Building Tomorrow's Backbone: The Rise of Decentralized Infrastructure | KBW2023

so ladies and Gentlemen please be gr Applause hey can you hear me good afternoon everyone uh thank you for joining for this panel my name uh and a quick introduction on my side my name is Augie elog I'm a partner and head of Asia at CMT digital we are a 26-year-old trading and investment firm based out of the us we've been investing broadly in the space for about six years we've raised three funds and we partnered with over a 100 companies in the space I am a as I mentioned a partner in head of Asia which means I spend most of my time in this part of the world and uh it is my pleasure to be moderating this panel on building tomorrow's backbone the rise of a decentralized infrastructure along with our esteemed panelists Eric Wang from analog and Hillmar ort from gelato before we jump into it guys maybe you want to give give a brief introduction on both yourselves and a couple minutes on your project for those in the audience who are perhaps less familiar for sure hey hey everyone thanks for having me here um so gelato we call ourselves web free decentralized backend and what we do is we provide web free services to over 400 projects uh in the space to help them augment some of the limitations of today's uh evm uh based blockchains um these include we provide smart contract Automation Services um we help projects abstract away the complexity of dealing uh and sending transactions to blockchains to their customers using account abstraction services and then also we help them leverage offchain computation and data to augment the capabilities of what their smart contries can do so you can think of us as kind of like a a decentralized network of nodes that help you supercharge and create super powers to your smart contracts and yeah our customers include projects from defi nfts gaming and and many others oh great hi everyone U my name is Eric and uh I live in one of the greatest cities in the world which is soul and uh I'm working on a project called analy an OG so at Analog we build cross chain messaging services so essentially we what we do is we connect different smart contracts and different applications on different chains together um and we've been building for about a year and a half now raised about 14 mil um and since then we've basically been heads down trying to build out a few products the first product that we're launching is called analog watch it's essentially like the graph across chain so it allows you to query data from any chain that you want uh into your own uh DB front end or into your own set of smart contracts so it basically allows you to do cross in query uh as well as Crush in WR um the other product that we're launching is called uh what we call the Omni chain SDK so essentially it takes separate smart contracts and connect them together um and that way you could essentially build uh take uh your smart contracts and build Omni chain apps that connect your liquidity or user base or whatnot from different chains uh and unite them together that's cool thank you both of those are very cool uh so let's get into it guys uh you know I think the agenda for today infrastructure is obviously a very meaty topic uh I think what we'll hope to do today is two things one we'll double click on some of the more um topical areas within decentralized infrastructure today namely sequencers app chains layer tws and we'll also maybe ask a few Project Specific questions based on some of the challenges and focus areas that you guys are dealing with uh but before that let's start with the basics right uh and maybe understand better the problem statement why is decentralization at the infrastructure layer important Eric maybe you want to kick us off yeah I think a lot of it has to do with exactly what the product that you're building is promising so when people generally build blockchain based products um their user base tends to uh desire some level of decentralization when they use it because that's one of the biggest value add to those products so when de developers are launching those products then uh people desire for example the validator set to be decentralized well they desire for example the node set of nodes to be decentralized uh they desire for example when you get data Dash should be decentralized right and um all those things are just it's kind of like a tacit agreement in a sense between um what you sell your product as and what the user wants so if you're building like a layer one or Layer Two generally speaking um you have to have many more decentralized components however uh it doesn't necessarily mean that all blockchain based products necessarily need to be decentralized so for example a lot of people use the Wormhole Bridge the Wormhole Bridge is a bridge with only 19 validators uh if I remember correctly and uh the people who use that uh they read about Wormhole and they're generally well familiar that these set of validators are pre-chosen from a selected group and uh if they're okay with that then I think uh they're fine to proceed however uh I think for most users they desire something a little bit more um uh that follows the decentralization ethos that can't be censored that can't uh really be taken down when uh certain people collude and that's why like building decentralized infrastructure is very important yeah fully agree um I think for me what maybe let's just start toine uh INF web3 infrastructure for starters so there are like different layers here right so you have the block Builder sort of execution layer at the very bottom which is Hey how can we how do we uh order and sequence transactions on chain where are the transactions stored how do we get access to the data of these transactions to Applications right H there you have like validators you got RPC providers for example right then you have another layer on top which helps read or index that data to make it easier to sort of consume right this is where like indexers come in then you have um services like oracles or Automation Services like gelato which help you actually send transactions from offchain to onchain with data that can be originated from onchain or offchain um and then you have for example Rel layers and Rel layers are basically like services that actually help you write to blockchains and send and and trans transport data from other services to chains and then you have like stuff like cross messaging protocols which are kind of combinations a bit I see them as combinations of oracles and um and Rel layers um and so where decentralization matters varies a lot of what sort of services you um Talk about and um for me the cool thing about web 3 is building hyper structur so you can build protocols and applications that are not controlled by anyone that just fulfill a certain process like a swap a landing protocol or or what not but minimize the value extraction that they actually uh uh take as a as a fee and uh if they need to run they need to run like autonomously not controlled by anyone they need to have censorship resistance um they don't they shouldn't have single point of failure right um and they um should be reliable in terms of there shouldn't be one party that goes down and then it stops working right and so this is why I think decentralized in this these are the reasons why we have decentralized infrastructure that enable things like hyper structures and where de centralization matters varies a lot based on which part of the stack you actually talk about maybe just one quick followup there for you both um I mean is there are there any parts of this stack that you finished describing him that you feel need to be centralized either temporarily or indefinitely I I you know I don't know uh the devil Devil's Advocate point of view could be that centralization even on a temporary basis is a good stepping stone to decentralization and maybe decentralization needs to wait for uh incentives to be put in place or other such things not sure if either of you have a view on this yeah I can maybe start so centralization is great right so centralization like we are not in here to decentralize decentralization is not a goal is it's not a means to an end right we are here to make something censorship resistant we are here to make something highly reliable and available but we are not here to just make it decentralized for the purpose of decentralization right it's not it's not an end goal so sometimes centralization is great and sometimes things should be centralized so I'm a big fan of if you need performance and scalability as the most important attribute of a certain service then you should just be highly centralized for example my humble opinion is that indexers should be super centralized like there's I don't see a big reason of having like not like not available on chain right there is a lot of benefit of actually having decentralized data but like if you just read it offchain then you can just choose from mul providers if one starts lying to you just switch easily over to another one right so I think there this is not really a real benefit you gain benefit of decentralization if you have like censorship resistance and reliability uh and these kind of things um and another thing that I think is very hard that should be decentralized it's very hard to decentralize is everything that has subjectivity involved and non-deterministic behavior so data from apis for example right if if you want the coin gecko price on chain then guess what you have to trust Co gecko because this is It's The Entity that provides you with the data right or some other API provider so there because it's nondeterministic it's also very difficult to decentralize and there we have to just live with some tros yeah um I agree with a lot of what you said there um so before I did this job at Analog um I worked as a head of growth at parody and at parody we uh basically helped a lot of blockchains or layer ones on board uh into the polka dot ecosystem and U what I've discovered from my experience there is that a lot of a lot of chains struggle really hard to kind of get to that point of decentralization and basically after Bitcoin and ethereum uh and not just within the polka dot ecosystem basically any layer one or Layer Two to some extent started off with some large degree of centralization so for example when polka dot first launched they launched with essentially POA for a few months uh before they were able to upgrade the chain and change into some sort of uh POS consensus mechanism and um I I think uh to a large part uh as blockchains and these uh ecosystem and infrastructure becomes more complex it gets harder and harder to decentralize on day one one and that's why many of these infrastructure builders make an initial decision to become more centralized at the beginning however uh there's a few caveat that they need to follow so basically they essentially need to have like a really good road map to uh that point of decentralization and also the points like the the kind of value as that they bring initially has to be like strong enough so that uh even the components that are centralized are uh kind of forgivable in a sense by the audience so uh for example like for layer twos layer twos have centralized sequencers right now and uh for centralized sequencers uh they are not able to fully censor a transaction because when people actually uh don't get their transaction included by the sequencer they're able to post it to layer one still for the most part so uh however uh in the long run uh they like these users are going to be very dependent on one entity to kind of do their uh bidding essentially or do what they want to do so um in that perspective then uh as these uh ecosystems mature more and more they're going to have to kind of uh decrease that level of centralization as long as the users and they have some sort of tacit agreement on what that process looks like makes sense and I definitely want to loop back to sequencers perhaps a little bit later in the conversation uh but Switching gears maybe I can ask a couple of uh specific questions Project Specific questions to each of you uh Hilmar maybe we'll start with with yourself and gelato um you know I think as you described right gelato is very much a a well-known and well-loved Tool amongst the developer and DOW Community uh but I would say the the fundamental premise of you know leveraging a network of nodes to automate smart contract execution Etc uh has a much wider applicability with a much larger user base versus today I think is a fair statement and so I guess the question for you is what types of users do you see using your backend today how do you see that customer pers developing over the next few years and what needs to be built from an infrastructure perspective in order to service that new segment yeah so we see the evm as kind of like a new serverless platform that is uh decentralized and has certain capabilities mostly consensus and trust that uh traditional serverless environments don't offer and what gelato doeses um sort of very technically is we help augment the limit ations of these evm blockchains um for projects that still want to retain this properties of censorship resistance and reliability but um want to build applications on let say ethereum or something right and that are powerful and actually have a lot of uh the capabilities that normal smart contracts don't have and our most of our user base is uh still I would say defi um for example like the Maka dos of the world the pancake swaps of the world right the BV finances of the world everywhere you have recurring conditional event driven processes happen that need to be executed with certain data um there is where gelat basically comes in so all the non-user specific transactions that have to be sent all the operational things in the background all the plumbing right this is basically done here by gelato in this case for example uh but not only that like we we expanded from automation also into gases transactions and helping projects improve the ux of their um of their applications for example Zora the nft marketplace uses us for gasless transactions or cross messaging protocols actually like Conex they rely on our infrastructure to get transactions included on 14 different evm chains because we have a broad sort of spectrum so we are also servicing F platforms and infrastructure protocols um also social and gaming usually from the X perspectives but they are very very web free native I would say our audience is very we free native but what uh is I think something which I find very bullish about uh the current sort of uh sentiment that we are in right now in the market as we in the bare Market but I'm talking like for the first time actually in the history of gelato we are talking to like NASDAQ companies that come into our Discord and want to integrate our Technologies to build their own web three products right so and we are not talking about like two developers anymore that are just building up a new token or like a project we're talking about like developers from like publicly listed companies that come in and want to leverage these tools to build applications so I think uh we are seeing more and more a shift and focus to talk to Enterprises actually so big companies that want to leverage web3 and the tools so I think there we need to just as an industry make a shift from assuming um our customers and projects that integrate us as like infrastructure services are these Deens are these web fre people that know all the quirks they are now Enterprise B2B developers that are very well aware of web 3 and they know all the ins and outs so they're very well educated but Their audience is a completely different one Their audience is a web to uh industry a web to sort of uh customer base so I think this is a large shift that is happening right now for us internally and uh yeah we still have to work to make all the communication material and everything sort of tailored to that but that's make that makes me actually very bullish because we are seeing that transition right now I have a question for you you mentioned maker uh just now and um as we may know maker is making a a onchain proposal well not onchain a proposal right now to switch to a salana based uh Fork um and run their operations there uh what do you think about that and how are you going to uh support them or are you planning on supporting them if they make that switch that's a very very good question so there is a movement H and we are seeing it very clearly there's a movement to app chains there's a movement for protocols and projects and applications to move to their own execution environment that is not ethereum but is entangled and connected to ethereum in terms of liquidity and minimizing their trust um but why do they want to do it okay they can get transaction fees down right uh but they can also have more control over the ux and how users interact with it and all these services that for example also we help augment on like these public available blockchains they want to access these services in a more sort of cohesive and inherent manner on their blockchains and I'm talking about transaction life cycle management so before each before this transaction get included run this logic after this trans transaction is included run that logic right so basically supercharged evm Powers what we kind of like try to do and that's a that's actually a big movement I'm seeing and that's why we are also positioning ourselves um and uh very soon actually I think in toen 2049 we have an announcement in that regard as more in the L2 rollup world and I think I see these projects moving to their own app chains and there we want them to be able to leverage galal Services much more tightly integrated they can today on like an ethereum where it's just like a module that is added on top but I to answer your question I see a lot of projects actually doing something similar not talking Solana I think I'm not 100% convinced that that's the right move but if you need low latency distributed systems like pias they also for salana then salana code base is a really really good one so it's just uh quite an overhead I would say that's actually probably a very nice segue into maybe the last part of our discussion which is let's talk about um for the the lack of a better phrase some of the battles that are raging at the infrastructure layer maybe we start exactly with this topic which is modular uh or rather monolithic blockchains versus app chains or Sovereign chains right as you described ilar I think um there are natural trade-offs that developers have to think about in terms of um how much control they want to exud over their users user experience versus the Monumental challenge that exists if you choose an app chain when it comes to bootstrapping your own security ecosystem Etc and so I guess the question for you both would be um any perspectives if you're if you're a developer thinking about Which choice to make um any perspectives on how to think about these Alternatives and to your point made earlier what tooling still needs to be built to make the app chain path easier yeah so um I think this is a conversation I've had with quite a few teams basically helping them make that sort of decision okay so whether do I build on ethereum do I build on Solano or do I build my own chain um I think for the great great majority of Builders they don't really need to build their own chain because uh generally speaking it takes quite a long time uh and effort to uh build up the proper infrastructure or be able to use the proper infrastructure learn rust learn go or whatever um that's needed to actually kind of get that entire app chain running properly as well as be able to make that app chain interact with other ecosystems so um for the set of users that kind of that should build their own chain I think those are generally users who um have already like a huge Community uh so for example like if maker for example wants to Fork off of eum Main net or whatnot and build their own app chain I think that could potentially be an interesting decision because they already have a large set of users that support them and fra as well is building a Frack chain as well because they have a large set of users St supporting their stable coin and they want to kind of build their own defi Community off of that um the other set of users that where it makes sense to kind of build an app chain is basically somebody who um does not is not satisfied with kind of this oniz fits-all idea of um being entrapped by something like evm or another smart Contracting language because there are inherent limitations essentially of programming uh and building applications in those languages uh so generally speaking uh if you're building an app chain you could build on the like the runtime level and then you could basically use some sort of modular stack let's say uh and then combine like let's say there's some dii modules like that you want to put in or there's some nft modules that you want to put in some Bridge modules right and then so you can combine the right kind of tooling necessary needed to kind of make your blockchain as performant as possible for your particular application uh but however that's only a small subset of potential users there uh I want to disagree here with this statement um and maybe have a more spicy take than this so I think actually we will soon come to a point where it's not even a question about like hey should I launch my app train or not it will just be an app Train by default it will just be and this won't be like hey I need to run my own Cosmos based chain and I need to spin our validators or whatnot it's just I will go to this dashboard I will click this button my app chain is there in a second it's up and running I just get an PC I deploy my smart contract all the services that I need that I explained back then when I sort of talked about the stack are available to me by default and I have a full control over the execution uh environment uh what should happen in the transaction life cycle how the uxx should be handled if the if users should be paying fees how do users identify themselves and so on so I think they are being um there's a huge progress right now in the space uh we are personally also working on exactly this and um I believe that it's still early obviously right uh and we still only see a handful the first networks launching like Bas Zora um there are I think we are talking to 20 20 other chains that are basically in the process of transitioning to the app chain and they are not only huge chains huge applications but I think it will be the default uh very soon actually and then the discussion about like which chain I should launch on stuff is over it's just like you are your own chain you might be sovereign chain or you might be more entangled into ethereum um and then the question is just okay what is missing yeah what are the tradeoffs of your own chain right that's basically what is missing the infrastructure is there you just need to sort of get it over with you just need to migrate it you not you need to like a have a platform that makes it available to you but the question is okay you lose Atomic composability what is atomic composability it's basically uh flash loaning from AA going leverage on maker swapping on Unis and going back right this is without shared sequence quencing um it's very difficult to achieve right but the whole internet is asynchronous right you make a request to another server you wait the message come back comes back so like the whole internet was built with this premise of mind and I think actually not many applications need this Atomic composability and you can have uh a lot of cool user experiences without without that with all the other benefits you have like scalability no fees cre your X and identity so that's why I'm I'm more bullish on this makes sense the other um you know more contentious area that I'd love to get your guys' opinion on that uh you know Eric alluded to a little bit earlier in the conversation is uh sequencers and to what extent they should uh be decentralized and maybe for the audience's broader context to the extent you're not familiar the you know layer tws uh are really coming into their own uh TPS on layer twos in the ethereum ecosystem has surpassed ethereum main chain and sequencers play a key role when it comes to to broader data availability and transaction ordering basically you know um sending the state back to to layer one and so the question is a little bit um and the debate ra raging among some friend groups not not my friend groups but some is to what extent these should be decentralized right should it be something purely decentralized almost to the point where it's permissionless or should sequencers be managed and overseen by a centralized entity let's say coinbase or something in the middle curious to get your guys's take yeah um I personally think that sequencer should be decentralized and there's really two sets of two ways to in a sense decentralize that so the first way is essentially with a shared sequencer so essentially uh how that works is all these layer twos um they share a common set of of you could say validators in that sequencer set and then uh there's a defined kind of rule set to join that sequence a validator set and um that way basically you can kind of trust in a sense that all of these layer twos are treated in a uh you could say the transactions are ordered and treated in a certain way uh when it comes to kind of submitting it to the uh ethereum layer one chain and the other one would of course be like a decentralized set of sequencers but within the I guess within each layer two itself they have like their own set of validators uh I think most likely uh they will kind of most likely initially it would kind of be just like private set of sequencers like by each layer two because uh different layer twos have their different um I guess business agendas and such uh but either way uh one of these kind of Futures should probably be realized soon because there's a lot of debate right now there's a lot of uh kind of concern about if you're like a user on optimism or some or arbitrum or whatnot uh whether my transaction will be like MD but basically by the people who actually run the sequencer so the centralized entity who runs the sequencer will be able to see every transaction in the men poool and they'll be able to order it the way they desire and um if you like if that entity ever gets hijacked or misbehaves then uh they'll be able to kind of maximize that me and basically trade against uh the own users who are working at that layer 2 which makes it like very difficult essentially to uh in the long run kind of operate a centralized sequencer yeah so I disagree I think um most sequencers should be centralized um and will be centralized why the reason why we are doing l2s is so we can finally scale decentralization always introduces latency and we want to have low latent High scalable cheap uh ecosystems where the um um chains where the execution that happens is trustless uh using validity proofs for example with ZKA um rollups right where we don't have to trust the Integrity of the rollup of the sequencer operator but we can enjoy the centralized scalability right and obviously if you use ethereum as a data ability layer you have full full overview of like the sequencing and the ordering and if a sequencer starts meing right then you leave the chain right then you go to a chain over a operator that doesn't M you in the to the extent that the the centr sequencer does so I think we will see a world of most of these especially Enterprises like Enterprises won't share sequencers right Enterprises will be this is my sequencer it complies with the kyc AML regulations of South Korea only individuals that are registered here can actually interact with it and not anyone else and they will interoperate maybe with a few other local sequences right maybe you have like small shared sequences in that area but um otherwise we will have these defa ecosystems that we have before where have like maybe layer twos like arbit optimism where multiple applications aren't there and they can be potentially more decentralized but I think the majority of them will actually be centralized because they can and it's and it's cool and it's still trustless there you have it two sides of the spectrum and with that I think we're just out of time thank you all for listening and thank you hmar and Eric for participating and for sharing your views thanks thank you everyone thank you ladies and gentlemen once again give a big round of applause for our moderator and our two guest speakers