How Electron Cuts ZK Proving Costs by 95% with Garvit Goel

Transcript

if you look at the last three years of onchain data on ethereum protocols spent roughly $60 million in verifying their proofs on ethereum our protocol reduces this cost by 95% so rather than collectively the whole ecosystem paying $10 billion every year they'll be paying like 500 million for all the rollups combined we have created a universal aggregation layer where all proving systems and all ecosystems get combined into one proof [Music] I'm here with garve go who is CEO of

electron electron was on the show about seven or eight months ago back in April and we discussed electron as a ZK proof aggregator it is a platform that allows uh protocols to reduce the cost of ZK proving by a significant amount and since then plus has been released they now have a test net and our on road to main net and I want to get Garvey on to uh give us an update on electron what's been happening and what's the future for the protocol so thanks for coming back on thanks for having me Seb it's great to be back here and there sure are a ton of updates uh there's a lot going on

here at electron and uh have happy to go deeper today with you on them so before we get started for those who didn't listen to the other to to the previous episode like they should go back and listen to that but you know give us a high level of What electron is building and why it's important sure so at electron our goal is to reduce the cost of verifying proofs on ethereum and other major El ons like Bitcoin and all and the problem statement somewhat is that right now the cost of submitting and verifying your ZK proofs on ethereum and Bitcoin is extremely high uh running into hundreds of millions a year if you plan to do it consistently as most apps and rollups

plan to do so we reduce that cost by like multiple orders of magnitude like up to 95% but what is it about like proving on ethereum or Bitcoin that makes it so expensive to verify those proofs so uh it's not that the verification is expensive it's that the block space is so limited for these l1s that if you do anything consistently again and again over time it becomes really expensive so verifying a proof on your phone would be actually pretty cheap it takes a couple of milliseconds to do it on a laptop but when you start uh occupying block space to do that it just becomes terribly expensive uh very

quickly and yeah we want to put those proofs in uh blocks so that other protocol can come and check them I mean that's the whole point right the point is that you put a proof on a blockchain that way another protocol can come and check and it performs on action based on that proof like the certainty that tokens have been deposited in a contract or uh that like some like client uh is providing a proof that like some action happened on some other chain so that you can there there therefore perform some action based on the proof precisely it's just how you settle a need up or a z bridge on the table yeah cool so yeah and just to kind get an idea of the costs here like how much

does it cost to verify uh a proof on ethereum for instance and what are some of the benefits or like the the the the cost improvements that we see using electron so uh Bas on current gas prices it's roughly $30 to verify one ZK proof uh but if you a roll up or any other ZK product for that matter you will be submitting these proofs ideally every etherum block you want to settle every ethereum block so with every new block you send a ZK proof so which means you some paying $30 in gas fees every ethereum block which is 15 seconds which on an annualized basis turns out to be I think roughly $100 million if I'm not

wrong and if you look at the last three years of onchain data on ethereum protocols spent roughly $60 million in verifying their proofs on ethereum if you look at the ethereum scaling road map we want every road up to a ZK rollup settling on ethereum so if you occupy most of the ethereum block space for ZK proof verification like 60% or something like that then this cost will be like $1 billion every year just for proof verification by these protocols uh and uh so that's the problem we aim to solve and our protocol reduces this cost by

95% so rather than collectively the whole ecosystem paying $10 billion every year they'll be paying like 500 million for all the rups combined this is not per rollup all the rollups all the ZK applications collectively will only have to pay 500 million uh which is massive savings which we can then put back in the hands of the users uh and the protocols yeah that that brings up a a great point that I've been thinking about a lot lately you know there's this great chart that people have been circulating on Twitter of like how much base generates in revenues and how much of that is paid back to ethereum and I believe what Bas pays ethereum uh on an annualized basis is something like

$1,000 or some ridiculously small small amount um you know right you know ethereum is is generating revenue from CK proof air ification if we're reducing that Revenue like by significant amount sort of looking at the broad scale of uh kind of revenues that that ethereum is generating uh if those revenues are going down then like how does ethereum continue to be profitable I I I think I'm not sure why base is beinging so low but it as far as the ZK rollups go I can say that they are paying a lot more so

I don't remember the exact figure of hand but uh I think scroll has paid double like a you know I think 20 to 30 million on an analyzed basis in verifying their proofs so obviously in their case they have users who can basically pay for this cost so they don't have to pay it out of the pocket but uh I think even base as they try to submit fraud proofs on ethereum frequently even their cost will go up it's possible that uh being an optimistic Ro they've chosen not to submit these fla proofs on ethereum yeah so let's talk about uh Quantum which is the network that you guys launched in testnet uh in the last couple of months what is quantum and

how's it being used right now sure so Quantum uh is our Flagship product it's the it's the core aggregation layer that we have for ethereum and Quantum has been live for the past 140 days 145 days on testnet aggregating proofs from some of the top ZK products and protocols including risk zero World coin scroll igda and we have been aggregating these proofs and submitting them on the ethereum test net these are real proofs from these protocols that they submitted on mainnet before so these are not dummy proofs that we' have created or in a in a fashion that is optimized for aggregation these are real life proofs

uh that we've picked up from these protocols mainnet and Ram that through our test Network for the last 150 days we've been doing that and uh every time they submit approve uh which is an aggregate proof for all of these individual protocols we save a little bit of fees and uh if you create the sum total of the fees that we have saved so far that uh yesterday hit $1 million in the last couple of months of operations which was some of the one of the cheapest uh duration like cheapest period for proof verification on ethereum uh it it usually is a lot more expensive than this and uh yeah so for Quantum we are uh as I said we have a stable test net now and we just looking

for audits very soon and looking to Feit meet like in January or February next year cool so there's a couple things here that I want to dig into and one is the different proving systems that you guys aggregate we talked about this on the last uh episode we did together you know you sort of educated me on the different types of pring systems so we have like plunky 2 RIS zero uh Groth 16 Halo um what what is the core difference between these different proving systems and and how difficult is it for you guys to aggregate all these different types of proving systems into a single proof that can be verified on ethereum yeah we spent a lot of time on this actually and it's one of the core

problems that we have solved at electronic so like to answer your first question these two systems are at an abstract level they are all the same okay if I were to like create a on line definition it would be the same for all of them but the underlying mathematics and how they achieve it is totally different and uh it's so different that there's hard to find any common uh nuances across these maybe one or two very simple themes which you see common but mostly they're very very different coming from very different uh historical reasoning and backgrounds to arrive at those proof systems so that's super super different and that's why super

incompatible to aggregate with each other so we spent like a better part of 4 months just creating a recursion Engine That Could support these multiple different systems in the same aggregation layer and so you know when we meet people most people just assume that okay we must be creating one aggregation layer for each proving system which in on Main net must operate independent of each other people just resume that but it's not true uh we have created a universal aggregation layer where all proving systems and all ecosystems get combined into one proof which is very important for cost

reduction optimization so yeah it was a long uh uh ZK exercise which allowed us to come up with this system which does Universal appication are there other proving systems than the ones that are most commonly known in say the blockchain space like the ones we we mentioned earlier that are used maybe in like different Industries different types of applications uh that uh that are still left to be uh added in into um electron's core infrastructure yeah so I think coming up with a new proving system is not hard you can just create date one create some variants of that and launch launch

another one okay and so you can do that and I'm sure there are more than 25 different proving systems out there currently we support a proving systems on our aggregation lay but what is more interesting is that see we understand that people will continue to launch more and more proving systems for the next decade at least okay so we cannot be just tie to one set of proving systems so the key Innovation that we made sure is that adding a new proving system to Quantum is super easy it takes less than a day of development effort to add a new proving system so let's say you let's say you decide to ship Bas 10.0 2 years from now okay which is like a some

amazing proving system it will take less than a day to add that no new audits will be required at your end no new audits will be required at R end and that's it in half a day that you proving system goes live on mainnet and can start start aggregating proofs from your from your ecosystem so uh that's the way we've designed it and uh it was so the ease of addition of a new proving system was one of the key parameters we optimized for as we designed this aggregation layer and uh we are happy that we were able to achieve this quick integration is this the first time that this has been achieved aggregating different proof systems into a single

proof absolutely absolutely there is nothing of the sort out there right now okay like you know like uh um like are you up for a turning reward or something or the see I mean this seems pretty insane I mean like I'm I'm I'm I'm not great at uh I'm not I'm not the greatest math guy but this seems like a pretty uh massive achiev yeah I mean I would like so first of all if an award is deserved it's by the engineering team so my my head was very limited uh uh in applied in a limited capacity here but I want to add like uh this Innovation was uh we didn't actually

have to go into that kind of R&D to achieve this uh so for example we started with a certain proving system and it's easy to go in another proving system using that and then apparently it's possible to go from that proving system to another in an easy way so you can create these uh pipeline and Pathways to actually achieve uh a final aggregate proof which can support aggregation of all of all a lot of them and people had written these recursion sockets even before we came along uh of how you can go from one proving system to the other right so yeah I think it was just clever selection of code a clever selection of uh of engineering

road map to achieve that yeah okay cool cool um so like coming back to Quantum like practically speaking when a profile wants to reduce its uh its proving costs or is verification costs what's the process for plugging in Quantum so we have a straightforward integration process in fact our team is happy to provide that support as well it takes usually less than half a day to achieve it and the process is something like this so without electron in the status quo you are sending your proof once you generate it you're sending your proof directly to ethereum where it is accepted by a smart contract and it is

verified and the rest of your business logic continues now what happens in the in the electron model is rather than broadcasting your proof directly to ethereum you send it to our aggregation Network and then we collect a bunch of other proofs from similar projects like you and then we create this aggregate super proof which we then broadcast to ethereum and then once our contracts verify this super proof on ethereum your application contracts on the Chain can query our super proof verification contracts to check for an inclusion proof and that's it then you go about doing your business just like

you were doing earlier okay so you're yeah and instead of sending the proof directly to ethereum to uh to be verified you're sending it to electron and then electron Aggregates that proof sends it to to the chain and then you can utilize the verification contract to uh electron verification contract to check that proof and so integrates that uh into your application yeah when you say like electron netor just to understand here is this happening uh offchain is there like actually a network of nodes that are doing this proof aggregation like what's the kind of censorship resistance and properties of of of uh a Quantum um yeah yeah when I say aggregation

aggregation is basically ZK proving it's a specialization of of that and so that's why just like any other ZK Network we also run the set of ZK Prov off chain and that's why just like any other ZK company we also need to decentralize our Prov Network we are working with several folks like galot is a partnership we've announced uh we are working with a bunch of other uh providers as well who will help us decentralizing approving Network on that front we look similar to any other uh uh ZK Network there's one key difference though uh which is like for rollups they have to have the sequencing requirement uh so that's why this whole debate on centralized and UNC centralized

sequencers and the reason that they have to do that is because they have this concept of a state uh the good part about agregation is we don't have any state so we we not required to do any sequencing so we don't have that problem in terms of uh a centralized sequence or not uh as of today we are just on testnet so we are operating a single aggregation machine ourselves inh house uh running on cloud uh but of course uh as we gear towards main net we'll be decentralizing this folks like gavot FMA they will be invited to run these nodes and uh at that point I think the censorship resistance also kicks in because because if let's say we decide

to not provide the service there will be some other node provider who will be having a desire to generate revenue and they'll just provide the service to you so you can use one of them on top of that I want to add that let's say electron the electron po Network as a whole decides not to provide service to a particular uh ZK project they can just choose to send their own proof on ethereum directly without using the aggregation they can always do that the cost will be higher so in that sense aggregation layers don't have that censorship Powers anyways uh because the alternative is just going to etherum directly there may be some use cases which are dependent on low costs to

achieve this at which point the decentralized PO Network becomes important yeah right I mean you you know that's a good point you there is no censorship from the perspective uh being able to send your proof directly to ethereum there there is everever from the perspective of being able to Aggregate and benefit from the lower cost um you mentioned on the last episode we did together that uh Pro generating a proof was a pretty computationally trivial task to do then in fact like you could do that on a mobile phone uh is there a future in which like people could participate in the Pro Network by uh like you know one of our other portfolio companies bless

are doing this uh decentralized computation Network where you just run a Chrome extension and you can run computations for this network you know is there a future in which anybody could run um some software on their computer on their phones and participate to proof generation further uh extending and expanding the kind decentralized nature of the Prov network is that something that is desirable or maybe not like what are the kind of trade-offs there so first of all uh yes I mean that's the goal that uh we want to get red the proof generation infrastructure requirements as much as possible having said that like uh I personally at least believe and I'm sure a lot of ZK people

will disagree with me on this I don't see the requirement why uh you know like a normal individual should generate uh proofs for somebody else I just don't see the point it's like decentralized storage maybe uh which never took off because you can always use uh cloud services or drive or whatever just to s s whatever you want is it's cheap enough it's very very easily accessible so even in future cloud machines and Cloud infrastructure will always be easily accessible so I don't see any reason why projects would want to offload that to the client side uh this will obviously not be true for privacy use cases where you want to withhold your data as much

as possible so generating the proof on the customer site becomes important and uh but again they'll be only generating proofs for their own use case not for anybody else which will become feasible uh in the next couple of years I believe u yeah so what do you think is the value then in like decentralized compute networks like what what are the some of use cases that you see there that are maybe more interesting than proof generation yeah I I think mostly it's uh ensuring uptime in censorship resistance I think that's the key value that we have that that at least we we are consuming from our uh providers there's another value ad here which is

aggregation of Hardware so since crewing is expensive these companies can do it for a bunch of projects thereby ensuring High utilization of the hardware uh so that's an additional value proposition on top of that yeah uh I see it's just like I think validator operations like if you look I think these ZK proving companies uh Pro networks will uh essentially start to look similar to validator node companies that we have in the proof of stake ecosystem right so you know in the case of um in the case of quantum I believe right now there's something like

nine uh networks or nine protocols being aggregated or something like that um so we use a veil scroll world coin Loop ring dgate real gun uh ZK P2P and and maybe a few other ones you know what like I guess the the the thing you want to optimize for is I mean you want to have as many networks and as many proofs being sent to uh electron as possible so that that further reduces the cost per proof because if you're if you have like a fixed cost for proving on ethereum and there are more uh proofs being added to that super proof that's going to just like reduce the cost of proving you as an individual protocol um so like

getting to that uh getting to that like Optimal number of protocols is is kind of like the end game um but while that's happening and while like there is just like a smaller number because this is so new of protocols using um electron or or Quantum in this case you know that that cost is going to be higher than where it could be potentially but then there's a question of like okay how many times per hour what's the ideal kind of time frame for sending your proof to ethereum um yeah how do you balance that out to like kind of Ensure best time for Price uh like that optimal

range yeah so this is the uh biggest question we get from our customers of and this is the most interesting question so there are several aspects to this and I'll break it down uh so first of all this is like a scheduling problem okay there are so many people submitting proofs on ethereum with different time intervals uh they have different uh monthly budgets for verifying these proofs and some of some of these proof schedules keep changing as well that you start with once a day and then you increase it to once every hour then you reduce it to once every week so these things keep happening now how do you so how do we solve for this so now what's going on

what we noticed on on the main net is the cost of verifying is so high that most projects just cannot afford to send proofs more than once every 30 minutes right we look at the onchain data and most projects are submitting it with this kind of an interval now once we start aggregating these proofs and reducing the cost these protocols can significantly increase the frequency with which they are submitting these proofs so for the same monthly budget you can now settle on ethereum let's say once every minute as opposed to once every 30 minutes so when when we hit the market

what we are able to tell all these customers is that we will increase your frequency of proof submission no matter what okay so that allows us to have that flexibility uh in terms of onboarding customers and solving for this problem because we can just come go to a customer and say that dude no matter what your monthly budget is keep your monthly budget the same just increase the proof submission frequency and we will do that for you so it becomes a clear value proposition without any trade-off for these customers uh which makes it very easy to onboard the folks for this service so what what is like right now kind of

the yeah how often are are super proofs getting generated like how many proofs are in that super proof and you know what's the cost uh per per to get included in the super group yeah so right now we submit a batch every 15 minutes uh sorry every 31 minutes uh we submit one batch every 31 minutes and so far we have aggregated more than 100,000 proofs so far uh which means uh and since we have uh eight to nine proofs in every batch this translates into a roughly 88% cost reduction uh on the test net yep these

are the SS right now yeah and people can go to uh they can go to export. electron.and they can see all these stats and it's basically you know it says here like total gas costs saved has been over $1 million um 110,000 proofs 112 th000 proofs have been submitted and you can see like literally you can see the proofs getting submitted uh the protocols that are um that are participating at at the moment so we have I da Avil scroll turn cach we mentioned those a while ago um and also the proving systems that they're using so like I da is using risk zero Nel is using plunk um uh cool so like yeah let's talk a little bit about so the

different um protocols that are that are using Quantum um and and there are kind of various use cases like how what what what's the gamut of of use cases that you guys are seeing I mean in terms of use cases we are seeing all of them ZK BMS score processor ZK Bridges privacy protocols uh I think if you've as I said if you're a ZK project you are a potential customer for us so the whole of the ZK ecosystem uh we are seeing like a subset of that on our uh in our customer Pipeline and uh I think the biggest customer we have right now is the latest one is risk zero uh so okay so to back

uh back track a bit uh right now what we are seeing is there's a transition going on where all the people who had shipped ZK circuit and custom circuit based architectures they're now switching to ZK VM and they're selecting anything between risk Zer sp1 value uh you know Lita Foundation Nexus all of these folks and we are seeing an increasing demand come to us from the ZK VMS so uh we've been approached by pretty much all zq VMS so far uh risk zero is obviously announced and couple of others uh are have already been signed or are they are in the process of being integration or getting signed with us and we are seeing an increasing demand come from these

particular segments so as these zvm take their own customers on mainnet uh this question of high proof verification becomes very important very suddenly and uh so earlier what some of the ZK VMS uh were expecting to do is they would do their own aggregation inhouse okay and uh because they thought that doing aggregation is just a ZK problem where you you just combine a bunch of proofs it's it's done but as we suspected uh many months ago and now ZK VMS are realizing is there's a bunch of there's a whole network that you have to put together and a bunch of services that you have to put around the aggregation piece in order to be able to fully

uh leverage the cost reduction without compromising on the proof submission schedules of their customers so these ekms are coming to us they want uh us to become their defao provider potentially become enshrined in the zkm itself and just take care of the aggregation pie so electron becomes the default aggregation layer for a lot of these zvm uh this is something we'll already seeing so that's like the biggest most interesting segment for us there are a bunch of folks which are already on Main net like zit for example is one of those uh and uh so if so obviously they would either have to go rewrite their entire

architecture move away from circuits and go to ZK VMS but they can just start using us today uh because again we support their approving system there are some like uh interesting use cases as well so for example igon da is uh they don't submit ZK proofs on ethereum but they do something called as BLS verification and ZK proof verification and BLS verification have a big component which is same for both the parties so we can actually support aggregation of that as well and we have that supported on the test net today uh that's the set of customers we are either working with or plan to work with in the near future are there other

companies doing Z Super proofs right now like with any significant traction uh there were a few that started out same time as us there's Nea of course that was doing it uh but I don't think so they have uh any tested volume anywhere close to us or the number of partners that we've secured uh polygon agler is obviously there but again polygon is very close to us we work very closely with them most most of the founders are investors in our company as well and uh so it's quite likely that polygon Adler and electron will be kind of working together to make sure that yes that together we aggregate most of the proofs in the market I mean it's just like it feels like a winner take all kind of

position to be in because the more customers you have the the cheaper the proofs and then there therefore like it becomes much harder for any other uh participant to come into the network and try to uh reduce the cost of proving enough to compete with the one the competitor that already has like the majority uh of the stakeholders in the ecosystem and kind of like the lowest proving system it's like are the economics here am I getting these right or like is there something I'm missing yes absolutely so that's the biggest mode uh so aggregation is actually not a ZK problem it's a business problem you want to onboard as many customers as possible as quickly and that's your Mo not the technology that is commoditized

already having said that this is where the biggest challenge for an aggregational layer also emerges because what we expected and what we are seeing now is a lot of these ecosystems they are competing with each other and once we start aggregating proofs from one ecosystem the other ecosystem uh has a lot of questions about how that aggregation is going on before they're on board with us so some of the architecture changes and decisions that we're making now uh are actually driven by that that people have an architecture which does not favor one party over the other and we've seen this increasingly

and I think this is the part that uh me and the B team are having to work through is to keep the incentives align not make sure that no one no particular part is getting any favorable treatment because that just puts other people off so yeah it's a huge political uh uh narrative game you're playing and we have to tread very carefully not to step step on any toes over here yeah I I find that surprising because from from my perspective it it seems like what you guys are doing is so uh like it's sort of it's agnostic right it's just basically you know anybody who's proving

to a the who's verifying on ethereum um should see that alignment or like shouldn't see like an alignment problem um unless you're verifying on some other network so it's not a problem obviously we we super ethereum align and anyone who's building on ethereum sees that but these rollups are also competing with each other ZK VMS are also competing with each so even they're all based on ethereum still competing with each other which is where we have to be careful uh yeah and in fact like support in fact we don't have any problem going across ecosystems because if you want to do this on let's say Bitcoin as well now Bitcoin aggregation layer will be

different than the ethereum aggregation layer because Bitcoin has different set of load ups and again they have to submit the proof on bitcoin not ethereum so supporting different l1s is actually not a business or political problem it's the same ecosystem where the uh where you see this kind of problem uh yeah yeah interesting um are you you you probably about like Bitcoin OS and how they did the first CK proof verification on ethereum or sorry on bitcoin will you guys support uh Bitcoin at some point we we plan to uh in fact we are in touch with Bitcoin Usos or as we speak and a couple of other ZK rollups building on bitcoin uh the great thing is they are

using two wi systems similar to the ones being used on ethereum so the Z part of it should carry over quite easily the set of customers will obviously change and uh the economics will change because ether Bitcoin has different pricing model as opposed to ethereum uh but it's a space we are watching very closely and uh we plan to support it as soon as we see that ecosystem getting close to an inflection point yeah we we just did an episode on uh on epicenter uh recently with um with Yago from from Bitcoin OS and there's it's really great to understand so how ZK proving works over on Bitcoin compared to ethereum so taking a step

back here what are some of the novel use cases that you're seeing and that people should look out for in the next couple years with ZK the one that I am Poly on the lookout for is so we obviously saw the stable coin ecosystem establish itself as a clear use case now we have so much liquidity on chain so I think the next step should be in my opinion the tokenization of other assets which are also money which is uh let's say stocks S&P 500 and uh some of let's say Tesla gets tokenized or something like that and so this is uh this as an idea this has been explored

in the past as well okay and even I worked on in my previous startup we were trying to tokenize real world assets like trying to push uh Asian Securities and tokenize them the problem we saw back then was that there was such so many high heal assets present on chain that you could just buy and get like a 200% 300% 1,000% return or get like a very high apy like a th% apy on your stable coins or whatever on chain so it was very difficult for these kind of assets to compete but what we're seeing now is most of those assets which were supposed to be safe and high yielding are gone uh we have mean coins which are super high risk super high return but

people understand that like if you are a company with significant assets on chain you're not going to invest your treasury in mem coins but you would potentially put some of it in t bills maybe put a small percentages and S&P 500 so I think the market for that would open up now uh and the regulatory side has also cleared at least as far as I I'm not an American so probably people over there could commend better but as far as I understand it has become from a regulatory standpoint things are going to improve going forward so it would possibly become easier to toonize these assets and I'm watching that space very closely uh because I I see a lot of uh companies from emerging markets for example from India Singapore Brazil

Argentina that would want to access this onchain uh Market essentially to uh let's say launch a public issue or something like that I'm personally very excited about that and I hope that becomes the next big use case we see out of crypto and ZK in general yeah cool and uh what's what's future plans for for electron and Quantum and when main that when token so we are actually gearing up for audits as we speak uh we have already intimated some of the Auditors we in touch with that okay audits are coming so they should share their price quotes with us so if you're an auditor yes please do reach out uh we are inviting quotes as we speak and I think uh we're

currently maybe middle of February is something which is looking like a tentative test net date uh main net date sorry uh and we have a couple of people only from the test net that will be hitting main net with us uh so looking forward to that yeah cool well thanks so much for the update it's been great chatting and uh yeah during this podcast uh three super proofs have been generated on on at Quantum so I've been watching I've been watching the proofs uh get generated so yeah very exciting and um yeah uh really pleased about all the progress you guys are making thanks again uh absolutely man pleas pleasure being here thanks for

inviting me and uh looking forward to uh getting me that b soon yeah cheers [Music] [Music]