Every Rollup Will Be a ZK Rollup | Uma Roy - Succinct

hi welcome back we have our next guest speaker Uma hello um my name is um I'm the co-founder and CEO of succinct and today I'm going to be talking about how every rollup is going to be a ZK rollup so I think something a lot of that gets talked about a lot in the ethereum community is the notion of rollups are the end game and maybe a lot of people agree with that but hopefully the point of this presentation is to convince you that not only are ZK rollups are the end game but are actually here today right now with what we're building as so today most of the rollups that are deployed in production are optimistic rollups think rollups like arbitrum base optimism main net and the reason that is the case is that when those rollup teams were working a few few years ago to deploy rollup technology ZK was still hard uh it was very difficult it required a lot of cryptography and so optimistic Roll-Ups were the fast go to market option but optimistic rollups actually have a lot of hidden costs uh the long the obvious one is a long time to finality where you have to wait 7 days to withdraw your funds and that's really bad because it locks up Capital it degrades user experience and it makes interoperability between rollups really hard ALS also these optimistic rollups have very complicated Contraptions for their interactive fault prooof game uh which has this challenge period and you know interactive challenges that are economically incentivized and in the end the systems end up being very very complex and actually hard to reason about so as I was mentioning why are most of the top rollups by tvl optimistic rollups today that was because back a few years ago launching a ZK up was really hard developers needed to implement cryptography and custom ZK circuits to make things like a ZK evm and proof generation was really really expensive uh so it would make the end user experience bad because transactions would cost a lot because you had to generate a proof for them and finally optimistic rollup teams went to Market early they had a hard time switching to ZK even if those things were now different and ZK proving costs have gone down a lot because there's a lot of sunk cost in their existing stack at cinct we're building a ZK VM called sp1 where you can just use ZK by writing normal rust code so s1's kind of a breakthrough in the ZK Paradigm where you no longer have to code custom circuits or do any complex cryptography you can make a ZK evm by just writing normal rust code like taking things like w or rev and getting them to uh in sp1 and getting out ZK proofs so companies now can just make a ZK evm by using sp1 and then they can use the syn pro Network to generate proofs very cheaply and cost effectively lower the mic just a little because it's covering your face oh cool okay and um finally now that ZK is really easy and very cheap um existing rollup teams can just take their existing rollup stack which is written in normal code stick it in sp1 and get a fully working ZK rollup n so we actually built this out for the op stack it's our uh integration called op cinct and I'll go more into this later in my talk about how it actually works and kind of who's using it so let's dive a little more into the performance of ZK VMS and how there's kind of this mors law where ZK is getting cheaper and cheaper and faster and faster every year we launched our ZK VM sp1 around one year ago and back then we were able to prove around 60,000 cycles per second and those are risk five Cycles which is the instruction set that we prove since then over the past year we've gone from being able to prove 60,000 instructions per second to 4 million 4.8 million to be exact instructions per second so that is a hugely dramatic increase in our throughput which also translates to way cheaper proving and way faster proving and we're also building building this prover Network which is basically a hosted proving solution for sp1 so sp1 is a piece of Open Source software you can use it you can run it on your laptop and generate ZK proofs but in practice to make things cheap and fast you need to run ZK on ZK proving on gpus or other custom Hardware to make it very efficient and so approver network is this solution where developers can send a program and an input via an API to request approve uh provs and gpus all across the world basic generate the proof and send it back to you and get paid and it becomes this protocol where everyone can get ZK for as cheap and fast as possible and that's what we're building so with sp1 and the prover Network I like to say that we're in a new era of ZK which I like to call ZK 2.0 where now anyone can use ZK in a much more modular way in ZK 1.0 which was kind of the old way of doing ZK where you get ZK rollup teams like scroll and ZK sync and polygon each of those teams would have to build all parts of the stack so they would have to build their own rollup implementation their own sequencer node software block explor Etc and then they would also have to build their own proof system and they'd have to build a bunch of complex cryptography and circuits for a ZK evm and they'd also have to build their own prover and host all that in for themselves in all it was a huge lift and that's why all these DK rollup teams took a really long time to go to market and were many years behind the optimistic role rollup teams with ZK 2.0 what happens is now everyone can use sp1 or ZK VM they just write normal code and then they can use our prover Network so that they don't have to have their own proving infra so you can just tap into ZK but with normal code and then you can build whatever you want you can build a ZK evm you can build a zks svm you can build whatever VM you want whatever state transition function you want really whatever software you want to prove is finally provable and this dramatically makes using ZK easier the go to market timelines way faster and also expands the types of things that we can ZK proof and so our hope is kind of in the ZK 2.

0 era ZK is finally easy fast cheap and can go get adopted everywhere in the stack and will allow every rollup to be a ZK rollup so hopefully I've convinced you at this point that you know with S1 or ZK VM and the Prov Network ZK can finally live up to its potential and now I'll talk a little more specifically about what we're doing to bring ZK to every rollup with OP cinc as I mentioned earlier so op synct uh takes the op stack which is a very popular rollup framework and sticks it inside sp1 to get a full ZK rollup which we call op C synct and in particular there's a lot of advantages you have 1 hour fast finality you can prove everything and it's pretty fast and reliable um with cheap costs however we also have another offering called Opie succinct light uh which is perfect for rollups that are more cost sensitive and this replaces the interactive fallproof games that are used by a lot of current optimistic rollup teams and replaces it with a single shot fault proof where instead of having this interactive back and forth dispute uh in the case of a challenge you can just verify a single ZK proof and here the kind of trade-off is that you get much cheaper costs cuz you aren't proving every single trans action you're only proving something in the unhappy path where something goes wrong um but you still get slower finality than the full ZK option so maybe around a day because you have to still allow for time to generate the challenge proof and submit it but you improve upon the 7-Day finality that most optimistic systems have so Opie synct is kind of a good middle ground between full ZK and the interactive fault proofs so in general with Opie syin light I think one thing that pretty interesting to note is that I believe it's a strict Improvement among beyond the interactive fall proofs we have today today is interactive fall proofs like Optimus and arbitrum are using in production are very complex multi-round economic disputes with a ZK fall proof you can just resolve your dispute in a single step by verifying a proof this means that not only are your finality Windows shortened from 7 days to 1 day it also means that your bond sizes that you have to put down for this economic game go from being hundreds of eth to tens of eth which is much more efficient and so you get lower Bond sizes faster finality and just an overall much simpler code base which is really awesome so we think that ZK fraud proofs are future proof and kind of the endgame direction of interactive fault proof games and then op the sync which is kind of full ZK where every transaction is ZK proving gets you even better uh you get really really fast finality so you can have finality in 1 hour one day maybe even tens of minutes you get cheap cost effective approving where the cost of proving per transaction ranges between a 100th of a cent to a tenth of a cent so that's very very affordable especially compared to ethereum Da costs and then finally because your state transition functions written in normal code and our ZK VM lets you use ZK with normal code if you want to add a new pre-compile or you want to customize your rollup that's no big deal you can just do that you can write the code stick an sp1 and everything will still work so we think with both op succinct and op succinct light every single op stack rollup kind of has a path to this DK future so if you're a large L2 with strict security and interop needs or a defi chain that has a lot of high value transactions where you don't care about one1 of ascent here or 100th of ascent there op sying a perfect solution you can become a fully zkf chain if you're more cost sensitive L2 with thin margins like a gaming chain or something where there isn't a lot of financialized activity then op distinct light with ZK fall proofs is a great solution for you and in our opinion a strict Improvement uh to the interactive fallproof games that we have today and I have an image here of a electrical engine which is dramatically simplified compared to the complex gas engine that I kind of showed earlier to me the gas engine represents interactive fall proofs they're complicated they're clunky there's lot of steps going on it's hard to reason about obesi synct light makes ZK fault proofs very streamlined and very friendly and we think that's the future that all rollup teams should take and I guess um there is becoming more and more broad agreement on this ZK rollup future where every rollup is going to be a ZK rollup and vitalic says he was even thinking about that in 2021 and uh the main concern is still approving costs but especially with that ZK Moors law chart that I showed you earlier in this presentation hopefully you're convinced that proving costs are an exponential trajectory to getting better and better and actually we even have a few op sisting chains running live today and in practice the cost per transaction is around a 10th of percent or 100th of percent so it's at the level of existing l2s already and so I think the time is now for chains to start adopting either full ZK or at the very least ZK fall proofs to kind of go towards this future that we see for ethereum rollups and kind of as evidence of what I was saying of people already using this technology we actually have a bunch of customers for both offerings that are using this today so we recently announced that we're working with mantle which is a ZK um which is an L2 with a lot of tvl around 2 billion in tvl one of the largest l2s and they want to convert their chain to be fully ZK with op sucin and we have another chain called Falla that does te e AI agent experiments on their chain that's using op Sy synct in production right now in addition to that we have a bunch of other customers that are going to going to use op Su synct light in the upcoming months as they think ZK fault proofs are strict Improvement on the security they have today and there's a ton of infrastructure Partners across Ras and Alda that are also supporting both op synct and op synct light so In Sum hopefully this presentation convinces you not only that is ZK is the endgame which is a common phrase heard throughout the ecosystem but also that ZK is here right now it's easy it's cheap it's fast with Opa synct and people are adopting it today and we should be moving towards that future to have a better more interoperable ethereum thank you thank you so much it was a pleasure to listen to hear you speak so