Fireside chat with ZKsync I Alex Gluchowski & Kartik Talwar at Pragma Denver 2025

[Music] and next up is our third and last fir side chat for today and for this I'd like to bring on Alex who's the co-founder of matter Labs a lot of you who know them as ZK sync so please welcome Alex on stage Alex excited to be uh doing this again with you um always a pleasure the best part is that every time we do this uh there's a new update going on in the the zy syn ecosystem so we just get to continue our conversation from from years which is awesome um we're gonna talk about the elastic Network today and uh my kind of first question to you is uh tell us what the elastic network network is and uh that's been a lot of need things coming on sure so elastic network is the um a zare rollup class which is the third iteration of the ZN protocol um what is a rollup cluster it's a relatively new emerging concept um it's a collection or a network of of chains for UPS that is designed for enabling seamless interoperability seamless uh interactions between the individual accounts on those chains and this is how it differs from just a random collection of chains like it must be designed from from the beginning um to remove the boundaries while keeping the identities of the chains uh intact like they have they can have their autonomy or sovereignty they can have different properties you can think of it as a form of non-fungible sharding but the the connectivity this this Unity of these chains is really key important and in this case Every Chain can have a user with different state or different balances but they still behave like one or it's just seamless to no they behave completely independent so each of the chains is powered by uh our framework for building blockchains for the ZK stack it's the exact same framework which is powering zyyn era the first um uh chain in this network and the all all of the chains are completely independent so you can think of them as just you know separate layer twos layer ones it's awesome um the thing I'm actually kind of curious about is historically we've kind of known what ZK sync was trying to do with with the ZK VM and kind of being one of the first rollups um but if I'm not mistaken this is a new sort of Direction where you're allowing more chains to be part of the network instead of kind of making ZK sync the chain actually scale up and and uh handle a throughput um what kind of made you to get to this decision and sort of what's kind of the framing around what made that even more possible um it's very simple we're following the market the market for general purpose chains is quite quite saturated like optimistic rollups started relatively early they they managed to get a large market share then there are alternative lay want the um it I don't see a need of launching one more uh general purpose chain that does not have something unique something special that it brings to to the table on the business side or you know like they have to specialize they they have to bring something unique uh uh but uh we see a huge demand for change being built uh for special purposes like application specific chains exchange specific chains uh chains launched by Banks and other financial institutions um chains launched by communities like they we have gaming chains that are uh uh they started off as communities on on different blockchains and then they realized at the end that they want want to launch their own chain uh it it seems a very clear path towards a future where we will have hundreds if not thousands of tens of thousands of chains um all working together just like internet you know if you think of the internet um it doesn't run on a single machine on a single server or even on a single data center it runs on many servers they all connected you can browse you can navigate to every website uh in exactly the same way we're using the same browser the same interface the same UI but all of these websites are different and they perform different functions and they run on different Hardware uh but they are unified by by this shared uh interface and this is something we're replicating like I think this is really the end game for blockchains for for ethereum um as a settlement player as well no that's actually a really good analogy so I have two follow-ups from what you just said one is in this world what is the role that ZK era plays and then the second piece is what exactly is allowed to be customized here for others building on ZK stack is it just that if somebody says this chain is going to be for my community but everything is still possible those are still general purpose chains or they can change different properties and if so which ones um you you you you can you can customize practically everything except for the core engine that connects the chain chains together you know for for the system basis of the chains and the virtual machine that must be secured by Zer knowledge proofs because otherwise you defeat the purpose of of a zik op clust and we can talk about this um and maybe just just so that the audience has a slightly better what falls into um like being customizable so yeah uh so you can customize let's start with data availability do you use ethereum for data availability and you are actually a rollup or do you use something external maybe Celestia or Avail or I da and then you become a validum or maybe you use your own custom data availability because you're a bank you want to control and keep the data private and not just close to the world then you fully control the sequencer you can decide do you want a centralized sequencer which is going to be high High super highly performant low latency uh thousands of TPS or you want a decentralized sequencer and this is going to be choice for a lot of applications because you can then give the uh Power of sequencing to the community and you can structure it can launch your own token which will power the staking of uh of this decentralized network and will ensure sensorship resistance and and and decentralization much further um you can um um customize uh the rules of the chain like who you let in who you let out like how what types of transactions are allowed what is your base token could be e but could be also your custom one uh and so on that's awesome I mean that exactly is the flexibility people would like so yep full Freedom so in that case what is anywhere from the the workflow to the ease of now having chain stock to each other here and um kind of can you compare that to the other experience that people have where they're going between cross chains or let's say we use super chain as an analogy how does this differ between interchain communication um so interchain communication is is something that is going to Define rollup clusters and they're going to make a difference in like this is how how what what what's what will make one or another unique um the uh the kind of the in this end game that we described that there are thousands of chains the interrupt must be completely seamless in order for this to function is the this internet of value you should be able to connect to any of the chain like you you you as a user you will probably have your funds on one of the chains on your home uh location and you should be able to do what you do now today with your uh ethereum wallet connect to an application and then just click interact and you perform the function and you will be aware where is what is this application where what chain does it is it like on um what security problem what is it offers because they they might differ validium is very different from a rollup but the interaction must be completely transparent and seamless in uniform you just do like one click confirm something happens you send tokens over uh some some interaction occures and some results come back and the the difference of zq rollup cluster compared to something like uh optimistic rollups is that we use the magic of ZK to guarantee two things infinite scalability so there is no fundamental limit on how many chains can join the system and what their throughput can be which is not the case for optimistic uh roller coasters uh and isolation you like whatever like if something happens on one chain something wrong you like maybe whoever launched the chain is rugging you it should not affect anyone else should be contained any other or any user should not like well it cannot it should not affect any other chains obviously the users of the chain which is failing will be affected right and so it's your RIS your account on a different chain under the same private key or you mean just but exactly but your your account uh on any other uh chains uh will be unaffected and you will and and your your tokens your your state there will be completely free and the way this is achieved is very interesting um it's like what ZK as technology is making possible is blockchains that can't be evil if you think about this this is basically like this is what what we build blockchains for this is how Bitcoin started this whole idea of don't trust verify don't depend on any Central Authority you just um are equal peer with everyone else in the network and collectively you make sure that Bitcoin is uncorruptible it cannot be evil right so this works for Bitcoin because the full nodes are very lightweight this works for ethereum also because the nodes are lightweight and and the culture of ethereum keeps it this way um the verifiability must be there but it becomes very hard when we go in the territory of like chains launching with thousands of TPS or you know like when you have like hundreds of thousands of chains because you cannot anymore run physically verification for all of this chains so now you have to trust someone or something and in our case you will have to trust something it's math cryptography open source code which you can verify from scratch or rely on other people verifying you can guarantee okay this is I'm I know the code I'm running and then all of these chains cannot be evil and will do whatever uh they are supposed to do according to to you know like the properties that they expose how does having now external data availability work here because you are effectively trusting that data is going to be secure and verified for some so that you can have the proofs work on the chain is that a good way to understand this um so the uh data availability um can be completely trustless and this is the preferable way to do it which would like the way to to do it on ethereum if you're settling on ethereum your chain is to use ethereum's data availability the blobs or even call data for publishing the State updates of your chain in which case you like any user can run ethereum full note can guarantee that they get the data so like it can completely not be rocked um that will not work for everyone like this is not Limitless ethereum data availability will have some bounds so in order to exceed that you will you will like some chains might want to opt into be validum publish data elsewhere then only the data censorship resistance properties will be dictated by your U data availability solution so like if you publish on Celestia and then you you trust the majority of Celestia validators for example right but then if something goes wrong if they are becoming malicious the only thing they can do is to shut down the chain but they cannot force you know they cannot fake transactions they cannot interact on your behalf they cannot steal funds or corrupt the state in any way so it's like like even with validium you have half half with they halfway there for full trustless and and cannot be evilness no that's a that's a that's a good way to think about it because you can't force change you can only just verify that a state was changed um maybe another simple quick question um do you run into any issues with cross chain transactions or messages when different chains have different base tokens that they're using for fees and um not having a balance on one or the other like how does that work in this kind of super cluster um that that that's a challenging problem for many chains uh it's easier forys zik sync uh protocol to handle it because we have native account abstraction from day one so any chain can interact with basically you can pay fees in any token you have to support that the chain has to kind of enable um either the the the sequencer of the chain directly enables accepting tokens of different kinds or you're on a pay master and you can convert it just in time within the transaction and this makes the composability very easy so you can use basically any like you can use stable coin which will pay and like you use stable coin you are making transaction that will go across multiple chains and you will pay in each of them in the currency in the base token of that one chain and you will just convert it just in time and this is challenging the like the the interrupt design is a very complex space um you have to begin with a right user experience like imagine like a perfect user experience what I described in the beginning you can connect to any chain one click everything works and then you have to work all the way back backwards from there and like okay how do we reverse how do we actually get there um I guess kind of on that similar note it's obviously a big discussion topic uh in our space right now but um let's talk about sequencing and decentralizing that piece um kind of how are you thinking about it how do you think about the timelines and just generally kind of what the plan would be and um any kind of announcements that you've been working on to make that simpler uh well we our team recently published a big milestone on decentralization with uh a consensus mechanism without which is permissioned so without staking yet but it's it's working and it's on the test net and the next state stage will be actually uh fully running it and implementing it for for era with staking the token probably with staking ZK um the chains will be able to also do whatever they want with with this like this is this is critical for era and you asked me about era S I didn't answer that question era plays a crucial role in the elastic Network as the first Z roll up and the place which acts basically like ethereum of of this elastic Network this credibly neutral space where everyone can deploy and have highest sensorship resistant guarantees in real time our goal is to eventually make it like full real time you know like make it even better than ethereum itself maybe use cryptography or te's or uh um encrypted M Pools for like 100% making sure that every transaction can go in this is important because all the other chains could be run by you know individual operators or validators this must remain the community space where the Voting is happening on the governance of uh of the elastic uh the entire elastic Network you know like we have um close to half a million uh individual addes holding ZK like it's a very very broadly decentralized Network and it's also going to be the place where uh a lot of Integrations happen uh already happened like we have Integrations with all a major Bridges uh onramps uh oracles um uh other infrastructure which then can be all the other chains especially app chains can directly tap into this without having to integrate all of them all of it themselves which is a very very costly very expensive and cumbersome process and maybe just to clarify are you saying that this is the only way for every chain here or do they have the option to do this on their own they can do it on their own but like why why would you want that cluster I mean like you you you're building a house you have a option to tap into the grid and just get electricity from you know like from from National Grid you can also build your own generator or your own solar panel if you want to but like it's a lot easier to just step into into an existing grid so for some special purposes you know if you're building a hospital you probably want a redundancy power generator but like most residential houses will just step into the uh the grid this it's the same situation here most you know like some chains might want to build their own direct Integrations for whatever purposes maybe some special business relations or whatnot uh but most app chains don't need to bother about this it's going to be very very lightweight for them to start a new chain just like as easy as starting an application just deploying on on somewhere but you have all the benefits of the chain and the and then you can run like very easily launch a decentralized network and the incentives for launching that and launching your own token and having like proper decentralized process and value acral mechanisms with your community are powerful and I I expect most if not all successful apps to eventually become their own app chains and this is this is going to be like a really really big moment right now the home for for the app chains is like it's not quite ready like Z syn the elastic network is not quite ready the optimistic rollup clusters are not quite ready like none of them has interrup life z s is Forest long we have at least the shared Bridge life so when you launch a chain you can directly tap on you can build on the shared Bridge once interrup is working it's going to be available for everyone not you know special treatment of the small group of chains that are uh having some preferences um no actually that's a very helpful way to kind of think about it Sor I was mistaken on one of the pieces there so thanks for clarifying one kind of thing I'm curious about is obviously we talked about a lot of benefits but these things are trade-offs at the end of the day so what are the bottlenecks in this model or what is actually going to not be as easy compared to the other Alternatives and I love to kind of have you tell us about the other side too um sure so the right now we expect all the chains in the network to be ZK rollup like ZK chains chains secured by ZK um which comes with tradeoffs specific to to to the for clarification does that mean they're running the ZK stack and that's the for now it's it's only the ZK stack this can change in the future but for now it's it's ZK stack so like there are limitations with the ZK St ZK St is right now not evm uh compatible not evm equivalent at bite code level this is changing like very soon we will ship a uh uh an upgrade which will enable bite code compatibility where like all the ux is completely going to be uh uh abstracted away and and will be as seamless as as on layer one other chance um the uh proof costs are currently non- negligible they are like something in the range of like 0.4 cents for for an average transaction on era today uh this is going to go down massively with the uh bo jum 2.0 which we're going to present uh uh I think on Friday or or Saturday at the uh if Denver uh the the new benchmarks it's it's very very interesting very awesome uh it's going to bring costs massively down and we expect the uh further cost reductions over the coming month and and years uh at that point it's going to be I like it's hard for me to point to to any disadvantages like I really believe that ZK is the the end game for uh for ethereum for all blockchains and I think we have a broad consensus in the community about this that's awesome so maybe just to complement that or have you digging a bit more as my last question like what can we expect more this year um in addition to evm equivalence and evm equivalence interrup uh more updates on the on the front end side stage one maybe stage two we we like we we'll see like I I I I became a lot going on this year I'm I'm now more confident that we can come to stage two because there like stage two is scary for for any rollup cluster but I now I think we have found a a good um tradeoff space there um decentralized sequencer and much higher throughput we we published the road map if you just Google zik syn road map 2025 you you will see it all in one picture so everything that people have been wanting gets delivered this year that's this year is going to be big big year for ZK pretty awesome well Alex thank you so much for this amazing conversation thank you carik please give him a big round of applause [Applause] [Music]