. So you're storing the state somewhere else. And so that could be a database, that could be a file storage layer, that could be wherever. But then because you have the cdn, you're able to cache popular files or data. And so what it enables you to do is, yes, you could be modular, where this system can work with any type of data the same way any serverless framework can. You can let it work with any traditional database. You can let it work with any file system. All it needs is an HTTP endpoint and you know, sometimes environment variables and then it can go fetch the data. But the way a CDN works is it's the network is intelligent where it knows what data is popular so it just caches that. So instead of just caching files, it can cache whatever data is necessary for the use cases and infrastructure you're running. So if you're running serverless computer edge functions and it's, you know, and it needs certain data every time, the network doesn't need to go to the storage layer every time. It can just it exists on the network because of the cache layer and the way that these services work. But that's really like, like what's interesting is and like just I think it's helpful to understand edge computing is a big trend in web2 but all these edge platforms, all it really means is they've went from like a one server setup to like a multi server setup and like geographically distributed setup. And there should be some sort of geographic awareness. So similar to a cdn, if you're using a serverless function it's running on all these locations. So instead of just caching a file on all these locations, now you're actually caching the function. So you can any node could execute it and if they need data to execute it that they don't have, they can go get it from another node or the underlying origin storage or database layer where the data actually lives. And so that's how the modern web works. And where these edge layers come into play, it's like the super performant execution layer. But what makes it super fast is you don't have to carry the data with you. And that's really where state bloat and all these problems come in with smart contract platforms where every node has to carry all the historical state forever. And that's really where you start to slow down and get bogged down. Especially as you scale when you're stateless. Pretty much you don't have those concerns because there's no state growth. We don't that's stored on other networks or other change much. In the same way if you use AWS lambdas you're still using a DynamoDB or AWS S3 and things like that. And to clarify, fleek network might offer some light storage in the future. It's very common, like Cloudflare introduced R2 but it's more of like a CDN based storage and it's Very performance driven. But going back to that point about Web2 edge networks, right now, edge networks in Web2 are typically sub 100 location and they're really just setting up infrastructure in different AWS data center locations. And the reason for that is it's not economically feasible for them to do it any other way. So edge computing today in web two is not real edge computing. It's basically gone from one server location to 30 server locations. And that's where fleek network, we really feel has an opportunity. Because it's the same Uber problem. If you want to go set up all the infrastructure yourself, it's not economically feasible. So all these edge platforms are just stuck needing to deal with the locations they're provided by other cloud platforms. But then that puts them at a baseline cost of whatever the AWS cloud platforms are charging. So then their costs are even a markup on top of the already high AWS costs with fleek network because we are decentralizing that layer similar to Uber. Now anybody who sees the incentives of running infrastructure and just like Uber, it makes sense to them and they think it's a business opportunity, could run a node, add that to the network, that improves the network. Because now we have more locations. And compared to these 30 location or 200 location edge networks or CDNs, we should have an order of magnitude greater coverage, if not more. And so we'll be in more locations closer to the end user in more instances and in more requests than a typical cloud platform. Maybe in New York or big cities where these cloud platforms run infrastructure, the performance is similar. But as you start to get out into the fringes and locations where they don't traditionally have data centers, but where a lot of the Internet's growth and new users are coming from, then yeah, there should be a lot of instances and services where we could outperform Web2 cloud platforms. But again, that's largely due to us just having sheer numbers and better just coverage than these ones that are sort of pigeonholed and forced to build on top of aws. So actually these other edge platforms, like a Vercel, we don't view as competitors. Those are our ideal customers. Like our, our goal here is to provide an infrastructure that a platform like that, we could say, hey, instead of doing all your serverless and, and compute needs in AWS Lambdas or cloudflare workers or whatever it is, you could come here and, and we could provide you an alternative that will lower costs so you could either make more money or pass those savings on to customers. You don't have to sacrifice on performance, you might even get better performance and you don't have to you know, worry about any of these other concerns that you might have to with, with corporate controlled infrastructure. So that's kind of where, I don't know if that answered your question about edge computing, but it's really just like a multi server geo aware infrastructure setup. **Speaker B:** Yeah, no. Okay, let me throw this out there and like tell me if this like understands what you're trying to do appropriately. Right? So the way that traditional. Traditional, right, like 3 years old at this point but like edge networks work are that. And let's imagine an ideal edge network that like not only has unlimited budget but can like stand up data centers like wherever they desire without a problem. Right. So the idea is that what they're doing is building these data centers around the world so that they can like be physically close to as many people as possible to like deliver the same content across the entire network. Right. And what they do is as like a corporate centralized entity, they're responsible for coordinating between all of these servers and making sure that a request on this one in like some remote province of India matches the same thing that comes out of the data center in New York. And what your like opportunity and understanding is is that like this looks so much like a decentralized network, but it's just not decentralized. And what if we took like the things that we're learning out of like blockchain world and out of smart contract world and out of like crypto economic incentive world and use that to say let's build an edge network but that instead of having a centralized entity we come to consensus together through blockchain. And so I guess is that kind of getting the vision correctly? **Speaker C:** Yeah. The only tweak I would make is the nodes in any edge network it's not just serving content, that's the CDN part. But now with edge networks you can also handle the compute needs for these applications. So you know, maybe it's a simple function to auth someone into your app or you know, just do a simple calculation. Like serverless compute is probably the dominant or choice in terms of compute. Like instead of going server based and using EC2 most modern devs just for all their compute needs use serverless, not all the time, you know, server and like, like we're not trying to be the be all end all. But for serverless and you know, edge compute the main benefit are very similar to the benefits of a cdn. You could just have these state, stateless functions that can essentially just run in multiple locations. And this way you get the performance benefits for the same reasons you get the performance benefits for a CDN and regionally caching and serving content. But now you can apply it to other types of web services and infrastructure such as compute or in the future database needs. Now you will, if you are building a big database and you have huge needs, you probably wouldn't put that on the edge but for simple things like authing into an app or you know, a form on your homepage or things like that, that's where you know, edge databases and key value like that, that's where there's very big benefits to running that, those types of things on the edge because it'll definitely get you better performance than more traditional setups or one location, you know, type. Yeah, setups. So. But yeah, I think your understanding is completely correct. **Speaker B:** Okay. I was trying to keep this like in the frame of Web2World because I thought that was a like compelling way to like look at this crypto project. But I'm to pivot hardcore into some like pretty deep blockchain stuff because like what I, the, the more you talk about what you're building, the more struck I am by that. This looks just like the Ethereum, like long term roadmap. And, and when I say that, I mean I, it is like so obvious to any of us that like all of these blockchain systems are coming like are, are convergent on the same scaling method, which is like creating some sort of base chain, then like pushing execution off. And as you explain fleek to me, I'm just so struck by okay, think about the Ethereum base chain. And yes, it has some minor amount of compute and yes you can store stuff in there, but you probably wouldn't. It's too expensive. But we have these execution chains that are enabling applications to be developed and leverage. I could keep going on that analysis and I think there's definitely something there. But my question to you is like when you're building this, like you're building crypto technology into like an old problem and how much do you see yourself like competing with like the other L1 blockchains? Do you see like fleek being sort of like a, a base layer by which like your token could be used as money, by which like other people like come in and build applications on top of or, or do you see this as like kind of a different thing that just if you squint looks the same but like, because it's solving different problems like It'll never be fleek versus alt L1. It'll like these are just two different categories. **Speaker C:** I would say definitely a different category. What I would say is we don't view ourselves as competitive to any existing Web3 protocol or L1 chain. We view ourselves as a very needed augmentation layer for all of them. Because like, yeah, like there are things smart contracts are way better for. Like, because we don't store state. So you know, if you're building an NFT or you know, like a token or you know, something with value, those things make way more sense to put on a smart contract platform. But what does, especially a strong settlement layer like Ethereum. But where like we see ourselves fitting in is like exactly what you mentioned. You have all these, these approaches to off chain execution and compute and infrastructure that eventually settles to ethereum. But every L2, every roll up, every approach to like these more modern, you know, approaches to building chains and, and scaling solutions, all that infrastructure today runs on aws. And that's the thing is like, so now you, like if you look at rollups and L2s, they're essentially virtual blockchains. So you don't need your own validator set. You're kind of just running the infrastructure in one server on aws, doing all the execution and then rolling it up periodically to Ethereum to save on execution costs and gas costs. But all that infrastructure running on aws, the biggest pain point for rollups right now is the cost to run these in this infrastructure on cloud platforms. So running something like a ZK Prover, I just saw a tweet the other day that the cost to run a ZK Prover on cloud platforms today is around $10 million a year. That I don't, I don't know how accurate that is, but I have heard and in conversations we've been having with projects who run ZK provers, it's definitely in the six figure or above range per ZK roll up. And so that's where we, that's the part of the Web3 ecosystem in terms of the stack we're trying to play is for and also for a lot of these networks, like if you see the Alt L1 to L2 migration, the whole reason for that migration is it was very, it was much more cost effective and also way faster time to market and way easier to maintain an L2 or a roll up as opposed to an alt L1 chain where you need your own validator set, you have to pay for your own security budget and it's not at all how Web2 works. Everyone rents their infrastructure in Web2. Very rarely do you see anybody setting up their own infrastructure. Now in Web three you have everybody trying to build their own infrastructure layers before they even have product market fit, before they even launch a product. And it takes them years to get to market. It's extremely expensive. It takes a lot of specialized knowledge and skills to do it. You have to know things like consensus, networking, P2P. It's just, it's very hard. And so, you know, where we kind of like to think of ourselves is like you're seeing this trend of like new protocols offering things as a service, you know, like eigenlayer offering economic security as a service. But it's still, you still need to build your own network and node software to build an avs. So like for us, for example, we're having conversations with a few projects and building AVSs and fleek network is an interesting alternative for anyone who's considering building decentralized middleware infrastructure. Because now in the same way you can sort of rent economic security from eigenlayer, you could rent very edge optimized geo aware node infrastructure and just pay for usage. So instead of renting servers to run nodes, you could just, you know, run that stuff on fleek network. And so what we're starting to see is, and where we're trying to focus is we want to, if you look at all these other protocols and where the infrastructure actually runs, a lot of it runs on these, these, these traditional and centralized cloud platforms. And that's where we think we can provide a better option and alternative for not everything, but for a lot of infrastructure and a lot of use cases. We think fleek network offers a very interesting alternative that is way more performant than what you're going to be able to do building your own network and having to build all these performance optimizations into your network yourself. But the way we see it playing out is maybe you wouldn't build your full thing on fleek network. Like I said, you're going to need to store data somewhere else. We facilitate that and make it very easy to build apps and have a file system and storage to access. But it's similar to how you could pick your own DA layer for building a roll up. We allow you to pick your own storage layer. But the idea is that for parts of your stack, we feel very confident that we can offer you an option to, instead of running all of it on AWS or in LA does or whatever that parts of that stack, the cdn, the Serverless Compute, the hosting you need for your front end. These web services we should be able to provide. And you most likely as a protocol or middleware project, need that infrastructure yourself. And every application building in your ecosystem needs these things. Everyone needs hosting, everyone needs cdn, everyone is using Lambdas or some form of serverless or Edge Compute. You know, things like Next JS app hosting. We're about to debut at the end of this month that we're the first protocol that could actually performantly host and serve a full Next JS application. So those are the types of things where it's like we're doing things that no other protocol can do. And so that's really where, yeah, we don't view ourselves as competitive. We don't see anybody really going after this part of the stack. But yeah, it's like if you look at the modern web, it's all these different microservices accelerated by a shared performance layer. Cloudflare being one of the most popular. But in Web three it's like now you have all these different microservices facilitating and serving all these different parts of the stack, but you don't have the shared performance layer in Web3 that is largely Cloudflare and AWS today. So that's really the part of the stack where we think we can, you know, service that part of the market. And that doesn't really step on anybody's toes because all these protocols are using Cloud for an AWS to serve this part of the stack today. So that's where we're really trying to focus and add value. **Speaker B:** Yeah, man. And it makes so much sense. I'm not trying to say necessarily that like when from my perspective it looks like Fleek is competing with any Altel ones. Instead what I'm saying is more that like over time I think that a lot of L1s, including Ethereum are going to look a lot more like what Fleek is building today. And I think that as like we move forward from thinking like L2 scaling is just let's create another Ethereum and stick it in like and those get like more exotic and interesting and the ZK VMs like allow just more and more arbitrary compute. It's just like again, I mean this more as a comment to like all blockchain design seems to be pretty convergent and I think there's something like magic about that. Like we're onto something 100%. **Speaker C:** We took a lot of inspiration. I mean I came from the Ethereum ecosystem. I love Ethereum as a protocol. But I just think two very different network architectures like SWE is actually like we use NARWHAL and bullshark consensus that SWE came up with, but it's just a very good consensus algorithm for an edge network use case where you don't need a global state and to come to global consensus, especially in an edge network where the work is very localized. So you could have each node sort of operating almost as its own, sort of like roll up and it's doing work and then periodically it's submitting it to the chain. But we use these cryptographic primitives that Parsa has come up with that essentially allows us to do that in a way that offers pretty high, you know, guarantees that, you know, the work being performed is actually what, you know, is anticipated. So for example, all bandwidth on fleek network is verified. So we stream all the data. So if any one bad bit of data is served, it's known in real time and it's prevented before it even gets shown to the client or the end user. Yeah, device. So those are the types of things where we've had. And Parsa has come up with some very novel solutions where if you're building like something like this in a decentralized manner, the biggest issue you have to overcome is essentially that you have new added overhead that a centralized version of this does not have. Where they control all the servers, we do not. These are all random operators. So there are things we have to do that a centralized version of this, like a Cloudflare does not have to do. But that is where Parsa has done a lot of amazing work with the fleek network team to reduce that overhead to like fractions or single digit milliseconds. And so that's really what enables us to do this at scale and at a performance level that could potentially Meet or Exceed Web2. But that's really where the struggle has been for other sort of decentralized web infrastructure protocols that have come before in terms of just not being set up in a way that even lets you potentially reach that performance level. **Speaker B:** Yeah, no, awesome, man. And I would love to dig in on the back end and what the minutiae of what you're building. Maybe on the next pod just for the sake of time. But I think the best way to go out is normally I give you guys an opportunity to just tell people about why they want to use your product. If they're interested. How would they get started? But I would like to ask this question in which. Please answer those things as well. But really for my own like education. And it's like both talking about today and also imagining like 10 years in the future and more of an end game state. Like you talked a lot about how what developing like a roll up is today is like basically like taking on all of the responsibility for creating it and then running it on aws. And like not only is that a pain in the ass, but it's incredibly expensive. Can you talk a little bit about like what is that experience like on fleek? Like why is that like easier, better and like especially when we're, we're thinking on the long term, genuine question, like why, why is it cheaper on fleek than it would be in aws? Like what do you guys do that's able, that you're able to provide like better costs? **Speaker C:** Yeah, that's a great question. And really it, like I mentioned earlier, 75 to 80% of IT costs are related to human attributed things. So we think if we could run the system algorithmically and use the incentives of the system to get rid of most of that human bloat, that that should bring the cost down a lot closer to the hardware and electricity cost to provide these resources. But if you look at other networks, like I use Pocket Network as an example. Pocket Network is like a decentralized RPC network. So they do what Infer and Alchemy do, but in a, in a decentralized setting. And so if you look at their costs compared to infera, they're 80% cheaper around. The reason for that is because Infera and Alchemy are built probably largely on AWS or traditional cloud platforms. And so they're just needing to pay those costs and mark those costs up to make a business out of it. Pocket Network is largely run on bare metal. So they're node operators because they're economically incentivized. Similar to, you know, Bitcoin mining being mostly ASIC based. When there's money to be made, people seem to quickly optimized for, you know, making the most money. And so Bare metal is a lot cheaper and a lot more profitable to run a node for Pocket Network on than it is to run on a cloud platform. And so the mostly every node in that network runs on bare metal. And so when you do that now you're starting at just the hardware cost and the electricity costs or the data center costs to rent, you know, space in a data center. And that's your cost to run a node rather. And so if you look at Infura recently, their free tier, infura's free tier, they actually send those requests to Pocket Network instead of their own infrastructure, because it's cheaper for Pocket Network to serve those requests than it is to serve those requests from their own infrastructure. And so that's really where we are starting to see proof that you could get significant cost savings and advantages by running this infrastructure on a primarily bare metal network, as opposed to building a similar product in a centralized fashion on top of a cloud platform, which is, you know, similar to, like, Vercel. Vercel is a beautiful product and platform, very popular today amongst developers, but they're built entirely on aws. And so what's the biggest complaint about Vercel pricing? Because they're starting from this unfair position of AWS's baseline costs. So that's really where we think we could see costs advantages is primarily by optimizing the network to incentivize people to run the infrastructure on bare metal. And that should give us a much lower cost basis for what it takes to run these nodes and provide the resources that these nodes are providing in terms of CPU bandwidth, etc. And package those and ultimately be able to do it at a much lower price point. **Speaker B:** Yeah, that is super cool. And I think that. I think it really speaks to, like, some of, like, the most powerful and like, the early visions of crypto where, you know, like, the. The story of the dao, right, starts with slock it and like, this idea that crypto could be used to, like, allow people access to, like, tools and whatever they needed without, you know, having to coordinate through, like, a company or whatever. And, man, just, like, in this moment of, like, reflection of over the last, like, hour of our conversation, I just think it's really cool to see a product that not only is, like, clearly doing very well, I mean, as a user and just like, seeing the way people talk about you and the way Twitter talks about you that, like, you're not only doing well, but it's in a way that, like, really carries the flag forward in terms of, like, you know, the things that I fell in love with this space and the things that, like, Jesus Christ. But, like, on a daily basis, it seems like some main character is trying to make me make it not worth it. And so all to say is that, like, what you guys are building is so cool. It's so cool on just, like, the size of opportunity, the business level, the Web2 level, the, like, crypto level. And just, man, congratulations and good luck and. Yeah, thank you. **Speaker C:** Yeah, thank you. You asked a lot of very good questions. **Speaker B:** Cool. So Before I let you go, can I ask you guys, like, where can the audience find you? Where can they find Fleek? And, yeah, I mean, if they're interested in, like, starting to build on Fleek, what's the best place to get started? **Speaker C:** Yeah, so for today, so the current platform is Fleek xyz. Most users are today still on the legacy platform, Fleek Co, but the plan is over the next few months, we'll kind of fully sunset and phase that one out and move people to the new platform. Um, but on, you know, on Twitter, we have the Fleek XYZ account, which is more for the the platform, and the Fleek Underscore Net account, which is a bit more focused on the network. And so I'd say that's probably the best place to stay up to date. But, yeah, it's great to hear that you like the current product and platform, but I promise in a few months it's going to look and feel and have a lot more useful features and functionalities. So, just to say, it's much appreciated and we love, you know, getting to talk to users, but we have some plans in store to hopefully, you know, make it a much better and just more useful and valuable platform in the coming months. So, yeah, awesome. **Speaker B:** Well, you've already sold me. But once again, like, I'll say it like, we gotta have you back and I'm so excited to hear what's going on in a few months. So, Harrison Parsa, thank you so much and have a good rest of your day. **Speaker C:** Thanks a lot. Of course. **Speaker D:** It was nice seeing you. **Speaker C:** Bye.