hey everyone this is Cena with another episode of into the bite code my guest today is Andrew hang the founder and CEO of conduit conduit is a rollup as a service infrastructure provider they're working with teams like proof of play zoraa mode Network and others so have a front row seats to how ethereum is scaling today so in this conversation we talk about how application developers can architect their smart contract so that they can scale on their own rollups we talked about rollup economics um how conduit actually works under the hood uh the conduit market place and a bunch of other topics I hope you enjoy I always return to this this example but um there's this game that you know the number one crypto game pirate Nation um and like you know they're we're lucky to work with them at conduit and they're doing anywhere from 20 to 50 million gas a second for you know 5,000 concurrent players you know uh I think six figure daus um and even that is like they're launching a second chain kind of I I guess at this point when this releases they'll already have done it but in a couple days um and you know I kind of think there's just this insatiable demand for compute um and a lot of developers today are kind of stuck in that scarcity mindset of like oh that'll never work on shade um and I think it might take a new kind of Builder it might take Rev iing that opinion but I think once you have 50 to 100 million gas a second on a single chain I think that enables unique opportunities I've I've heard you describe what you're working on as onchain compute right uh as opposed to Cloud compute or server compute and I'm curious like this question of building applications that uniquely leverage what's possible here and the scale that we increaseing have um do you when you think about these applications are you are you envisioning things that are entirely onchain and you know are Crypton native applications or do you see a way in which onchain compute starts kind of coexisting with Cloud computes within the same applications like what is how do you see this like future world and the types of applications you're imagining in your own head yeah I I mean think like web 2 and web 3 compute already coexist a lot like if you look at like the Unis swap front end right that's hosted probably on versel somewhere um and you know I think it's really hard to believe that um they won't coexist for a long time um I think it's a question of uh what part of the stacks do you bring on chain right and maybe this is the most critical part is like I think the new primitive of like crypto is like the idea of self- customing or the fact that you can have essentially a neutral platform kind of govern the relationship between you and kind of the objects on chain in a way that's transparent and like encoded in the VM um and so some arbitrary party can't just come and like flip the rules on right um and so I think when you're talking about bringing more things on chain I think bringing the pieces of your application that really matter like even in the pirate Nation case it's like U you know when your character is moving around on the screen that's not an onchain transaction but when your character like wins a battle or like you know opens a loot box or something that is onchain um and that's recorded and transparent and fair and they can't take that away from you um and so I think it's about finding the right piece of the stack to put on chain um in a way that delivers on the promise of crypto right which is like you do get this transparency you get these uh kind of rules that are enforced by a blockchain that no one can come in and like kind of take your stuff from you and I think what kind of onchain compute actually scales is your ability to bring more of that onchain right because I think if you look at a pirate Nation it's like a million gas per transaction for them um and if you're capped at two and a half million gas which is like the gas Target for a lot of these layer twos you're doing two and a half transactions per second that's like I don't know know like 10 players playing con currently or like maybe 100 um and so it's only by kind of scaling this gas limit that you're actually able to start to see people uh kind of bring these experiences on chain in a way that just wasn't possible before um and so yeah I think to answer your question I think it's about what pieces you want to bring on chain I think you know there are clearly things that webu are better at today like you know hosting front ends or things like that I think there are certainly projects you can use to try to decentralize some of that but I think ultimately it's it's kind of going to be like how like the rules of how these objects on chain can actually interact how they can change hands um and how they can grow and and kind of um uh kind of expand from there yeah yeah like we've had this order of magnitude cost reduction from Main net to l2s and that's enough to more casually kind of send transactions around and um you know we've all become a little bit more fluid with how we use blockchains but I think if you get another you know couple orders of magnitude decrease from there then it completely changes the sorts of things you can think about we we definitely think so I think like anytime you're talking about 10 100x 1,000x yeah and like this is the meane behind like gigas um and I don't know that we're at gigas yet I think we're Maybe at G gas and like sinking a chain um not online performance but I think once you start creeping towards there you know we're at 50 to 100 today how do we get even higher um I think it does actually enable new things to to happen on Chan in a way that is actually very competitive with the web two experience um and I think Beyond talking about the gas limit for a single chain um you know Web Two doesn't like there there are vertical scaling limits here um any web 2 application that gets big like faceb Facebook is running on billions of servers right um and so I think naturally I think there's kind of two Dimensions here there's one how do you get more parallel on a single uh chain so today as you know the evm is like single threaded um there are potentially opportunities for parallelism there it kind of ultimately depends on uh the Ste and and kind of how you're using it um but can we design applications to take advantage of kind of a parallel evm and that design looks very different than how applications are designed today uh and then two interesting like application developers need to kind of get into this parallel execution mindset because you can't be you can't you you need to kind of like Shard your States and be careful not to just like touch Everything at Once because that can't be parallelized exactly exactly so like the you know if you touch the same storage slots and there's dependencies on transactions that has to go serly um but if you're able to paralyze kind of the state access where they're touching independent sets of State those can obviously kind of go go in parallel and so for um I think apps have never had to think about this because like every you know evm is serial today um but if you want to go even Beyond like 100 million gas a second like I think a very obvious win is like instead of executing these serially like let's execute them in parallel um and so that requires rethinking from the ground up kind of how do I design this maybe I Need A Shard kind of this contract set out um and then importantly like how do these then interact with each other in a way that offers a unified kind of app experience um I think it's somewhat similar to um you know what we're seeing like proof of play do where they have one chain now they're launching a second chain there's going to be a third and fourth chain um and so not only do do I've distribute kind of the state across a single chain but like multiple chains and then creating a unified experience top of that um and so ultimately like again we're so early in the app design and like the techniques for how to actually scale an app with this new primitive that um you know I think what we're most excited by at conduit is working with like the highest quality folks are actually pushing the limits there and like working hand inand with them and I think um if you believe that apps are going to stay single chain that like parallel doesn't matter I almost think like it's a little bit bearish on crypto because you know it's so limited in what you can do um and you know the full case for crypto is we just bring tons more on chain um and you know these techniques are going to help us scale that yeah so think thinking more about uh scaling from an application developer perspective I really like the kind of uh different options that you sketched out so one is to have multiple l2s or multiple l3s and that's that's a pretty straightforward mental model right I just kind of take different parts of you know larger chunks of my state and almost think of them as separate you know separate databases basically and then my product is responsible for weaving them together into a coherent user experience um the other thing is that I have I have states that maybe actually needs to be able to uh you know an individual Atomic transaction needs to be able to act across different parts of the state and I'm kind of being intentional about um how you know within within the micro structure of how I'm writing data to that state to make sure that that's parallelizable and this you know this isn't possible on L1 and my understanding is because you kind of don't you don't know in advance when you're executing a transaction what what storage slots it's going to hit and and um you know we don't have this notion of access lists which is necessary for being able to do this but do you think on you know I guess one of the promises of L2 has been that we can kind of approach the design of the VM and the runtime from scratch and like introduce net new features like like being able to know in advance uh you know what what state a transaction is going to touch and like whether it can be parallelized is is this like the sort of a thing that's uniquely possible now that we're on L2 I I think it's uh you know there anything that's possible on L2 is also you know within reason possible in L1 it's it's mostly just a you know is L1 kind of aifi and how hard is it to get changes in there and I think maybe what we've seen is that it's kind of slow um and that's you know in some ways by Design because there's like you know uh it's say security uptime reliability really important for L1 and uh you know they're they're probably making the right trade-off there um I think traditionally um there's just been a lot of fear um on innovating on on this area and like in some ways um like major l2s right they're their business model is like constraining compute so that they can charge more for it um and and we kind of have the opposite kind of line of thinking which is like we actually just want to enable as much onchain computer as possible um as cheaply as possible and so it's just kind of a different thinking of how to use the technology um but I think uh yeah generally you're right which is the problem with um an arbitrary like storage is very slow on the evbm um and it's how you store data um and it's very hard to know what you're going to touch ahead of executing it um and you can even try to provide me an Access list but like depending on the ordering of your transactions that access list might be wrong right um You don't know who's going to stuff before you and how that's going to modify your execution so this is just generally a very difficult problem particularly um just for for a general like kind of settlement layer um I think like what I'm excited by are applications um and like in an application World in theory you can actually optimize some of the stuff because you know more about kind of the access pattern um and so I think that's an area that we love working with kind of our clients on um where we can kind of co-build kind of and kind of architect with you how to best uh take advantage of of this new primitive um and you know if you're at um Frontiers uh kind of a couple weeks ago which was you know great great event hosted by Paradigm um the uh rise chain I believe um gave a great talk on kind of parallel evm and um they they've done a bunch of benchmarks and um I think historically for like ethereum mean that I believe they were able to get like a 2X kind of parallel Factor um on execution which is like I mean it's it's good but it's not like you know traditionally you would like you would want to scale with the number of cords that your computer has right so like instead of a 2X you would want to see like a 10x or like a 48x or something like that and so like a 2X is really limiting and that just kind of tells you like this is the challenge of putting every single application on one chain um and having them composed with each other is actually kind of the worst case for how do I make this go faster um it's it's kind of just like an impossible problem and so I think for that reason for for a lot of other reasons I think applications will uh at least kind of the most ambitious ones will kind of want their own chain um and you know they'll be able to customize it they'll be able to design it and take advantage of them knowing what's going to execute um and I think that will importantly enable like much higher gas throughput um and kind of better experiences on CH and part of Designing it is deciding on its parallelization properties so I would say that today like parallelization is uh probably in the background where like you know Step One is let me just get an application on chain the evm is actually quite fast if you you know uh can cash things in memory um and or like make modifications to to know what you'll be able to access ahead of time um I think like parallel comes once you cap out like a single thread and so you can do about 100 to 200 M gas a second uh depending on like the workload and like your infrastructure setup and optimizations and so once you need to go beyond that I think parallel is actually kind of the next Frontier um and you know we'd like to work with the best team trying to achieve that um and it really does come down to like um yeah do you have the Ambitions and like how badly do you want to make it happen because it's very much kind of on the bleeding edge um and will involve a lot of work um and like that's where again we're excited about customers like a proof of play we're like this is like the thing that they do they they need a scale pirate Nation the game works people find it fun and like scale scale scale and so working with them is incredible because like there's just such a tight Loop there and we know that the work we're investing is like gonna actually make it into production um and so I think that's kind of the thing we're most excited about again it's just finding these teams kind of on the frontier that want to push the limits and have a have a great kind of plan for how to you know attract the demand for that much compute as well yeah if we kind of zoom out of parallelization for a second and think about the question of scaling more broadly and I have this sequencer and I want it to max out the throughput to to the extent that's possible how would you sketch out the the range of approaches angles on that problem that that you and other folks are like exploring right now what's going to like what's promising what's going to bear fruit how how is this actually going to work yeah um you know I think like uh a lot of folks are doing great work in the ecosystem so like Aragon was probably uh so originally called turbo gu right was probably one of the uh original people that took like a performance-based approach towards running a node and I think they have a lot of great things there and then I would say almost the spiritual successor and like successor is probably the wrong word but like um W is really interesting from performance um kind of uh uh kind of perspective and uh originally for L1 and now looking towards L2 and um I think really I wouldn't say there's one thing I would say there's like there's the phys of the problem which is you have this big Ste tree um that goes only up over time and gets more expensive to access as like more stuff happens and so it's I wouldn't say it's any one cut I would say it's like probably like a thousand incremental things that you need to work on and in particular I think one thing that we're excited about here at conduit is just the um kind of the the fine-tuning for a particular application because again like when you're on your own chain you can optimize for the specific kind of access patterns um and so Something That We're you know we've worked closer with proof play on is like how do we tune this exact thing to your use case um whereas like again on a generic settlement layer like you kind of don't know what's going to happen you can optimize it today for one thing but like tomorrow it's going to be something different um so what does it mean to optimize it for the specific use case like specifically yeah I mean this goes into a lot of um engineering like algorithm stuff I think like a big performance thing that comes up is like the database and so like as the state size grows eventually there will become a point where you have to go to dis right um and like you know you can't it in memory anymore you can't store it in memory anymore it's like too expensive like if you look at proof of play their archive State size is like 10 terabytes um and so like I don't know if you can find a 10 terabyte memory instance on AWS um and even then like it's growing at a rate where you know it's it's going to outpace whatever Hardware um improvements we get there and um so I think like a lot of what we're doing is like how do you optimize the database um so that you know as the state grows um you can kind of maintain the same amount of throughput and it's a mix of algorithm stuff but also like really lowlevel database tuning um and maybe even one day like you know today get these using level uh DB or like Pebble DB level DB as far as I'm Weare is no longer maintained Pebble DB is like it's fine um but there are like next Generation databases that are being created today like um if you've heard of firewood from Avalanche Labs um or uh even like reth and Aragon are using this thing called M3 or M mdbx or something like that um which are like next Generation kind of databases um and were these developed for blockchain use cases or outside of this just in like General software database world so firewood was um and then M mdbx I'm not sure if it was purpose-built I do know that that's where they've seen the most product Market fit so de facto it's kind of like a blockchain um database um but yeah I think like you run into very obvious like I think ethereum by seeing at you know one to two million gas second has never had to run into these issues um but I think almost like immediately when you hit 50 million gas second you start getting this these really bad kind of performance characteristics and so I think for us it's been like you know having really good Web Two it's almost like more of a web two engineering challenge in combination with some of the web 3 stuff um and uh going super deep on like optimizing benchmarking getting the right tunings um Etc to enable the kind of throughput that uh you know uh you know we're doing today on production yeah what I I am really curious about just the engineering of this like how what does it look like for you to be tuning this database like what does what does your kind of inside conduit infra setup look like to kind of Mock and and kind of like test this this setup yeah so I think this is again we're just having the the kind of most ambitious and best builders on conduit is really helpful is like they give you the data set that and they have the requirements that kind of push you forward and so a lot of this tuning was just done on proof of play um and like syncing the chain um and trying to keep up with the head of the chain and so when you have like a natural data set like that um it makes it really easy to to kind of um bench mark at least for that specific use case I think generally for us again because the evm is a generic computer thing and like you can even think about like in web to right databases are also kind of a generic compute thing of like you can store kind of anything you want and there different it's like uh for each of these specific use cases you're going to have to find a different tuning or a different um kind of approach um and so we have a great one for the proof play the next one might look different and then like um eventually we want like a systemized process of like these are kind of the yeah forun like these are kind of the major um kind of verticals of like access patterns as we would call them and like when we make database changes like let's run that Benchmark against all of them and just see where the performance Stacks up that's cool so is the is the right way thinking about this that there's like execution which is the parallelization stuff and then there's the database which is the the tuning these different kind of new databases what else goes into scaling nl2 yeah well without giving too much Secrets sauce um there are also like algorithmic changes that you can make to the evm um and I think a lot of it has to do with just understanding like again I'm there's no secret sauce here I would say like the biggest thing is just to do the benchmarks to do the profiling and then to optimize from there and like there's a lot of uh and again it's kind of specific to the use case so what we do on like one customer might not be like what we need to do for another one um but generally I think uh we found the deeper that you can kind of go on the evm and the internals and how it works in finding the bottom the easier it is to design kind of algorithms and or even like making its way to the application Level of smart contracts of how to best like utilize this um and so today given the nency of the bleeding edge where like the computer's getting better and like you know maybe you know you might need to change higher application is designed to take it advantage of it I think you know there's no Silver Bullet um and that's where I think just like working with us and like working in a tight Loop is kind of the way to achieve that and like maybe I'll give one example here of like uh you know storage is a really big one where like certain contracts um on certain chains might have like you know you can technically have like infinite storage in a contract um and that actually tends to be really bad performance-wise because you just have this like gigantic amount of state and then um it's only growing over time and then even for like an archive State like you know you need to store that multiple times um and so one thing you know kind of maybe a more obvious change is like you probably want to Shard out your contract State not just for parallelism but just for being able to take advantage of uh kind of these different um kind of caching mechanisms and so if you have one contract that has like all the state in the world uh that's going to be kind of hard hard to scale so like one area that we're kind of going back to teams on and like is important from a parallel perspective and a like eventually like this will you know uh kind of be a challenge to scale so so a team would ideally start thinking about scaling and talking with you guys and like thinking about the rollup architecture when they're actually architecting their application like this is it's it's part of the same process ultimately the earlier the better um because once you have your smarton contract and application built it's like it's going to be extra work to change it um and I think like where we've had the most success is where we get like kind of embedded with you in the process early and we can help guide how to build that um kind of onchan um frankly those are some of the customers that we're most excited by is like they have these ambitious goals and like can definitely kind of deliver on on the GTM and now we just got to figure out how to make it work right there's probably some version of you know start with doing unscalable things like build build the MVP see see if you can get to product Market fit and then for v2 like design with scaling in mind I think there's a bit of that and then I think like the again the customers that we're most excited by here in some ways already have the pmf and they're hitting like the scaling challenges yeah of like you know we either like we have this in a web2 way right where it's offchain today and then like we periodically settle onchain but there's a very obvious like we want to take this activity and put it on chain um or it's something like um uh yeah or it's something like you know we we're rate limiting the number of users that can actually use it because we know we're running into these gas lines and so help us you know actually bring more users on by scaling um versus uh yeah kind of like limiting what we can actually build here into the bite code is sponsored by splits are you TI tired of sacrificing security for usability splitz believes is still way too hard for teams to self- custody their onchain assets they're building a new kind of Internet native Bank on top of ethereum splitz makes it easy for teams to manage the whole life cycle of their finances from structuring revenue sharing agreements using payment flows like splits and waterfalls to managing those earnings once you receive them using pass keis and smart accounts splits is being used by teams like protocol GI Zora song camp and others I'm a big believer in them and recommend checking them out you can learn more at splits. org yeah okay maybe Switching gears to interoperability and um I'm curious if you can just explain to me how you think interoperability is going to work so today we're kind of In This Moment In Time where there are many many layer TW each user has a different wallet different set of assets on each of them we kind of use either the canonical Bridges to move between them or things like a cross where you know a a kind of solver of sorts goes and gives the asset to us more immediately on the other side for a small fee um but I feel like this is all kind of temporary right and a lot of the parts of this stack are kind of moving and shifting like even optimism you know superch chain kind of announced their or shared their interoperability spec a couple weeks ago so as someone who's kind of working on this and across these different rollups um what is this going to look like do you think yeah um you know I I understand the desire for interoperability um I still think like it's uh I think and maybe this is a spicy take I think like most of what people want when they say interoperability is like I want be able to easily I'm on like this app on this chain I want to easily use this app on another chain and I don't really want to have to think about it um and I think if we can solve that that's like 99 like 99% of like what needs to be done um and then I think like you know it's kind of about like what tradeoff you want to make to achieve more than that right and so like in some ways third party bridging yes there's a ton of fragmentation there's a lot different like Bridges um and they have different security properties I think that maybe mostly gets figured out in the market over the next few years um and I almost wonder like in the same way that like Global composability is something that people talk about I almost wonder if like almost in some way like Global composability means that you are kind of the same app if that makes sense um or like maybe should be the same app uh and I think like if you look at web 2 like there's not a ton of like composability between like major verticals because they're quite differentiated right you might share some data here or there um but it's not like it's the dominant access pattern um and I think when you're talking about moving like assets and stuff like that they're obviously like defi it can be great to be composable um but if I look at kind of the uh uh kind of the landscape today it's almost like you have ethereum where obviously a lot of liquidity lives and then you have arbitrum one which kind of is the dominant settlement layer um for for defi on later toos um and then interoperability you know it's going to be good even in the super chain case like we're very excited for it I think it's going to be low cost and it's going to enable easier asset bridging I don't know that it's going to give you the same level of composability that you know being actually on the same layer entails um and so I think like I think over time like these layers differentiate to the point where like is a user on pirate Nation going to want to automically you know open a loot box with like depositing into like a Lyra Vault or something right um I just don't know if that is actually going to be the dominant use case but that also seems to be what a lot of crypto Twitter or like a lot of protocol designers seem to be optimizing for um and I almost wonder if it's kind of a premature optimization for a problem that doesn't exist quite yet and like I think the dominant again the dominant um if feels to me access pattern here is going to be I'm on one chain I want to easily or like I'm on an app on one chain I want to easily use an app on some other chain and not really have to think about it and I think like tooling here starting to get um actually much much better how how how does that particular case work so you know I'm just going to call it like uh you know like a privy for example I think recently integrated like relay um and relay is a you know pretty partner that we work closer with they do a lot of cool stuff but the idea is like if you can kind of in the background easily um you know log in somewhere and then just like use another app and like not again not think about it maybe it's a oneclick transfer or something like that I think that starts to reduce the friction to a point where you know today I guess like without that it's like oh I'm on this chain I need to know about the chain I need to find the bridge to the other chain I need to bridge across I need to wait like five minutes hope it goes through and then maybe I can use this other app whereas I think this new product experience is like oh I can just one click start using this other app somewhere if it's on a different chain and I don't have to think about it and so ultimately it needs to be abstracted away um and you know there will be maybe a cost associated with that I think given the amount of competition and just how much is happening I I think those costs kind of really go down over time um and I think that again it's maybe similar to like I just need to log into a new app right or like anytime I use a new thing thing I need to add my payment credentials there's a similar amount of friction there um but just being able to oneclick again Bridge into another app um on another CH I think is um getting pretty close to like the ideal user experience okay another question I have is around economics so do you think of rollups as primarily like a scaling technology or do you do you also think there's something important there about how it changes the economics of running an application and like kind of opens up a different business model to to developers how do you think about that prompt 100% I feel like you you've been uh you've been listening to my content um but uh no it's it's a it's a I think it's a great Point um and uh yeah I mean I think like again I don't know that apps like proof of play would be possible on somebody else's chain they've actually gone through a very interesting Evolution where they start out on polygon again it's fine for like low transaction volumes we got expensive and there's congestion they moved to arbitrum Nova where you know they became 40% or more of the TPS and started hitting the point where you know they're they're the ones increasing the congestion on the chain and so they're bidding against themselves it's only getting more expensive and you know in one day they spent something like 70k um and on their own chain it's like well actually I own the chain and so like the fees can go back to myself and really all the onchain fees I'm paying are the settlement costs right which can be quite cheap particularly for you know U like an Alda chain um and so I think that uniquely enables the business model to the point where you can treat a rollup as your own server versus um using someone else's chain is like a very very expensive AWS or something right um where you know the they they have a different business model because they're you know it's this open public righton database and uh you know uh they're showing that data forever and like they decentralization reasons everybody needs to be able to run a node so if we don't make it expensive enough then we're going to get spammed Etc whereas I think on your own chain you know maybe there's still the spam problem I just think there's probably a lack of people caring and you can you can even raise fees to the point where um you know you kind of come bat spam but like the important thing is like the fees roll back to you right so if proof of play is uh kind of uh kind of driving a lot of the transactions for the game with Rel layers then like the fees just go back to their Rel layers which allows them to actually have this like circular system where the onchain fees are are basically Nega and so I think that relay what are the Rel layers here so the Rel layers would be like if I'm you know if I log into the game I kind of sign something that enables so like when I open a loop box or something that triggers a transaction on chain that you know comes back and like opens my loot box for or something um and so uh it's kind of one of the ways you can get a gasless experience on chain for their users um this is like account abstraction type stuff exactly um so I think that ultimately like bringing that what was what would have been like especially as they scaled right just kind of this variable cost that would have bankrupted them has now become like a conduit kind of hosted fee for the infrastructure and like a very minimal kind of onchain kind of cost footprint um and so you know I think actually proof of play in the business model is uniquely enable kind of by their own roll up and I think a lot of people are starting to see that where you know particularly if you want it's not just like more gas but it's like if I were to pay for that much gas on someone else's chain I would go bankrupt like very very quickly um and so in some ways it's becoming the only kind of model for application specific rups that want to push the lims yeah that makes sense so they're so are they running is Conduit running the relay or they're kind of incentivizing an open kind of network of relays and they're just paying them out of the transaction fees that their sequencer gets so proof plate runs the Rel layers today um and in theory that can be like decentralized right all you're doing for relayer is actually just uh you're taking kind of like the signed um kind of permissions from the client and then Translating that or like sending that on chain so it's actually like a pretty minimal kind of process um over time for things like you know pppp and like even their cross chain stuff right will eventually kind of be more more decentralized um I think again where we're starting today is like a little more pragmatic and making sure that the game works and like the user experience is good and then eventually getting to a point where all this is kind of like fully decentralized um and there's some exciting things there around interoperability particularly on like I guess on the uh so they're they're on the arbitrum Nitro stack um and so there's some interesting things around like Fast withdrawals and being able to um you know send messages kind of across chains in a way that um you know is faster um that can help enable some of the a little more seamlessly um but you know got to got to walk before we can run and ensure that you know the the customers have a good experience and then like building into that later so if I think about the economics of a rollup in some ways you can think of the kind of transaction fees and maybe meev as the revenue and then the costs of data availability settlement as as the as the expenses right and the difference between the two is what you make as a sequencer is that how you would think about it and do you think given you know is this a real lasting thing or does competition kind of Drive these revenues down to the bare minimum that they can be yeah I I mean I think that uh the the revenue models are going to be different depending on the change and I would say like looking at even the proof of play use case right if you look at their settlement fees basically negligible and then you could imagine like sequencer revenue is probably negligible too right and so the revenue model there isn't the sequencer fees it's actually going to be maybe more of the uh for lack of a better term like the web 2 kind of SAS side of things right is like we have some unique technology that enables them to go fast and to maintain that throughput we obviously have a bunch of other infrastructure surrounding that that like kind of power the game and I almost think that like um when we look at like fees I almost think that like that's not the greatest long-term model um because like if you look at ethereum during the bll it's like extracting these gigantic fees but that's just clearly not sustainable right uh and I think the endgame for crypto is like we have sustainable um kind of almost like quotequote boring um business business on chain that um you know just kind of do their thing and generate revenues whereas I think traditionally it's always been like oh like this you know what is the next thing I can dump my money into to like maybe get a 10 or 100 x and that making me very price and sensitive towards like what the fees are right and I think realistically like the speculative process is an important part of developing the infrastructure getting interest getting capital and Builders into the ecosystem it's not going to be a long-term uh it's it can't be the dominant thing let me let me put it that way forever um and so in that world I think like what we're doing is in some ways disrupting kind of the current metag game of how you think about things where even for an arbit one or like an optimism or a base like the dominant model is I want to extract revenue from our users for paying for transactions right um and I think if you want to build cool things on chain and like reach the next level I actually think you have to invert that and it's like you want to make that basically free you want to minimize kind of your onchain footprint cost um while enabling the type of scalability and throughput to actually build unique experiences that can onboard people that you know aren't going to go into defi or aren't going to go into these other things and in some ways um in some ways like not saying the blockchain shouldn't exist or it shouldn't be obvious but like kind of Fading Into the background where you know you have these like strong guarantees but it feels kind of just like a standard web2 application yeah totally it's like a it's a yeah it's like the your margin is my opportunity type situation where even if one kind of Layer Two doesn't reduce it fees its fees the other l2s can and they can attract use cases that are made possible uh at more scale at lower fees and there's a kind of like Battle of network effects and kind of winning different segments of applications and users which probably I mean Trends this Trends the fees to lower and lower amounts all right I got a couple more fun qu personal questions to close if you're good with that so first one is you've been working in the valley and in tech for some time I'm curious what what kind of differences have you noticed as you've kind of started working in crypto like from a from a cultural perspective like the good and the bad like what are what are the kind of things that have stood out to you yeah crypto is a very interesting industry I would say that um I really like working in an industry that is on the frontier and the bleeding edge and changes very rapidly and it's Dynamic um and you kind of don't really get that as much in web2 unless like I think the recent AI stuff might be a little bit different but like for a long time right it was just like like how do I build a B2B SAS kind of company and is there like a tiny sliver of a vertical I can like kind of cut off here and ultimately that wasn't like super interesting and so I think in contrast with crypto I think in some way like good and bad thing about the industry it's like very hard to know where it's going to go right like um and I think even when I started the company it's like a month or two later FTX happened right and like I'm trying to H our first engineer or something and then FTX happens and he's like oh this whole industry is a scam this and I'm like oh well um and so I think uh I think there's the dynamic and kind of the freedom of it that's like really interesting um and like one of the reasons that I like continuing to build here is like you're on the frontier and you don't know what's going to happen and you have an opportunity to help sheep that um I would say some of the things that were surprising to me um and I think are really just a I would say are more of a like we're kind of still in nent industry is that um you know I think there are a lot of like Mission driven or like a lot of uh idealogues for lack of a better word um and I think that can be useful for things but we also have to Circle back and identify like are those same ideologies still serving us um and like are there ways that we can improve as an industry um to you know increase adoption to uh you know ultimately get this Freedom technology into more hands um and to be kind of more useful um and you know I think like that certainly when we're starting that was a challenging aspect uh uh but I think like the you know the industry has gotten more mature I think like our product and Tech is getting out there I think a common refrain we used to get a lot was like you know isn't a role of just centralized and like you know I think there have been a lot of um what's it called uh uh improvements and kind of the decentralization aspect where you know optimism has fault proofs and it can be fishless arbitrum obviously has kind of always had those and is launching permissionless fall proofs and so there's a little bit of like you know this is centralized today and like it's never going to work and like well we're you know the tech is it's complicated and like we're going to get there and um yeah ultimately it's going to be how we're going to scale ethereum so there's always a little bit of that um and then I think something that is very different from traditional Tech is I think the how fast people can quote unquote get rich I mean this happens in web to as well and like you might have seen that in the recent AI CES but I think the the level to which like economics influence things um and there's kind of this a lot of noise because of that um is uncommon I would say um and is kind of uh a unique characteristic of uh crypto where I ultimately think like one thing that I kind of strongly believe is that over time like I think the people who are able to navigate that and understand that yes incentives are one thing but probably not the main thing that's going to make me successful and then to choose you know the kind of pragmatic and and things that are going to give them the best chances success um and some ways like delaying that gratification for kind of the bigger uh kind of pie later on um I think that's those are the customers that we're excited to work with and ultimately in our experience have really built the best products and experiences um the ones who kind of focused on on building and that is again a super unique thing where you know I think sometimes we have the TX better product we're kind of better and then like there'll be a big token Grant somewhere else uh and then they'll kind of like go that direction and so there you know there's not always anything you can do in those scenarios um but it is something that uh at least in in kind of my um my experience is is quite unique to crypto um and I think as a as a Founder you just have to be aware of that and um yeah kind of not just understand like it's kind of part of the game a last kind of fun question is we have a mutual friend in Georgios and I think you guys have worked together pretty closely on things over time and I'm curious what what you I mean how you would describe Georgio to someone who doesn't know him maybe and then what have you learned from him as an engineering leader as as anyone as like a person also yeah um georgeos is incredible he's uh I would say he is probably the sharp one of the sharpest kind of technical people in crypto that deeply understands the tech at a at a granularity that um actually means like you know he knows what he's talking about um and that is actually like you'd be surprised actually a lot of people kind of don't know what they're talking about um and so he's consistently somebody that you know I can call on uh if I need to you know talk through something kind of on the engineering side of things um I think like what I've been most impressed with is his like growth uh so I think you when we F maybe I've told this story before but like when we first met he had you know just come back from burning man he actually had broken his ankle doing a wheelie and had to get like trucked out of there uh and um you know I was starting my eir actually at Paradigm and so I would actually go visit him in his apartment where he would be stuck in the Attic because he could barely walk around I would like you know go like whatever fetch water for for him cuz like he couldn't walk and then like we would like start start jamming um and think now like almost two years later I think like he's uh I would say like much more measured and I think kind of uh yeah kind of like in control for lack of a better word um and he's always been a very passionate guy and I think that's a lot of what's Driven him and and kind of uh how he's gotten to where he is today I think what I I've been most impressed by with him is like how he's been able to kind of transform into you know somebody much more measured and and kind of able to navigate things kind more more skillfully so um really been a pleasure working with him and uh yeah he's been an important part I think like what he's doing with WTH is pretty incredible I think there's probably one person in crypto who could have pulled that off and you know we're very lucky that he he decided to do it um and uh yeah I think generally just across the ecosystem he's made like incredible contributions um and I think you know where're we're very lucky to to kind of haveh yeah I like I like that story I didn't know that that he broke his ankle while doing a wheelie it it kind of explains a lot I guess is a very georgeous uh very georgeous way to uh be introduced to him and uh yeah I think uh I don't know if you like he he's also very into anime and so during that time you know obviously stuck on on the couch kind of introduced me to anime and uh I think we watched all three Neon Genesis Evangelion movies I think back to back until like 3:00 a.m. or something and this was like probably the 10th time he was watching it all the way through but um what's the what's the pitch on that why why should I watch that what's the story NG honestly I I was pretty I was pretty tired uh so I don't know if I was like processing everything but uh I think it was uh I think the the animation was great the the kind of plot lines were very complex um and uh I think George Jo has gone really deep on like the analysis of it and like kind of the you know the the the kind of themes and everything I think um I was I was just trying to trying to trying to keep up uh and so yeah I he's probably the better person at about that yeah all right man I think we can call it there cool man thanks for having me yeah really enjoy the conversation hey I have a small ask here if you've been listening to these conversations and want to support what we're doing here I would really appreciate if you leave a rating and a review for the podcast it might seem like a small thing but it actually makes a big difference to help other people discover the show also thank you and I'll see you again soon Back To Top