REBROADCAST: Decentralized Tech Livestream - Fluence: Compute Without the Cloud

Evgeny Ponomarev: PA

Jon Radoff: Welcome everybody. Your back on another episode of the the decentralized tech live stream. Today we're gonna be talking about decentralized physical infrastructure networks. It's what I'm building at Beemable. We're infrastructure for game servers and game tech, all the stuff you need to scale up and power games. And I brought into the conversation Yagini from Fluence, which is building a compute network and deep in as well. So we're gonna have a lot of common ground to talk about today. We're gonna be able to talk about the economics of this, why it's different than other like cloud providers in the past. So if you've been scratching your head wondering, like what is all this deep in stuff about, or you're just curious what Beemable is doing, or what the future of really figuring out how to scale up a new kind of network that's community-owned instead of just owned by these big siloed companies like Amazon. That's exactly the conversation today. So I think you're gonna learn a lot about the future as well as the present. Let's start with Yagini and just his background. Thanks for joining the program, by the way. Why don't you introduce yourself to the audience, just in what you've done, and then we'll dive into Fluence and then we'll take it from there.

Guest: Yeah, thanks for having me, John. Excited to be here. And hello everyone, I'm Yagini, co-founder of Fluence. And we build the decentralized computer platform, which is, as John said, a deep in network. And I'm coming from, I basically, Commitur Science background, I graduated from university in Russia, the learned math, and was born in academic center of Russian, I have a lot of friends, doing science and physics in math and chemistry. I've been doing software engineering, beginning of my career, then switched to more like a product management, launching project like IT companies, like a small IT company, a SAT project. There I'm there, did quite a lot of internet marketing and got quite early in crypto. So I first time heard about Bitcoin in 2011, was playing with mining, there was GPU mining at the moment, mining tools, so kind of interesting things and been following this space. But I'm actually started doing something full time in around crypto and what changed around 2017. And that was actually the time when we launched Fluence and we launched it, you know, we sort of the idea, the vision, but then we actually had to pivot a few times to figure out the best, you know, what would work. And, but the story was that, we've been looking at blockchain space and we're excited about how it can change the finance world and how it can change the internet infrastructure world. And, you know, having this IT background, I was like, okay, maybe we can have a new internet in front of, maybe we can have new clouds, like decentralized cloud computing, decentralized cloud storage, DNS, like a lot of actually aspects. There was potential to be decentralized and, you know, we got some progress with many of these aspects but not everything was actually decentralized after all these years. But we can talk about it. Yeah, so essentially, you know, I was working on these decentralized compute solutions or like ways to implement this to solve this problem for quite long time with students.

Evgeny Ponomarev: So yeah, I even don't even remember much about,

Guest: and you know, my past before joining the battery.

Jon Radoff: We actually got involved in Bitcoin at about the same time. So in 2011, I mined Bitcoin, but I did it on, actually, strangely enough for this conversation, a massively decentralized basis because I had thousands of visitors to my blog and you could write a little, at the time, this would no longer work by the way, but at the time you could use a little JavaScript program which if you aggregated it across thousands and thousands of blog viewers, you could actually mine a little bit of Bitcoin at the time.

Guest: Yeah, I remember that actually, I experimented with it a little bit, but it wasn't very productive.

Jon Radoff: Yeah, well, if you had a lot of visitors, you could do it. You could get enough. I mean, it was definitely good economics

Evgeny Ponomarev: if I think back to, you know, you could mine,

Jon Radoff: I probably could mine a Bitcoin after several months doing that. And that's worth a decent amount today, so I can't say it was a terrible idea.

Evgeny Ponomarev: So let's start with the high level.

Jon Radoff: So we have a lot of people that come to these streams who are game developers because that's my own following. A lot of people in different dimensions of Web3,

Evgeny Ponomarev: but it keeps feeling like decentralized,

Jon Radoff: physical infrastructure networks, deep in. It's kind of a, it's emerging as this hot category, but it's still very new to people and I think people are trying to figure it out. So I have my own thoughts on this, but since this is what you've been doing for years, why don't you start by explaining, what is a deep in, why is this revolutionary relative to the ways you could have accessed compute in the past,

Evgeny Ponomarev: and then we'll kind of take it from there.

Guest: Yeah, I usually say that deep in overall is the new generation of sharing economy projects or like a crowd sourcing resources. And we had this big wave of, it was more than 10 years ago, right? When Uber and Airbnb were created and sort of people started discovering, oh, okay, like we can, you know, get people having some resource to the marketplace and aggregate this, you know, the resources services provided by individuals into some platform where other people would rent from them and get the product from them. But this was like a web to centralized marketplaces, you know, fully controlled by company and so on and so forth.

Evgeny Ponomarev: But then in parallel, we saw how blockchains enable

Guest: these kind of models, but without intermediaries. You can, you have this neutral database in the middle of participants of service. And you can distribute payments easily, you know, without any fees of yacht on ramps or ramps and stuff. And then you also have better sort of democratized management of the system because it's blockchain, it's proof of proof of stake, so it's kind of like, can be managed by community. So in the sense like what Deakin is now is, you know, mostly infrastructure networks that powered by blockchain to distribute payments rewards and connect participants. And, you know, they are, you know, usually not centralized in terms of governments and authority and control, like a more democratized and centralized company. But they overall enable individuals and small companies to enter new markets and, you know, enter new sort of revenue streams that were not possible for them before, but provide some resource to the, to this platform, to this Deakin platforms. Oh, you need a ton.

Jon Radoff: So we owe a little bit of a debt to the past of pre-block chain ways that this could be done. So we should maybe acknowledge some of these things just to give people a sense of how they might be structured. But so I ran SETI at home, for example, for years, for example, of like, there's just a lot of data mining that you effectively have to do across a very, you know, large data set. So a way to do that is to parcel it out a little bit at a time and let lots of individual computers do it. So if anybody has run SETI at home, you have done kind of, not exactly, because there's this centralized component to administer SETI at home. And we can talk in a little bit about the distinction there. But if you've ever used SETI at home, you've kind of interacted with a version of a decentralized compute network, even though there's nothing to do with blockchain there. Another example would be like folding at home, right?

Evgeny Ponomarev: So folding at home is an example.

Jon Radoff: And anyone can go sign up for this. It's another one you can try where there's all these important, like, protein folding experiments underway where you can offer your computer's capacity to be able to help solve these problems. And the applications there in drug discovery and biology and things like that. And a famous example of it was, early in the COVID pandemic, there was actually a need to kind of really quickly get to a simulation of what was happening with the virus and do it down to a protein modeling level. So they actually got to the point where they could model a little bit of a time slice of the way the proteins associated with the COVID virus was working. And from that simulation, they were able to do a lot of really interesting vaccine work and drug discovery work related to that. So that's another example. In fact, I think up to that point in time, the virtual computer that was running the COVID simulations was like the biggest, one of the biggest super computers at that time. So those are a couple of cases. There have been other cases like where there's decentralized networks of various kinds, like Tor, for example, was kind of a decentralized network for privacy.

Guest: It just even be torrent, right? Even be torrent. It was early example that everyone knew, which was kind of delivering new ability to access to a lot of files and been music and videos, which mostly was not sort of legal to consume. But the technology itself is like peer-to-peer networks are very similar to what Deepin is now. It was just no incentive, no monetary incentive. I would say no direct incentive that you could get from an network by participating. But the narrative and overall, it's very similar thing.

Jon Radoff: Yeah. And Oscar, you're getting a call out from the community you probably already saw. They want you to talk more. Oscars are a producer. He keeps things on track. He keeps the camera moving. He brings in other content. From my personal experience, it's virtually impossible to do both of those jobs and be a great conversationalist. But of course, Oscar, we love it anytime you jump into the conversation and say something. So it was juggling 30 plates, but I'm more than happy to interject when it's needed. But that's a good cue for me to mention something at this point in time, which is, we also depend a lot on you, the community, asking your questions. So this will be a conversation between me and you, and Gennie, but we love it when you ask your questions and want to weigh in with a comment or have a question for me or you have Gennie. So you've got someone here who's been working on decentralized compute tech for a long time. And if you're interested in that or the blockchain intersection or deep in, this is a great time to ask your questions about all of the above. And I see we've already got 250 or so people in live audience. So there's plenty of people here. This is probably bigger than most of the talks that you could attend at one of these big blockchain conferences. Like you could go to deep in day or token 2049 or consensus and have a deep in talk. For the most part, you're not going to get an audience if you have a couple hundred people. So here you are, you could have direct access. So please ask your questions. And if you're watching this in replay, by the way, hey, think about coming to one of these live and participating in the conversation. We could even put you on camera, by the way. We're kind of crazy on this live stream. So if you want to talk to us and jump on the stage, Oscar will send you a link to StreamYard and we'll get you in. So please do ask those questions. So here's probably the big quote. This is one of the questions I get asked probably more than almost any question, including from blockchain investors you have getting. So I'm curious how you answer it, because you probably get the same question too, which is, okay, well, we had it in the past.

Evgeny Ponomarev: We had said it at home.

Jon Radoff: I get it. There's a distinction between a total centralized controller versus truly decentralizing it. But why do we need that? Why do you need a token? Why is tokenizing this economy matter ultimately? What are your thoughts on that? Why does Deepin need a token at all?

Guest: Yeah, so I think there's two different questions. Like why you need blockchain overall, right? And then why you need to, like with blockchain, essentially, as we said before, if we have huge amount of participants globally on the network that earns some money and want to transact through, like, by providing services to each other or to one single big aggregation, you want easy transactions to have been to send small pieces of revenue to any person in the world. And what changes does it really well? And then, like, democratization of overall governance also solved. Then the second thing is why do you need token, right? And different networks answer it differently. Overall, token is something that allows you to scale faster. It's essentially a tool. Your network can theoretically exist without token. Because this service is not a token. The service is video servers or anything, right? It's a software product. But the token allows to create additional incentives for both supply and demand sites and essentially accelerate, like if we consider dipping as a two-side marketplace, then we can accelerate the growth or like the flywheel by attracting, like, rewarding supply side, attracting more supply side, then rewarding demand side, and attracting more demand side. And that's essentially what all these early sharing economy projects like Uber did. They subsidized riders and they subsidized drivers. But they did it by using essentially their venture funds money capital that they raised. And they just essentially paid cash to audience, to bootstrap in new cities. And then they started reducing this subsidy. So if we can do similar things, so if we need to bootstrap certain sub-market, or like bootstrap the whole market overall, we can start from subsidies, but then we need to reduce them over time. But at the same time, because token is really connected to value, overall created by the protocol, the network, what you get is that early participants who contribute early, they subsidized those tokens, they actually benefit from the project growth, because the token usually, it's usually correlated. When your project is more successful, token is growing, right in price. So essentially, if you contribute early, you get some tokens, you get benefits later when we have like a more growth and overall project growth. So this is one and more aspect of having a token. The second aspect is governance essentially, right? Okay, we want this network to be neutral, to be infrastructure, to be not controlled by single entity. How do we do it? It's not really possible in the world, in non-blocking world. That's why essentially we need some tool to represent voting power. And we need to grant this voting power to people. And then in crypto, there's been a lot of experimentation, with DAO, with different voting schemes, but overall, on average, voting power usually represented by amount of tokens, either halted by person, or being delegated to this person. Right, so as a trust. So essentially, token governance component is very important part for token existence. So I think sort of these two aspects, really make the deep network bootstrapping and growth and management much more always token.

Jon Radoff: So I agree with everything you just said about the ability to, essentially, the first part, what you were saying, is the token is a more capital efficient way to bootstrap and then scale out and network due to that flywheel effect, if you can incentivize supply and demand, and kind of as the network scales, everybody's happy with the incentive mechanism, which you can decrease incentives over time, as the network grows, and it becomes more potent economically. The second part was about governance. To me, there's another thing that goes kind of hand in hand with the governance piece, which is just this idea of using the token to align the incentives of all of the network's participants. So in a traditional business model, and I'm curious what you think of this as well. A traditional business model is you have a corporation which generates profit for its shareholders, and then they create products, and then those products have to be increasingly profitable over time, either through growth and or better value extraction over time, and the customer kind of votes with their wall at every time they keep buying the product. The problem with it is there are multiple layers of disconnect between the customer and what they really want, and the profit generating entity, and then all those layers of management, which sit in between, which kind of have the principal agent problem. With a token, you have network suppliers, you have the demand side,

Evgeny Ponomarev: you have anyone who's interested in the future road map

Jon Radoff: of it voting in governance, which was your example. To me, that's actually in principle, a financial model that gets everybody working together a lot more effectively,

Evgeny Ponomarev: because the demand side can't call the shots

Jon Radoff: or the suppliers go away, right? And vice versa, the suppliers don't get to just dictate everything that happens. They have to be part of a cohesive road map. They have to be able to offer their equipment or hardware in the case of a decentralized physical infrastructure network in a way that's economically viable, hopefully through even some competition mechanisms around pricing. So it's that alignment of interests that I actually think is really interesting about a tokenized economy.

Evgeny Ponomarev: Where are your thoughts there,

Jon Radoff: anything to add or subtract from that?

Guest: Yeah, I think that's the right way to look at it. And you could argue that, people also can buy Apple stock as they buy iPhones. But in crypto, what's cool is, this nature of sort of ability to get those tokens for contribution or just get them easily on the market, any fraction of the token, like for $1 or for like 10 cents for anything, right? It essentially means if you have a big project with millions of users, it's much more probable that this, like a higher percentage of these users would own some of tokens versus these users would own your stock if you had stock instead of token. So the point is that you have this, the whole ecosystem, you have users, providers, you know, some core team members, investors, and if you have a token versus stock, versus like shares, the higher percentage of this ecosystem would get token, would have token versus haven stock, just because it's much easier to get token. And because you can get token by contributing, it could be like token cashbacks, could be like a lot of mechanics, just the customer, you just pay money, you still, you know, most probably when you interact with crypto project, you still most probably would get involved in owning the token, which makes you much more aligned with the whole kind of direction of the project versus when you're just a customer of some web too company.

Jon Radoff: So what's been the hardest part if we kind of bring it back to the fluency of what you're building, maybe let's introduce to the audience here, specifically, fluency, like, what's the problem you're solving? And what's been challenging about building the network that you're building?

Guest: Yeah. So what do we do is a decentralized computer platform, essentially, we are aggregating computer resources globally and packaging them into the cloud services, which we offer to mainly be to be customers. Similarly to big clouds like Amazon AWS, Google Cloud or Azure.

Evgeny Ponomarev: And I guess in the specifics of this product

Guest: is that like, we want to really build the alternative infrastructure to be cloud platforms, big cloud companies. And we understand that actually, you know, huge part of the internet of all digital services we use every day relies on these big platforms and corporations. And if we really want to build alternative, this alternative has to be very, very reliable. We have to feed, you know, SLA service level agreements like uptime and so on and so forth. So this actually creates quite a lot of challenges in terms of how we technically provide services and how we approach the supply side, like the computer providers on our network. And for example, that's why at the current stage of the network, we like, we only allow professional providers who own enterprise-grade servers in a top tier of facilities like in data centers, like tier four data centers. Because we, you know, haven't found the way, like we would love to involve more people to be providers to contribute their compute capacity in whatever form they have to network. But the problem is uptime and latency, right? If we have this capacity and we bring it to customer and customer, you know, to lose this access, then how do we solve it? And then it's actually not trivial, I think because every customer needs very different services and tolerable to very different downtime tops. But overall, like, you know, in clouds, in cloud world people invented sort of reliable, on demand and reserved compute. And then the invented spot instances, which people understand that it's something available to you temporarily, which may go down, like maybe just results in other customer at any moment of time. So there are certain types of workloads that you would run on it. But you know that this is not very reliable in terms of, in terms of uptime. So right now we focus only on, on demand and reserved instances, but then also we will bring spot and spot in when built in spot, we will have much more ability to bring compute, not only from top tier facilities. Essentially, we would tolerate downtime, because customer would be ready for it, and we would have more ability to migrate workloads quickly to other instances.

Jon Radoff: So there's a question from the audience here,

Evgeny Ponomarev: which is basically nobody really knows about Deepin yet,

Jon Radoff: which by the way, that's probably not a bad thing for those of you who are watching today. Like you're the ones who are picking up all the alpha here on a market,

Evgeny Ponomarev: which is, you know, today still quite small, really,

Jon Radoff: and the revenue generated by Deepin projects, mostly is still quite small compared to like big commercial enterprises. The question really though is, why don't people know about it yet? I'll give one answer and would love to hear your own view, you know, getting.

Evgeny Ponomarev: I mean, one thought is, although the term is new,

Jon Radoff: some aspects of this have always been present in blockchain. So if we look at Bitcoin, for example, I actually had someone say at a conference to be something interesting, like Bitcoin was actually the original Deepin project, because what it is is its bunch of servers, who mine out Bitcoins. And of course today, Bitcoin is tremendously valuable as an asset, and even a medium of exchange sometimes. So Bitcoin had computers that were decentralized, that joined the network adhered to the protocol, mined out the Bitcoins by doing cryptographic algorithm, basically. Now, what Deepin is really about is, again, having hardware on the network that does something.

Evgeny Ponomarev: But the something now is something that other customers need, not just the ability to generate a token.

Jon Radoff: The token is the reward, but it's not the work being done by the nodes. It's not like you have to do a bunch of crypto algorithms on the node. You're doing, for example, compute or running a game server like we do,

Evgeny Ponomarev: or helium, which is Wi-Fi,

Jon Radoff: or Geo-mapping, or all these weather, like all these really interesting applications, where there's some hardware device that joins the network, does work, where the payoff is no longer cryptographic, inherently in nature, but it is doing work that pays a token. So you're essentially mining with Deepin nodes, is another way to think of it. How do you explain this, and why I think that people who go to big conferences like consensus and stuff, have seen Deepin, but the broader blockchain community, I noticed when talking to people from Web 3 gaming, for example, they're very unfamiliar with it. And if I mentioned the idea of nodes,

Evgeny Ponomarev: they're used to these nodes that were sold by various game projects,

Jon Radoff: where they're not really nodes in the way you and I probably think of it. We think of nodes as a hard piece of equipment doing work, but these other nodes, we're really just, you know,

Evgeny Ponomarev: mostly just NFTs that unlock the currency drip for these death spiral economies for the most part.

Jon Radoff: And that's not what we're talking about when we do nodes in Deepin. Deepin nodes actually do work, serving something that a customer needs done,

Evgeny Ponomarev: like a game server, for example.

Guest: How can we go awareness to that? I think this node sales little bit created this, a little bit of chaos in people's minds, because, you know, projects selling these NFTs as a license to run a node, and then people think that they buy a node, they actually buy a license to run a node, and then participate in the network. But I actually agree that that blockchain's been probably for us, Deepin, Deepin, proper Deepin projects, because you run a node, you earn some crypto, you provide a value to a network, and you earn some crypto. And it both the case for work and for stake, and actually today, I think people run hundreds of thousands, if not millions of nodes for blockchains, like, and that's how you can count it as Deepin. But then if you don't count, just blockchains as Deepin. Then I would say that Deepin is still quite small. You know, we definitely see the progress. We, you know, it's also about us collectively pushing the narrative a little bit and explaining people what it is. It kind of helps. We see it's growing, and, you know, we're doing all these conferences and podcasts around Deepin. But I think, it's just,

Evgeny Ponomarev: it was required some level of maturity

Guest: from blockchain world to enable networks like we have built it. Because we actually rely much on, on what blockchain technology was created, like, rollups, for example. Right now, like, Fluence runs on the rollup, which is layer 2 to Ethereum, which is the rollup, like, this is the blockchain, separate blockchain network that allows very, very cheap transactions, and secured by Ethereum, by, by parent chain. And, we use a bunch of like, a standard things that been invented in the crypto world, like, bridges, you know, from parent chain to our chain, like, multi-seqs, and so on and so forth. And it just took some years to have it built, so we can just go and use it and build our network, using all these tools.

Evgeny Ponomarev: And, you know, we needed it because,

Guest: we are not defy applications. So, like, our transactions are very, very small in fraction. Like, we want people to really be able to pay, like, just, you know, a couple of dollars and get compute. And then we also want provider to consistently earning revenue, or by small transactions from people. And when you sit on a public chain, you compete for the block space with other projects that sit on them.

Evgeny Ponomarev: And then it's, it's typical kind of crypto thing.

Guest: Whenever, like, any hype project appears, like NFT drops, or like meme coins, anything like this. You know, blockchains used to, blocks used to get full, these transactions, transactions fees increase. Now, if you're just using some service in one of the $10, $5 around, now your transaction fee can be super high. And, you know, you're not petty as a customer. So, we wanted to, like, avoid it completely. That's why we went to a roll-up. And roll-up is basically the chain that only asks who will sit on this chain. No one else there. But in order to have it, to use it reliably, we had to wait a little bit until this technology's, wouldn't make sure. Yeah. You're really pointing to,

Evgeny Ponomarev: it's really only been the last two to three years,

Jon Radoff: really, that any blockchain has existed at scale that can do high three-put,

Evgeny Ponomarev: throughput fast-finding

Jon Radoff: with low-gas fees, right? So, layer two's, you know, is one approach to that, as well as these other newer chains that kind of have it built in. But up until then, like, if it was built on Ethereum, and you really need a smart contract, type chain, to build these kind of technologies. Like, if you built on it, built a deep end on Ethereum,

Evgeny Ponomarev: three or four years ago, you'd be paying,

Jon Radoff: you'd have to pay, like, I don't know, $50 gas fee every time you got a little bit of credit for offering your computer. So, it wasn't really feasible at that time. So, I mean, it explains a lot about the explosion of deep end over the last few years. And then, companies like Helium,

Evgeny Ponomarev: for example,

Jon Radoff: are several actually that have started to move to Solana, which has some of those properties built in, like, beamables is shipping our deep end on Solana. For similar reasons, as you're using a roll-up layer two on Ethereum, you know, relatively low cost for becoming part of any transaction on the network that gets you there.

Guest: Yeah, actually, the funny thing that the deep end ideas overall are not new, right? They've been here for many years, and people really tried to build stuff, and they failed. And they failed mainly. Like, you basically earlier, you had the choice, either you go to some of the existing chains like Ethereum, and then your users pay, you know, many dollars transaction fees, and then, obviously, you cannot really, like, enable a lot of use cases with it. Or you build your own blockchain, which is very, very hard. Like, you have to maintain the network of YDitors, you have to make sure you're, you know, there's no attacks on your consensus layer. So, you start, and then you need to onboard users to your chain, which would be separated and isolated from other chains. So, it was really, really challenging. And you ended up sort of working on the blockchain part of your product, instead of working on your core product. And that's why we saw a lot of projects failed earlier. Yeah, but also, like, now, you have this scalable chains like Salana, you have multiple layers of roll-ups for Ethereum. So, now, it's actually much, much easier. And there is also, like, a bunch of abstractions built in UX. You can just, as a developer, you just can go to this roll-up of the service and spin on the new roll-up in a few minutes. So, that's actually a really great progress with developer tooling of blockchain, enabled it as well.

Jon Radoff: Crypto from X's chiming and why not base? What base is perfectly viable, too? Like, base is very little transaction cost. Gas fees for the most part. I mean, they're not a layer, too, per se, they're an EDM built on, basically, a clone of Ethereum other than a few changes that they made. But, like, it's similar in the sense that you have these EDMs that are designed from the ground up for lower gas fees and faster finality. And they all ultimately sort of interconnect back to Ethereum at the end of some long sequence of processes. Like Ethereum kind of becomes the ultimate settlement layer, clearinghouse for where a lot of this, these financial networks end up getting connected in. But yeah, I think base is another great example. We know the base team and have talked to them a lot about it. We have games being built on base.

Evgeny Ponomarev: So, they're great.

Jon Radoff: And you could point to a long list of them. Arbitram is, you know, that is another example where a bunch of deep-in projects are rolling out. What do you see, if any, as some of the blockchains where most deep-in activity is happening?

Evgeny Ponomarev: I think mainly what I saw was

Guest: Solana, Arbitram and Ethereum essentially. Somehow, bunch of projects, a pure-ton Arbitram, or used Arbitram technology to build their own relapse. And then other group of projects are just a pure-ton on Solana. And these are two big groups I'm observing. I don't know why. I think mainly, I don't know why Arbitram, like not optimism, for example, it's very hard to answer. It just happened somehow. And on Solana, I guess, it's the same reason. People are just looking for the chain with cheap fees, transaction fees, without the need of operating, and you know, your old notes of the chain,

Evgeny Ponomarev: or something like this.

Jon Radoff: I also want to give a preview for a little bit later in the program, about 10 or 15 minutes from now. I know some of you are waiting for the secret word we're giving out because we're doing our own gamification over at Beemable, where you can earn a point reward towards the air drop. We're going to do post-TGE. And if you listen closely, you're going to hear that secret word. What I'd love to see from everybody is mention, mention some questions here in the chat, or put some comments and ask them questions, I should say. So we love the engagement. We got crypto crypto. You've got a good name because obviously you're crypto native. You're pulling us on every chain. So we is another good one. I think that probably, and I love Soe. I think they are really committed to gaming, as an application class, because we talk to a lot of really great game developers who are going to Soe. They are going to Soe for a similar reason, which is Soe has this hyperperformance, fast finality, low cost, very high throughput, which is very good for the micro transactions that games need. But you are making a good observation, which is for a lot of deep-in projects, the same thing that would make a blockchain performance for micro transactions in a game. A lot of things that happen in a deep-in network are also essentially micro transactions. So, you know, Soe, I think in theory has the performance for it. I think what it doesn't have currently that you see in things like Salana or sort of the EVM ecosystem generally is not a lot of traction around other projects that have solved problems, and that you can sort of riff off of, because I think ecosystem and code ends up being an important reason in this particular space. So there's just more to leverage into your project if you go off some of those existing ones. But in principle, Soe could do it quite well, I think, on purely a performance basis. I don't know that it's necessarily one of the target markets that they're particularly focused on at the moment. I don't know if you have any thoughts to add on Soe in particular, particularly at GENI.

Guest: Yeah, I don't, I guess, from our perspective, you know, we look at blockchains from different perspectives, right? Actually, I would show fluents for a minute. So essentially, yeah, I said we built a decentralized cloud, but I didn't kind of explain what it is and how is it kind of plays role in the ecosystem overall. Because one of the main use cases is running blockchain nodes. Like we essentially provide servers, like, you know, physical servers, and virtualized servers, where people can deploy any workloads, anything that they run in Web2Cloud, like databases, backends, game servers, like anything, right? But in Web2, main use case, computational use cases. People run in nodes, people run in Ethereum nodes, two in nodes, arbitrum, Solana, anything. And we provide this in-frag globally. Because right now it's kind of a little bit silly. We all build in this Web3. It's all decentralized. It's all, you know, great and alternative financial system. But it heavily relies on centralized cloud. So when Amazon wants to turn off half of blockchain,

Evgeny Ponomarev: they can do it like in a minute.

Guest: So that's why having this in-frag, level-level in-frag, and so that's why, you know, for us, all of the chains is great. We hope you to support them. Like we can run any blockchain nodes to secure any blockchain, like for the customers who want to, who need these nodes. So, yeah, we also have this AI use case, which is coming. So currently we only focus on just the CPU servers, but GPUs are coming because it's actually, there's a lot of quite growing intersection between Web3 and AI, and a lot of Web3 projects demand GPUs, because their nodes also require GPU access. And we're also bringing it to people. But overall, the point I'm trying to make is that the chain, kind of the restaurant features of different chains, where you can build these of that type of services easier or harder.

Evgeny Ponomarev: But there is also, there is also no management part of any chain.

Guest: Like if you need to run a node for a certain chain, you can, you know, you need a compute infrastructure, which we are happy to provide to the whole ecosystem.

Jon Radoff: You're kind of connecting into an adjacent topic, which is just the composability of Web3. So that composability, meaning a lot of these protocols can interact with each other and work together. In the same way other protocols do, like you were just mentioning how Amazon could shut things down. But we can drop to lower level protocol levels, like there's no CEO of TCP IP, for example, like no one can shut down TCP IP. But that's how almost all computers on the internet interact today. There's not really any CEO of domain name services, right? Like that's actually a relatively decentralized, or a CEO of email, right? So, you know, there are some other protocols where computers on a peer to peer basis can connect with each other or form networks that allow certain services to be delivered. And that's kind of inherent in the idea of decentralization. Now, when it comes to computer provisioning, it has this new problem in it, which is not many outside of these cases, like a city at home or a folding at home, not many people are willing to just donate computers, especially computers that just end up being used as a way to keep the company, like a game or operating blockchains or operating any number of commercial applications, right? So, what you have to do there, unless you're going to build huge financial infrastructure in a centralized way, which would be the alternative solution. But if you want to do it, like TCP IP or email, or domain name services, then you have to build in the value exchange into the protocol itself. That's really what blockchain does. It allows you to, with smart contracts, it allows you to facilitate value exchange. So in the protocol level, you can actually get paid in a token or, I mean, some D pin also do stable coin pricing. We're gonna offer stable coin options as well and our network for people that just want sort of less volatility around token when they provision services. Blockchain lets you do that. It lets you bake into the protocol of financial exchange when that needs to happen, which certainly for access to capital equipment to run your software, you really, really need that. So that leads though to the second piece, which is now that you've got things that can do the value exchange and do things like computer provisioning, it opens up this whole universe of, now you can start writing software that does something like provisions a computer to help backstop whatever's needed. So instead of having to go away to a SaaS company for a service that's more canned or go to Amazon and just order your own servers and write all the middleware and connect it all together, well now you can start writing software that provisions computers on demand and brings them into your use case. And the composability of Web3 is what lets you do that because we're doing with the protocol level. So an example of that, for example, is it beamable as we're building our D pin, we're actually going to other compute networks that have access to good data center level compute that is appropriate for running AAA games and we don't have to just rely only on sort of the nodes that have directly access beamable. We're happy to source computer capacity off of anywhere, including other D pin projects that have access to compute capacity. And you can only do that when you have the ability to easily compose all these other networks in and when you can do it with smart contracts, it makes it much easier to facilitate the value exchange up and down that value chain effectively. Thoughts on composability or anything I was just saying

Evgeny Ponomarev: you have been getting.

Guest: Yeah, I think this, I like this, how we moved towards the world where I actually can permissionlessly on demand in the moment, rent huge amount of computers like in the seconds from all this D pin infrastructure. If I need this, if I have this spike in usage, if I have like a millions of users coming to me and I need to serve them, I can like quickly rent it from D pin network, from us as well, right, from, but there's a bunch of D pin networks like what we do. And it's because it's all implemented through smart contracts, it's totally doable. I guess the only current challenge is that where we need to move forward a little bit is that essentially like everyone is building on either a different draw up or different chain. And okay, like if you want to at the same time rent from fluency and another infrastructure project, you would need to interact with two different chains, which maybe, you know, it's just implementation of it first time and then you can see that you have slightly a challenge in, but you know,

Jon Radoff: I could use case for all these multi chain projects that are coming along now does kind of simplify integration between chains.

Guest: Yeah, so I hope that, you know, we will see more and more sort of universal abstractions on top of different networks and different networks within the single category, like networks that provide the same similar products. When you have like an abstraction over it, which is also permission less, now if I need to, if I have a demand for computer, just go to the abstraction and then it's been rented from multiple networks. So it's actually, if you compare to how I do it today, it's a very different experience because today I have to preset up the count in the cloud platforms and I need to make sure my quotas are there. So, and they usually don't give me good quotas from, you know, when I just sign up. So I have to talk to their support and to get ability to rent quickly big amount of resources. And there could be like a bunch of details, like I gonna overpay for longer period than I actually used. And crypto networks, like deep in networks are much more flexible. Like we currently, for example, we're gonna charge per day, but we're gonna, you know, switch to charging per hours and then ideally per seconds. We want to charge per second because what's the, like it's more convenient to everyone. Like if you, it allows to utilize compute much more efficiently and it's beneficial for everyone.

Jon Radoff: So really interesting conversation, question from Rosinate. I hope I said your name correctly there. But your question is, essentially comes down to something related to what you were just saying, which is affordability of deep in for countries like India where there's lots of customers, but they can't pay high fees because they can't afford it basically. Like let's talk a little bit about general affordability, at least of the use cases we're most familiar with, compute, you know, essentially it has its compute. How does deep in change that in terms of affordability and just global accessibility, you think?

Guest: Well, it's just simple economics. Like you, like we bring more supply and we, this way we make the compute cheaper, essentially. Like the more compute on the market, the cheaper it is. And deep in projects allow, like incentivize people to contribute compute, which means bigger overall compute capacity, which means lower price per run. Because essentially we saw how this happened in recently in the proof of work, crypto mining, right? Essentially it's a specific sort of use case where it's specific compute all these ASICs and specific devices, but still the people essentially competing towards zero margin, the cost of operating of these devices is equal to cost of electricity. And then they have some economics about like what, which Bitcoin they sell most of it immediately to cover expenses because they're quite high, but they also, you know, sell, they keep some of Bitcoins to sell later when it goes up to sort of earn actually earn money. So with, with deep in networks, it may be similar if it's like really reached the level of saturation when you really get a lot of compute connected and available all the time network. And then some of the providers would constantly

Evgeny Ponomarev: be ready to lower the price

Guest: and to compete by price with other providers to get more workload. Maybe like some of these providers would get cheaper electricity, maybe some of these providers would get better deals with data centers, maybe some of them will be data centers. So their economics is better. So you essentially pushing people by creating this big market and connecting supply and demand, you're pushing these participants to be as efficient as possible. So, and I think it's very important in context of growing AI and growing compute demand. And we see it right now like countries building data centers right around the world to onboard this need of compute capacity for AI tools. So that's why I think like this deep in compute network is very important. Because that's actually makes the economics of compute much better.

Jon Radoff: All right, I'm gonna give my own thoughts on this as well, which builds on what you're saying. And then I'm gonna say our secret words. So listen carefully if you've been tuning in to listen to the secret word and earn some points on the B-Mobile Network leaderboard today. But there's some other ways to think about the way Deepin does things, which is we're taking compute and essentially creating a liquid asset class around it. So if compute truly becomes a liquid asset class, not unlike energy itself, then you create a lot more competition in the market

Evgeny Ponomarev: and that drives down the costs for a lot of the people

Jon Radoff: who are customers of these kind of things. And what that also means is if you talk about like a developing economy for example, like it allows them to access compute, which is powered by the cheapest energy availability out there, which also lowers the cost. Also you do have the whole profit margin aspect of the silo providers like an Amazon, for example, where there's just a lot of quarterly results. They gotta keep posting up every quarter and a lot of management layers and a lot of overhead to operate those businesses, which you've actually reduced out a lot of that in Deepin network because it's really just direct operators joining the network and typically within Deepin, the network taxes and things like that that might go back to a foundation to keep kind of managing and investing in R&D for the network is much less than like the profit extraction from a profit generating company. So that also contributes to it. I have seen some economics by the way where pretty frequently with access to compute, the bandwidth component can often be like 90% less. The compute itself is 50 to 70% less. You've probably seen similar numbers. I mean relative to like cloud providers, so Azure and GCP and AWS and whatnot. So that ideally is opening up a lot more democratized access to these compute servers. But I love the thing you commented about

Evgeny Ponomarev: like the similarity or proof of work mining operations

Jon Radoff: because also a lot of the Bitcoin mining operations or token mining operations, they're now getting into Deepin because it makes a lot of sense for them. They've really built their businesses around essentially what's energy arbitrage. So they get energy for one price. They do some computation using that energy to generate a token and pretty much the MBAs that run these things are just running a spreadsheet every day. Like what's the mining that we should actually be doing today based on current energy cost for us versus the economic output. And for a lot of them, they will be getting into the Deepin business. So you can kind of think of Deepin is powered by energy and a little bit of equipment. But most of the cost frankly ends up being energy because energy is the real cost of operating computers over the long term, not just the capital equipment up front. So it will continue to have this energy arbitrage where through the forces of competition you converge on something that is really good for customers overall because the profitability to operate them ends up being much lower than these big siloed providers where it costs billions. So I said I was gonna give the secret word.

Evgeny Ponomarev: So I'm gonna give it out the word

Jon Radoff: and we'll put it on the ticker below as well so that you can all see it. The word is infinite. I-N-F-I-N-I-T-E. I think I just spelled it right. I hope I spelled it right. I-E, yeah, you got it. Okay, so that's the word. You have to get any way. We're kind of coming up towards the end of our hour and I really wanna thank you for participating in this. I think we were able to cover a lot of subject matter

Evgeny Ponomarev: for people to learn about D-Pen

Jon Radoff: as well as people who have been following D-Pen probably get some additional ways to think about it. While we kinda wrap up in the last two or three minutes any closing thought, any big picture ideas that you wanna share back to the audience here?

Guest: Yeah, actually my marketing team is asking me to also say secret word.

Evgeny Ponomarev: I'm so excited.

Jon Radoff: It's both to secret words everybody.

Guest: Go for it. So our secret word is building the metaverse. So building the metaverse all together without spaces. And you participate in our pointless program just enter this as a code and you get, you're gonna get boost there. There we go. Oh yeah, it's written in.

Jon Radoff: Maybe afterwards. Yeah, well, hey, D-Pen is probably gonna be the thing that actually builds the metaverse here at a more fundamental low level. So I think D-Pen is actually the bit, this is my closing thought that I'll share with everybody. So D-Pen to me is the use case that allows blockchain to go from essentially being a set of financial products on chain which has been great. It's been very transformative. But I think most subgenres or subgenres applications in blockchain today have been retargeting a very crypto native audience. And that's everything from DeFi, decentralized finance. And there was a much earlier point made about games. Like crypto actually is like, why don't we get great games on it? Well, for one thing, it takes a long time to make a good game. Like five plus years to make triple a games like the ones you mentioned. But yeah, like making it more of a financial transaction to get into these games and use your wallet to get in when all these big mistakes that have been made in crypto gaming so far, that has just been more of a speed bump or an impediment to accessing it. And it's targeted largely a crypto native consumer. Now a lot of games now realize that they have to target the general gaming market, not just create crypto games. So now the crypto piece is almost being buried in most Web 3 games and you discover it later. And I think some of those will expand out.

Evgeny Ponomarev: But the reason I think deep in is such a compelling category

Jon Radoff: is that the ultimate customer of any of these services is not really a crypto native person in any way. If you need compute, if you need Wi-Fi,

Evgeny Ponomarev: if you need the weather,

Jon Radoff: like these are things that everybody needs.

Evgeny Ponomarev: It is not a customer that has to care

Jon Radoff: about blockchain or crypto. It just so happens that tokenizing the node operations, the network, all of those aspects we talked about earlier is a great way to bootstrap and build these networks for the customer who is not crypto native. And this is pretty revolutionary within blockchain. It's a category that truly targets a customer who doesn't need to care about blockchain to benefit from it. And I think that's a pretty exciting change for the deep end market. You have Gennie, thank you so much for participating in the conversation today. This has been a lot of fun. We covered a lot of ground. And thanks to all of you who tuned in today, we've got over 600 people in the live audience learning about Deepin, I hope, for the first time. And go check out Fluence, learn a little about what they're doing, go check out Beamable Network. We're building a deep in for game servers. And we'd love to see all of you again at all of these live streams. We cover this stuff, AI, web3 gaming, general game development, decentralized tech. I do this four days a week, and we love it when all of you are here. And if you're watching it and replay, join us live sometimes.

Evgeny Ponomarev: Thanks, Gennie.

Jon Radoff: Thank you, John. That's been wonderful discussion. All right, until next time, everybody.