Bringing ZK Proofs to Developers Everywhere | Uma Roy

Transcript

the existing Paradigm is you have to have this team of like 30 to 50 people who are all like very specialized and then you hand code like a ZK system for the zkm so you like write this very complicated circuit it's very hard to audit and the problem with having this team that's built this very complicated ZK circuit is that if someone else wants to take your technology and use it as a rollup stack or framework it's very hard cuz like they can't modify it I think that's why when you look at the existing landscape of rollup stacks and like Ras broadly you'll see that op stack and the arbitrum stack is really popular because like it is very easy to modify it's you know um they they've been used a lot and so I think the yeah existing

shortcomings of ZK rollups today is that you know it's really brittle and not that flexible the polygon Community grants program was launched with 1 billion tokens all four polygon Builders season 1 is live right now and features 35 million in madic to support the next generation of polygon products builders on polygon these grants are for you join the aggregated future today by applying at polygon.net on zerox Research is a recommendation to buy or sell Securities or tokens this podcast is for informational purposes only and any views expressed by anyone on the show are solely our opinions not Financial advice bachio Ryan and our guests May hold positions in the company's funds or

projects discussed what's up guys welcome back to another episode of zerox research today we're joined by Uma the co-founder of sixin thanks for joining us Uma hello thanks for having me um before we get started do you want to tell us a little bit about what you guys are building over at sixin yeah um so at the synct we're building um the tools to make ZK easy and accessible to all developers uh in particular we're building a ZK VM a ZK virtual machine called sp1 that allows any developer to just write normal rest code and then use ZK and we're also building a decentralized pro Network so that any developer using sp1 can

generate proofs using the pro Network without having to have their own infrastructure and get proofs provided by a wide variety of these provs on the network what inspired you guys to build six what were some of the problems that you guys wanted to solve yeah so my co-founder John and I we had been working in the ZK space for a little bit uh in particular had been working on some particular applications and use cases of ZK and we had seen firsthand that ZK was really really difficult um you had to write these custom circuits it required a lot of specialized expertise you had to have a deep knowledge of cryptography and it took a really long time to take all these

things to production and we realized that ZK wasn't going to go mainstream or get adopted if it continued down that path um and the holy gra of ZK has always been that people can just write normal code and it just works um and we we also really believed in that vision and that's what inspired us to build sp1 which to just make ZK really easy and accessible to any person not just someone with an advanced cryptography PhD that makes sense um and the SP sp1 is a ZK VM how does what does that exactly look like and why are ZK EVMS needed historically in ZK uh you've

needed to write your function that you want to prove as this thing called a circuit um where you basically write your function in terms of these like primitive operations like additions and multiplications and you know maybe some other Primitives but it looks a lot like coding with assembly so it's super low level it's very difficult it takes a long time with the ZK virtual machine what happens is you have normal rust code it gets compiled to a particular bike code format so in our case it gets compiled to the risk 5 Isa so similar to how solidity in the evm gets compiled to the evm bik code and then we provve the execution of risk five um with your

particular program with a set of particular inputs so instead of proving your particular function or comp computation we have a circuit that proves risk 5 and then any program any risk five program can be proven by our virtual machine uh and why that's really powerful is now uh you can just write normal code compile it to risk five which is a standard uh compilation Target supported by uh rust's compiler so it's like a very standard Target we didn't have to modify the compiler or anything like that and then any rust code can get proven within sp1 that's awesome so kind of um cuts down a lot of the development time you don't need to know circuits to write it

you just need to know rust or maybe some other kind of like llvm but um so when you were making this what did you notice any of the tradeoffs it has over other VM environments or other types of like CK proof generation yeah so the main I mean the dream of Just Right rust get appr proof is it is kind of the dream um and so the tradeoff or like historically why it hasn't been done is that it's been really slow so of course it's like similar to any normal program where if you spend a lot of effort you write it all in assembly it might end up being faster than if you just write it in like a higher level language um and so historically ZK PMS have been much slower than if you wrote circuits um one

of SP one's like biggest Innovations was that it was actually really fast and performant and so it kind of made ZK BMS actually practical for the first time for real use cases and and there's like a bunch of techniques we have in sp1 to make that possible uh both on the ZK algorithm side um and also on the performance engineering side um I guess one of our biggest techniques is this usage of or this notion of pre-compiled where basically one key realization we had is that for most workloads that people care about for example uh verifying you know the ethereum state transition function for ZK rollups or verifying a proof of consensus for

bridging uh or really any of these things uh most of the time and most of the Prov overhead comes from very repetitive cryptographic operations like hash functions and signature schemes and those can be really expensive in a zvm because you are spending a lot of Cycles doing this very computationally intensive thing and we realize that um if you are able to have a specialized circuit for that particular operation and have that within your VM you can kind of get the best of both worlds where you have the performance of circuits where you need it and then you have the flexibility of the VM so for us uh in our zkm sp1 we have pre-compiled

for catac and Shaw uh these are like very common hash functions used in ethereum Worlds and for a bunch of signature verification schemes and that makes our VM super super performant and actually practical uh so in practice we've actually found that our VM ends up being a lot faster and a lot better than even the circuits we used to write uh because with our VM also we can optimize it a lot because it's so Broad and it's such a big platform uh that every Improvement to it improves so many different use cases that we can spend a lot of engineering effort optimizing it and so in practice we've actually found that it's a lot better than circuits and it's a lot easier to build in because it's just writing normal code uh so it's

actually kind of strictly better than old Paradigm um yeah in both in all Dimensions like latency speed cost and developer experience that's awesome it's not often that in blockchain you get like something like peretto efficient with you know the tradeoffs are like so much more worth it um I can definitely see how those like pre-compiled will make things a lot like faster and easier just a quick couple questions on that um do does succinct make all of the promiles can other people make their own and if they can make their own are they allowed to share it with the rest of the network as well yeah so sp1 is fully open source and one of the reasons we made it like

that other than of course that being really important for security and auditability uh like if it's not open source who knows what you're actually proving um is that we wanted anyone to be able to contribute a pre-compile um to sp1 and actually a few external teams have already done this for like their specific use case so for example if you have a particular workload it has a very specific computational bottleneck you can write your own pre-compile and then incorporate it into sp1 uh for right now actually the CNC team has written all the pre- compiles uh that are used um because honestly there's not that many functions that are bottlenecks um and so the current ones

we have are kind of enough um but in the future we hope we're like super happy to take external contributions from other teams uh for like commonly common pre-ms that we think will be beneficial to a lot of people related to you guys being open sours it's actually something that I read that you guys touch up on is that you're modularly architected I kind of want to ask about why this is important and I guess how other ZK VMS have gone wrong or they've done wrong in terms of being close Source yeah so I think first first of all being open source is just really really important for security and auditability like if you don't know what the constraints are then how do you know

what's actually getting proven um and so open source from that perspective I think is a must have uh open source is also really nice because as I said now the community can add their own pre-compiled they can modify it they can build on top of it so for example um some people want to verify sp1 proofs on a layer like Bitcoin or something like that or you know different execution environments and because it's open source they can take the proof system they can maybe modify parameters and it's just really customizable and flexible and I think for example you've seen being fully open source and modular has really helped something like the op stack take off and kind of become the

canonical rollup deployment stack um because it's easy to customize um you know there's no risk of you know there's no licensing risk or anything like that and so um for us like we view it very similarly like um our prover network is kind of like where we think we'll be able to provide and users with value so like sp1 is open source but running it on your laptop isn't really a fun time I think most people will want to use sp1 via the prover network uh so we feel really good about that and we think that open source just makes it so that there's more Community around it it becomes the canonical shelling Point VM you can have a ver ified in a bunch of different environments you can have a

rich community of like pre-compiled um and yeah I think that's why we open source is really key to like our strategy yeah that makes sense so it's kind of like you it also helps to bootstrap The Proven Network that you're you're also trying to build right now um and then you know speaking of that I think that's like something that's probably missing from the industry right now like a let's just say like a decentralized Prov Network like is that something that you can talk about a little bit more and maybe even like more generally for the audience the proof supply chain yeah so I think today generally um if you want to use a ZK um generally like most teams are running

their own provs in the cloud and one aspect of ZK is that the proofs are self-verifying so there's no trust assumptions with teams just running provs in the cloud right like that's totally fine they can't generate false proofs um but I think there's a lot of advantages of a network um one is that you know it is important in the end that it is decentralized for like censorship resistance and liveness guarantees and things like that I think more importantly um if it's decentralized and there's kind of like a competitive pricing mechanism for all um provs in the network uh it'll lead to like better price competition and then also drive down the price of proving uh

so I think that's another thing that's not talked about much for Prov networks is that it can actually reduce costs and reduce latency by inducing this competitive environment between a bunch of different provs or kind of competing against each other to generate proofs for people so in that sense it kind of looks like a Marketplace uh with really good decentralization liveness guarantees and like price guarantees um and yeah I think there is no currently live prover Network um we're working on building one uh as a part of you know sp1 uh today I like to say that we kind of have a prover network with one prover which is us uh so if you want to use sp1 you don't have to use it on your laptop in fact I recommend you don't use it on your

laptop because again that's not going to really be a fun time uh you just change one environment variable you can send proofs to the pro Network um and we'll generate the proofs for you and send them back and then as we get like further along that we're going to we've architected in a way where uh we're not the only we don't have to be the only provs on the network and as we go further along that path and develop it out more we're going to start onboarding more provs and yeah actually making it a real network with like many many different people participating what does that onboarding process look like onboarding new provs and what's the timeline for that I think that's something uh we'll focus on over like the next few months um in terms of

onboarding I think it's actually pretty simple uh you just talk to our um our networks like so generally what happens is you have an application that's trying to use ZK whether it be a rollup or a bridge it sends a request to RPC dosc doxyz uh that gets stored and then if you're approver uh you can like talk to the RPC you can claim proofs and say hey I'm going to uh claim this proof and then I'm going to actually generate it and then once you generated it you fulfill um the proof and you say okay I submitted my proof the proof gets verified and then you get payment for it um and so I think the onboarding process will look like pretty simple you just have to like basically have methods for

claiming and fulfilling and then actually generating the proofs um and then different provs can also for example do like Hardware acceleration of sp1 or really optimize where they're running the sp1 pr so for example you might have like a better costs if you're running your prover is not in AWS and stuff like that um yeah so I kind of like imagine that is uh how the onboarding process will play out um and for for The Proven Network have has any thought been given into like what what fees that the provs will be rewarded in and um is will will there be any like staking to make them

accountable yeah so one important thing in the network is that the provs so the provs you don't have trust assumptions on the provs right because like they can't generate a false proof unless there's a bug uh but the hard part is making sure that they don't grief the system so it's bad if approver says hey I'm going to generated proof and then doesn't actually end up doing it um and there's a few different mechanisms for ensuring this like accountability and ensuring there's not these like griefing attacks basically and we're kind of still thinking through the mechanism design on that side it's so that's still pretty early but we have some interesting ideas that um yeah maybe we will talk about in the next like few months oh okay that's interesting um uh

do you think that it will be more I don't know if you can answer this but do you think it'll be more you will use an internal kind of staking mechanism or maybe you'll use something like iron layers reaking um I think for enan Lay um generally for Igan layer reaking you have like existing ethereum node operators just running more services but I think proving is a very very Specialized Service that maybe it doesn't like if you're an ethereum node operator it also makes sense that you'd be running nodes for other networks I think proving looks pretty different from that right like you may need gpus you may need like specialized hardware and stuff like that so I'm not sure that reaking based networks necessarily make

the most sense um so yeah I'm not sure yet um as to like what exactly that'll look like that makes sense um I kind of want to talk about the sp1 and who you view as the primary customers for is it rollups and app chains and if it is is it new ones um existing ones I think our primary customer for like the next you know year or so is rollups I think every rollup is going to be a ZK rollup like I I've you know talked about this a few times like the current Paradigm of all this optimistic rollup and kind of this like infrastructure soup that's needed to get them working like you have liquidity layers and then LPS on top of

the liquidity layers and then like a bunch of different Bridges with like varying trust assumptions and then you know all this stuff I think all this stuff fundamentally doesn't really work and it's also really expensive like people like to think ZK is expensive and um I I mean the costs of ZK are rapidly coming down because it's like we're making progress on the algorithms of Technology but the cost of like all this you know Mosaic of infrastructure is not coming down because fundamentally there is like this cost of capital uh there's this cost of having Capital locked up for seven days and then there's this cost that needs to be paid to all these infra teams and it introduces like a lot of friction so I actually think optimistic rollups are very expensive people like to say oh today it's free

well I think that's a total myth like it's not actually free it's just the costs are much more hidden and then paid by the users and it's like a lot more off fiscated as what's actually going on but I wouldn't be surprised if it was actually like eight figures or even like low nine figures for big chains being paid out to like all these different teams and all the LPS and all the layers and all the bridge fees and stuff like that um and so for me I think like yeah the current Paradigm of optimistic rollups is not really going to make it um it doesn't really scale which is maybe not even that surprising cuz it's not I think like true scaling right like ZK is actually scaling it's like actually verifiable compute the optimistic stuff's more of like a it's like a clever

trick um and yeah so I think every Rob's going to be a ZK rollup I think the only way ZK rollups are really going to work is with a ZK VM like the existing Paradigm of writing a complicated circuit uh specialized for the evm that's like impossible to upgrade when the evm upgrades it's like really un auditable it's like hundreds of thousands of lines of code like I think that doesn't really work and so I think every ZK rollup is going to use a ZK VM and then I think like we want to be the best ckvm and so yeah I think for the next you know year so we just think we just want to convert all rollups to using sp1 like that's our our biggest Focus um I think there's also other use cases like

Bridges and interoperability so for example polygon's AG layer protocol which is like an interoperability protocol is built using sp1 I think they're doing like a lot of really awesome work bridging obviously is in like a similar category and so we're working with like Celestia and ail to bridge their da layers to ethereum using sp1 um and then there's a bunch of like Oracle use cases so I think those are all big use cases but I think rollups will be like rollups are a really high volume use case that I think we're very excited about that makes sense I have two questions uh following up from that one is one thing you mentioned is that ZK costs have come down quite a bit and one thing I feel like I hear pared a lot

is that ZK is expensive ensive and like optimistic grows are cheap can you I guess paint more of a picture of how much ZK costs have come down and how that's like what has happened over the last year year and a half maybe that it's become drastically cheaper yeah totally I mean I think even with sp1 I think we originally released it in like February or March and since then we ourselves have improved our own performance and cost by like 5X and we have like this long list of ideas and notion um to improve the costs by another 5 to 10x by the end of the year and so I kind of see that every day very

viscerally like we wake up in the morning we go to the office at 10: a.m. and our number one priori is just like make sp1 cheaper and faster and better and like every week basically we improve it by a little bit um and we have so many ideas on how to improve it way more so I think like for me actually like I've seen even the cost of sp1 go down by 5x in a few months and I see like what we're working on to make that that Trend continue so I kind of like know it to be intuitively true um in terms of metrics for example uh for the existing ZK rollups uh so like ZK sync polygon zkm scroll the proving costs are actually pretty small um compared to

like fees paid to ethereum for da or like you know verifying proofs on ethereum which can be expensive it's like 200k gas or 300K gas um and so the proving costs are actually already pretty small I think it ends up being like maybe between 0.1 Cent to 1 cent per transaction which is like pretty cheap um and yeah I think that was totally Unthinkable even like a few years ago and I think it's only getting like cheaper and cheaper hey guys permissionless the event for crypto natives is heading west to Salt Lake City and we're bringing the biggest names in crypto together for an incredible can't miss event join me in the rest of block works on October 9th to hear for more than 200 Industry

Titans including Nick white from Celestia shiram from Igan and Keon from Mona the conversations that permission list will be covering the hottest themes in crypto including modularity reaking the Bitcoin and Solon ecosystems Ai and crypto rollups l2s institutional adoption and of course the November election that will be right around the corner we couldn't be more excited about this event and it's happening in one of the most beautiful places in the US ticket prices will be increasing towards the end of June so get your ticket today make sure to use the code so I get some credit and bragging rights from the rest of the team uh use ZX10 for 10% off on your ticket one more thing we're also hosting the permissionless recka font in collaboration with bbx the theme this

year is a multi-chain future to build a better cross chain experience so I don't have to have a heart attack every time I bridge to a new chain all Hacker found participants will receive a free ticket to attend permissions so if you're a developer come join us in Salt Lake City team up with fellow devs and build the multi-chain future polygon Labs is developing the next generation of open- source ZK Tech to aggregate crypto liquidity and user bases empowering developers to grow in a unified web of interoperable chains with the a layer the a layer is a machine for Network effect don't worry about boost wrapping liquidity just build anywhere in the a layer to tap the liquidity and users everywhere to support the aggregated future and boost quality project the polygon Community grants program was launched with 1 billion tokens All For

polygon Builders season one of the community grants program is live right now and features 35 million in madic to support the next generation of polygon projects for Builders on polygon or projects interested in migrating these grants are for you join the aggregated future and feel empowered to grow in an infinite web of unified interoperable chains apply today at polygon.net ZK well you said that everything will be a ZK roll all rollups will be a ZK rollup um we'll talk a bit more about that later that sounds interesting but I just wanted to ask what you think of current ZK rollups and you said that they you mentioned a bit that their VMS are having issues and they're overly complex but in general like how do you

view that space right now yeah I think with existing ZK rollups obviously their teams are like super talented they started like a while ago um but the existing Paradigm is you have to have this team of like you know 30 to 50 people who are all like very specialized you know experts in cryptography and then you hand code like a ZK system for the zkm so you like write this very complicated circuit it's very hard to audit um it's very hard to upgrade right like the evm is always upgrading like every six months to one year you have to keep up with that and I think the hardest part is that now we're in this world where there's like app chains and people people want to really like customize

their rollups and the problem with you know for example having this team that's built this very complicated ZK circuit is that if someone else wants to take your technology and use it as a rollup stack or framework it's very hard because like they can't modify it um say they want to add a pre-compile like good luck that's that's not really going to work uh say they want to even adjust the gas schedule it looks very complicated so I think that's why when you look at the existing landscape of rollup stacks and like Ras broadly you'll see that op stack and the arbitrum stack is really popular because like it is very easy to modify it's you know um they've they've been used a lot and so I think the yeah existing shortcomings of ZK rups today

is that you know it's really brittle and not that flexible and I think that's also even problematic for the teams because like if you know a significant portion of the best Engineers on your team have to spend all this time in overhead um maintaining the ZK stuff that's l less time that they can spend on like making the sequencer better or you know improving user experience or you know doing other things um and so that's why I think basically every rollup will become a ZK rollup because that's kind of inevitable like I think that's the technological endgame it has so many benefits for interoperability and therefore like user experience but the way that's going to happen is not going to be through the existing way of people doing zare rollups it's going to be through ZK VMS and I think like op stock

uh for example will have like a ZK option in the future uh where there's like a full ZK version um of op stack and like people can just use that and it's like write it in Rust you want to add a pre-compile okay cool you just like write it in Rust yeah I want to e Echo that um that Trend that I do think that the current optimistic rollups will become ZK Roll-Ups in the future I mean I think it's already like built like they're making it easy in their modular stock to like switch off the type of proofs that they use so yeah I agree with that um curious to know like vers maybe versus other rollups so maybe with to syncs specifically um what type of Hardware is syct using to generate the proofs um for

us today sp1 can be proven on CPU and also GPU um so we have like both CPU and GPU provs um on our prover Network right now uh we're doing GPU proving so it's just like cheaper and um more cost efficient or yeah cheaper and more cost effici and then also faster um yeah uh Curious to hear how you guys are like approaching go to market for six how do you incentivize people to build using SP sp1 um what like incentives do you have planned I think so far there's honestly been like a lot of really organic interest in building using sp1 like I think a lot of these rollup teams

really realize that um they kind of will like need ZK or will need ZK and so we've been working with a lot of rollup teams um for example polygon I think is one of the bigger Partnerships we announced recently where polygon's new protocol agair is built with sp1 uh polygon obviously has like a ton of ZK expertise um they wrote this Library called Pony 3 that's like an implementation of fry and Starks which is kind of like the underlying proof system behind sp1 and we use that Library so we've been really happy like collaborating with them in like an open source capacity but I think as they were building agare they realized you know writing custom circuits for agir in

plony 3 or something like that was going to take like a lot of time and a lot of effort and really the hard part about building something like AG is making the protocol good doing the BD for it you know making the ecosystem and yeah it's a pretty non-trivial protocol to like get all the pieces correct um and so I think they realized oh we can just write once we have the protocol for us to be able to write it in Rust and just use sp1 to generate proofs is like such a no-brainer because it's fast it's cheap and now like two Engineers on our team can spend a few weeks writing it instead of having to have like this huge team of like 10 or 20 people spending months writing it and not having it be maintainable or upgradeable um and so I

think with polygon agir like it really solved a real problem for them um and let them go to market a lot faster and have like a much smaller team working on it um and I think for a lot of other use cases we've had it's like very similar where it's like Oh either we'd have to hire like a really big team or we wouldn't be able to do this at all and so S1 solving a real problem for them and I think we're just very focused on like solving real problems for people um similarly to the rollup teams I think optimism especially has built their stack in this very like modular futureproof way um they are big Believers that like okay when the ZK Tech is ready we'll plug it in and that's kind of like why they architected

their stack in that way and so we've also been working with them on getting like SP an sp1 version of op stock um and again I do think that solves a real problem for them uh in the future of like with inter operability and um you know their longer term plans to make all these like op stock Ras chains feasible um so yeah I think overall like our go to market is really focused on solving real problems for like real customers um and it's really nice CU it's very easy on our end we don't have to do that much like all the developers just write rust so we obviously try to help people and provide them support but yeah most of the hard part isn't usually the sp1 integration it's like actually just writing your program that makes

sense I um so you guys are building with OP stack to integrate sp1 and this is like I know the arbitrum team is obviously also working on there and to do some Z um proof system I um kind of want to ask a bit maybe this isn't particularly related to like it's more about rups but how do you view that does it make more sense to I guess start off as an optimistic rollup and then move on to adopt ZK proofs or well now we're coming to a point where costs have come down drastically or do you think it makes more sense to approach it like let's get it right from the very beginning that ZK is the endgame approach I think for example with op and arbitrum they were able to like go to market a lot faster

cuz they started as optimistic rollup or I guess in op's case they didn't have fault proofs for a long time so they just kind of started as a rollup and then um they got to Market a lot faster they got a lot of tvl they built out like deps they have like users and network effects so I think that makes sense and now I think they're going to transition to ZK rollups um you know hopefully using rtech um and ZK VMS broadly so I think that like go to market totally makes sense I think now with ZK though for example say you're building a new rollup or a new rollup framework or like an app specific rollup I think it just makes sense to start with ZK from day one because the technolog is already ready uh and it's honestly like maybe even

easier to build it than having to build out this whole optimistic fall proof system like the optimistic fall proofs are actually pretty complicated like the op team's really really smart and like full of really amazing engineers and there is a reason it took them such a long time to get fall proofs working it is a very complicated system and so I think if you're an existing team that has like already brought your product to Market has existing users in network effects transitioning absolutely makes sense to ZK now but if you're a new team it makes sense to start with ZK from day one because I actually think it's like an easier experience like we provide something that's out of the box you just write rust and it just works um I I think you actually see this a lot in like the Bitcoin rollup ecosystem where

like if you're building a Bitcoin rollup today I think everyone 's doing ZK because there's like they kind of have this like technological leap frogging frogging effect where uh for example in the United States first people had PCS and then they had mobile phones but then if you look in the developing world no one really ever had a PC they just straight went straight to mobile and I think with the Bitcoin rollups it's kind of like they never did the optimistic stuff because it was like their ecosystem obviously came alive I think a lot later um they're just all doing ZK from day one and so that's I kind of Imagine that all the rollups starting now would that you know we will'll really Leap Frog and just start with ZK from day one because it is like kind of

the end game of all this stuff and it's also much simpler I think that's great and I guess it must say something like you know optimism is looking to use sp1 polygon is using it for the arer and like you said they have a team full of experts so I mean I think it speaks volumes when they're using that using tools that succin develops to help like speed up their process and go to Market faster for themselves so I mean with that in mind um what would you say the market for ZK VMS looks like in a few years is it a like a winner takes all or do you see it being like a highly competitive space maybe like an oligopoly or something like that well I like to think it'll be Winner Takes all and like we'll be the winner um I think

there's a lot of like kind of weak Network effects from a VM where for example like if we have a really rich community of like pre-compiled like that's really nice or if we have a really rich community of Integrations or you know even for example like the security story like oh this has been used in production and we like kind of start to trust it um I think for example you kind or you there's a lot of developer tooling around it I think you kind of see that with like the evm to a certain extent um and you kind of see that with like any developer tool that there's like these you know once a bunch of developers use it it kind of snowballs into more and more developers using it um and having you know really great developer experience that you can like comp compound and refine over

time um and then I think there's like also weak Network effects from like the pro Network where it's like the more proofs the pro Network gets the more efficient it can be the more provs you'll have they'll invest more in like optimizing their proof generation costs and then um you'll get this really nice flywheel where like they will have the cheapest proofs and then more people will use proofs and then you kind of get this like flywheel Network effect as well um and yeah I think that's like what we're trying to build both with sp1 and like cultivating this really rich open source developer community and then also having this Pro Network where it really drives down cost and latency and then enables more people to use like more

proofs perfect I um we touched up on this a little bit but I kind of want to get more color on how a rollup would integrate s sp1 um could you expand on that a bit yeah so if you're a rollup like what does it mean to become a ZK rollup you need a implementation of your state transition function uh in ZK uh and so I think for example we work with a few new rollup teams like Tao and Sovereign and there's some other ones where from day one they're coding their state transition function in Rust and they're already like proving it with sp1 from day one or they have that very top

of mine from day one and then you need to make your Bridge basic basically verify ZK proofs so you have your Bridge contract on ethereum and then it needs to verify a ZK proof of your state transition function and then it'll have your rollup State Route and then you can use that to like bridge in and out of ethereum so if you're a rollup team you want to integrate with sp1 you just need to implement your state transition function in asp1 generate proofs integrate some sort of like sequencer to like call to the pro Network get the proofs and then submit them on chain and then also you know modify your solidity smart contracts to accept S1 proofs and um yeah get use that as a source of truth of what your rollup is uh on

ethereum um have have the pre-compiled been audited in um for bis sync or anybody else yeah our whole system is audited uh we've done a couple audits with two really top tier ZK teams um that have audited like scroll and axium and like a bunch of other rollups before um and that definitely includes the pre- compiles um like we care a lot about security and obviously if you're using pre- compiles it's very important that they're correct and then we've also um hosted a public audit competition uh with Cantina which is like run by spearit so this was like a public competition anyone could participate uh

I think there were like you know dozens if not more participants to like look at the code and find you know any potential vulnerabilities um yeah there was actually a lot of really smart people across the world that participated it was super awesome to see um and like super inspiring that like totally random people um studied our system understood it surprisingly well uh it is pretty complex and like they um yeah had a really great understanding um and so yeah those are that's kind of like the audit security um overview of sp1 and does it does it make it easier too because it's in rust or is that just another layer of of complexity added oh to audit it yeah um I think like the

like so rust is a memory safe language um I think mostly for auditing though that doesn't matter too much you're mostly auditing like the circuits and like making sure it's constrained properly uh so the circuits can be expressed in like whatever language really um I think it helps it's in like plon our circuits are expressed with like Pony 3s system for you know constraining those things and that's kind of like standardized so I think that helps a little bit um but yeah I don't think they're rust helps that much with auditing okay makes sense what do the sixin end game look like what is your like apart from sp1 and the prer network what what else are

you guys focused on what else is coming up I think for us like we're just really laser focused on getting sb1 to be like as cheap and as fast as possible getting it integrated into every rollup stocks or really making that vision of like every rollups a ZK rollup with S1 come true um like every application in crypto that wants to use DK uses sp1 um and then yeah building out this Pro Network to make the costs and latency of proving like as cheap as possible and also decentralized um so I think already that's like a lot of work um and we're really focused on that um I think something I'm excited about in like the longer term though is that ZK I think has a lot of applications even outside

of just crypto and blockchain so rollups is just like the shortterm one to twoyear plan I think on the three to five year time Horizon you could imagine a lot of use cases outside of crypto even in web to so for example today you know you have these like Credit Union companies that are like issuing you your credit score and right now like they have basically no accountability right where they can just like modify people's credit scores with without you know any recourse and it'd be really nice if there was a aspect of verifiability in that where you know that they're applying updates um according to like some set of like known rules that we've all agreed upon that are like transparent and verifiable so you can imagine like this idea of like

verifiable servers just beyond rollups being like really important uh for a lot of companies um that are you know really trusted today um there's like a tested image editing where like I think we're increasingly in this world where it's like really hard to tell like is an image coming from a real thing a real camera or is it just coming from like mid journey and so you could imagine that in the future you have like all these cameras basically signed that hey came from real camera and then you do these like verifiable Transformations on top of it so I think there's a lot of like really interesting use cases Beyond Roll-Ups in crypto that I'm really excited about in like the longer term um and yeah excited to see like come about

that that Vision where you know it it's used less outside of the blockchain network more for those kind of use cases you described does that would that require its own network or are we talking like client side proof generation some of that might so for example for the verifiable server they could just use the pro Network like they don't really require privacy they just want verifiability you can just use the pro Network for privacy use cases um that would require some amount of client side proof generation for tested image editing there's not really privacy it's just like verifiable transformation so that doesn't really require um clide proving but then yeah there's also a bunch of other interesting

privacy focused applications that have client side proving as well you said client side proving for privacy how how would that work would it would you need like proof aggregation or like how exactly would that work yeah I think basically you would generate a proof um you know in your web browser on your computer or in a mobile app and then that proof like wouldn't be that succinct and then maybe but it would you know hide things and make them private and then you could send it to the Prov Network to shrink it and make it more succinct I kind of Imagine That being how like client side proving would play out is you basically like do a really fast proof that's not that

succinct because there's in general this trade-off between Pro time and like um how it's distinct and like short it is so you you do a private proof very fast not that compressed and then you send the pro Network to compress it and make it really small um yeah that's um so it's kind of like wrapping proofs you're like wrapping one of it one is quick to verify one is quick to generate and you're getting the best of both RS basically yeah yeah yeah exactly okay gotcha um before we wrap it up one where do you want to Point our listeners towards how can they find out more about six synct or I guess like is there anything they should

specifically be focusing on yeah well if you're like excited to BU with ZK you can go to our docs uh it's docs. sn. XYZ you can go to our website uh and then yeah you can follow us on Twitter at succinct labs and follow me I'm at Puma the Uma um and yeah I think those are all pretty good resources um we have some blog posts talking about like sp1 why it's awesome um and yeah like it's very easy to try out like on purpose so yeah I always encourage people to try it out and like see it um and see kind of the magic I think especially if you've worked with ZK before it's really like night and day like the I I mean as someone myself who used to do that stuff

I think there's been so many moments where I was like oh my God I can't believe this is possible like I just import a library I import like the Sur library and rust or the Json library or revm or reth and it just works like wow my God this this would have used to taken me months and now I do it in like an hour um and so yeah for ZK people I always encourage them to just try it out and see for themselves because there's like that is like Irreplaceable feeling of like being like Oh my God this is like really a new paradigm it's like such a big shift that sounds awesome um I'll make sure to include those links thank you so much for coming on um yeah for sure thanks for having me are you currently building on polygon or interested in

migrating season one of the polygon Community grants program is live right now and features 35 million in madic to support the next generation of polygon projects builders on polygon these grants are for you join the aggregated future today by applying at polygon.net [Music]