zkAcademy #8: zkRollups-as-a-service with Altlayer, presented by Union Build

Transcript

[Music] [Music] [Applause] hello everyone my name is Amit kumur and I'm the chief operating officer at altl we're building a rollup siiz service platform so um evidently this part of the uh educational series is going to be on rollup a service so try to understand what it what it is uh what what what makes a ro of service and what it

actually offers to end users and end clients so before we do that um let me give you a little bit of background on why we are why people are building roll up a service and why you actually need them so many of the people in the space you must be familiar with with uh chains l1s like ethereum Solana and others basically they operate in a mic model where you have these nodes in the network that process your transactions so they basically execute transactions they reach consensus on the transactions so basically the set of the transactions and the state of system and they basically make sure that the data the transaction data is available for the people to check and verify the state so this is how almost every single uh chain

that's out there in the market today operates there's a second Paradigm which is often called a modular blockchain stack where the idea is to basically separate these components outside and not ensure that all of these components are running on a single platform or single L1 and so you take out execution and that basically becomes rollups right so you see the likes of you know optimism arbitrum zking they're all execution layers so they execute transactions and they're attached to a base layer which is ethereum the second part is the settlement layer so where basically is it is the network which basically decides on whether um you know a certain state of the rollup is correct so for example if you are looking at a

zq rollup then the settlement layer basically verifies the proofs generated by the CQ roll up and the third part is the consensus andd basically it's the it's the network that decides The Ordering of the transactions and make sure that those transaction available for someone uh let's say who wants to prove that the system is operating in the right or right way so over the last year or so it has been well established that modular design philosophy or mod designer stack is the way to basically scale blockchains and so almost every single I mean my belief is that almost every single Mony change will end up becoming a settlement a or a consistent D and execution is the one component that will obviously move out of the base

chain and the good thing is that once you move out the execution component you can actually build interesting things so for example imagine you are building today on ethereum you are bound by the runtime that ethereum offers you for example currently offers evm so you can only run contracts that you know um compiles down to evm but what if you wanted to run wasm what if you wanted to run a Solana VM or something of that kind you can only do that on an execution lay if it's outside of eum so rollups basically enable that possibility make it possible for you to basically run arbitary run time but still stay attached to a s settlement layer like there and because of this uh you know popularity of model you know chain stack

um my personally believe is that execution is the stack that will see the most number of or so far even by the way has actually seen the most activity in in U let's say in terms of develop activities so it has seen the most amount of work being done and because of that uh execution because execution has moved out of the base chain we started to see different kinds of execution lay being built so for example imagine I the simplest case would be considering let's say optimistic rollups versus Z rollups right so it your rollup basically execution lay has a certain type of security guarantees a security guarantee could come from a fraud proof based mechanism or it could come from a cryptographic proof which is which is

like usk rollups um a rollup or ex could have different run times as I mentioned earlier so could be runtime it could be running AVM it could be running wasum it could be running Solana VM or you know more recently you could even imagine running something like a mov an execution layer may also give you different may also have a certain purpose so you could imagine an execution lay like arbitrium optimism that serves a general purpose right so imagine deploying all sorts of contracts on that rup so similar to what you can do on them today while you could also Imagine building execution layers that are sector specific so imagine building building something like a Dy rollup on which you can put you know applications like a make a DA and you know unisoft

and the likes or you could go even more more um sort of fine green and you could say hey let's go and build an app specific rup so you could say hey I'm building a fully onchain game and I want to make sure that um this game runs u in a certain with a certain setting for example you know let's say I don't want to have gas mechanics on my on my network and can you do that you can do that only if you build an app specific execution L and these there's this last category which I call a female rollup so these are rollups that you can spin up and you can use them for as long as you want and when you feel like you're done with it you don't need them anymore you can take them down so these are basically temporary

rups these are rollups that you spin up only if you need them and when you don't need them you can take them down and then the last I guess treat of these rollups could be or execution layers could be on which chain do they settle so they could be settling on a layer one one like e theem or they could be settling on Layer Two like let's say arbitrum optimism and others and so basically by the way these are not the these are not an exhaustive I would say you know traits that you could have imagine for extion L you could imagine other things um you know for them as well but basically the idea is that once you have pushed out execution out of a base chain out of monic chain you get all the flexibility and that flexibility basically creates opportunity to build different kinds of

EX execution depending on what you actually want them to achieve so a good example would be you could imagine building an application specific execution layer that uses fraud proofs uh but relies on um let's say run CDM but is actually settling on a base chain which runs VM for example near protocol and you could do that only if you move execution out of that so it's all good however when you when you launch all these kind of kind of when the execution moves out and when you start to launch different kinds of execution layer you actually do see a problem and the problem is around fragmentation of the execution space so you can imagine if you're a builder now you have the problem to decide okay which kind of roll up which kind of

execution do I choose and pick and how do I customize them to suit my needs right and there are n number of executionist out there in the market today you know of course from the most more popular ones like let's say zika sync polygon arbit from optimism which are much more general purpose in nature uh to much more sector specific spe as well so you could imagine arum novva which is designed for U you know game specific or those kind of applications which require uh which require low gas fees and they basically rely on a on a certain type of deay and do not necessarily settle on there you could imagine different kinds of and there's different kinds of you know rollups that use certain in runtime so not necessar

evm but maybe aome you have rollups that do not actually have any settlement like Sovereign rollups right so basically now you have this wide V variety of you know extion Le out in the market from a developer perspective at some point you have to decide and choose which one is the right uh choice for you and this is where uh there are some pain points so one is developers if you're building if you're building a dab you have to deal with this all the heterogenity that comes from the execution layers that are out in the market today and there's no outs say Unified framework that could use to decide and spin up these instances of exist and even worse uh imagine if you

do not have a let's dedicated technical team if let's say if you're building an nft project if you're building a gaming project where you do not understand the technicalities of how rollups work some of these rollup sdks are being offered today they can be a little bit intimidating so it could be too low level for you to understand and use and so there are benefits of having moving excution out of the base chain and have this model approach but then also there are demerits and disadvantage as well and this is what I'm trying to mention here which is it becomes a little bit complicated for people to pick and choose to give an example why it's why it's important why how challenging this problem is uh DS are not a great example but you could pick any any any any instance of

any let's say app chain today generally speaking it takes quite a bit of time today to go and spin up an lb uh go and spin up an app chain or app rollup to be honest with you it looks like you know if D an example the reason why I'm putting this because it is one of the biggest application on the etherum space that is building an app chain today it's not necessarily roll up but you can imagine why rollups become interesting and why it becomes easier for why it's important to make sure that the mistakes um or the issues that we see in the app chain space should not be repeated in the app rollup space and so um back in 2017 a lot of people were building different kinds of l1s for example TSA polygon Solana and ethereum

as well theum a little bit much earlier but basically at that time it took about a year and a half sometimes two years three years to actually build a fully you know performing performing blockchain like an L1 and if you look at dydx which decided to launch their own app chain recently I think it was yesterday or today when the dydx main at went live it took them atast at least a year and a half if not more by the way so what I'm trying to argue here is that the time to launch an app chain is kind of the same or roughly the same as the time to launch an L1 which is not ideal because the whole point of some of the sdks that are out there was to basically make sure that the time to launch these U you know apps speci ex Chains would be you know would

be much shorter and that's not really the case of course in some cases of course if you're not customizing too much it should be straightforward but there are certain challenges by you know if you're launching an app and the underlying point is that blockchain infra is still in a stage where you still have to do a lot of hand holding you still have to put a lot of Manpower in building an app specific infrastructure and this is what we should change this is what I think app rollups have the ability to change and so um the question that I would like to raise here is if you wanted to build and make it easier for people to deploy or launch an application specific execution layer or application specific

rollups uh we believe that there's something to learn from Amazon web services even though many people in the crypto space may not like AWS but the fact of the matter is AWS is one of the backbones of almost every single company that's out out there in the in the space today right from startups to pretty much well established companies as well so there's something to learn from web to and I think this is what uh where ramama service comes comes into play so basically if you look at AWS uh when you want to let's say launch a launch an app and you want to host your app somewhere uh in in in the prehistoric or the pre AWS era basically you have to go and buy a hardware you have to go and

buy servers and then put all your application host application on those servers um and when that and it's also not very scalable because um imagine uh let's say your application you know drives a lot of transaction volume during let's say Christmas time you had to go and buy Hardware before Christmas and post Christmas when the sales have gone down you basically had to figure out a way in which you could basically get down or reduce your your expenses uh by removing some of the hardware that you're that you're operating and so the goal was AWS was to basically make sure that you know your system runs much more efficiently and in an optimized way so it basically helps you reduce the cost and this is kind of what we want to achieve with with uh our

execution L as well basically you want to make sure that any anyone who wants to spin up an execution lay of any type should not worry too much about you know burning millions of dollars of of money to build uh that rollup the second thing it has to be fast right in the sense that uh when you want to spin up this um this rollup it should be for for today let's say imagine you're building a startup it's a helps you in many ways right one is basically you can go and spin up an instance quite quickly and you can take them down if you don't like it or if you feel like you know you don't use want to use that anymore and so you want to have an infrastructure that basically enables rapid Innovation and basically scales with the application and that's something that

doesn't happen that well with the info that exists today in the market right and and the other thing that we would like I like to highlight here is that unlike AWS which offers a very centralized service an ideal model to make it possible for people to deploy application specific Ro option b a model that is somewhat decentralized somewhat trustless and somewhat permissionless and this is what many of the rup service Builders are doing today right so basically um for those who are new to the space And for those who do not are not familiar with what R service is and does basically they offer and help you run uh your rollups so imagine you are a game

builder and you are building a fully onine game and you feel like it would be best if you run this or your game on an app specific rer because you do not want to worry about G mechanics and things like that and you want to for example offer a gasless experience right you can only do that if you could customize the underlying rollup and so Bly you do not want to worry about oh who is going to run my sequencer how am I going to run my RPC endpoints my explorers uh my indexes and so on so you basically Outsource all of that infrastructure support back to a provider that is basically the RAS providers the r service provider they will basically run this service for you where you do not have to worry about hosting any of the node infrastructure that comes with

it and so um many of these Ro up as service operators that exist in the market they offer somewhat like a SAS model where you you as a developer you basically pay them some sort of monthly fee and they allow you to um you know host their rollup on their on their infrastructure now um here's an example of some what you know how this rollup rollup service um service could look like where basically you go to a dashboard similar to a SAS like dashboard and where you can go and say Hey I want to deploy a rollup I don't want to write any line of code uh and can I do that so basically this dashboard would allow you to

basically click and pick and choose and customize different you know things for example you could say hey I want to have my own gas token I want to have uh certain blog gas limit and things like that you can basically do that with just clicks and buttons you do not have to write any single line of code all the complexity of on underlying SDK for example let's you're using op stack or you know arbit orbit those those sort of things they all kind of get taken away from you so basically it becomes much more simpler for you if you want to go and deploy rollup if you want to get a feel of it you can go to dashboard. all technology and you can get get a sense of how easy it has become uh to deploy rollups through some of the RAS services that

out there in the market now um let me highlight again what are the benefits of R so one it's low cost should be low cost because you do not want to go and spend millions of dollars in U one in hiring the right kind of people to go and deploy this and two uh it should be low cost in the sense that uh it should be fast to spin up and fast to fast to take them down if you don't need them anymore uh many of these draft providers actually should offer and actually do offer a no code framework where you do not have to be an expert in you know in rollups and blockchains even if you're a marketing person or VD person person or a project manager should be a you should be able to go and deploy a rollup through a no code dashboard no code

framework and the other thing is um because it's a universal most Ras providers offer kind of universality basically what I mean by this is you go to Ras Ras dashboard or R platform and you'd be able to pick and choose let's Your settlement layer Your da layer and different other services that you would like to um you know use and so uh it becomes like a one you know one single platform that allows you to you can choose different bits and pieces that you need for your for your rollup and this way you as a developer you can focus on your code expertise so example you're a game builder you would focus on building good and funing and engaging games rather than worrying about the backend infrastructure and how that runs um now as I mentioned many of the

RAS providers or Ras service kind of sounds like a SAS model where basically you pay them R provide a certain amount of money per month and they basically offer you this nodes infrastructure services now uh there's one thing that you have to both as a user and both in this space as well you have to understand how they operate so um typically if you if you look at the SAS business model in particular there are different ways in which SAS products actually make money so you could imagine a flat pricing model where anyone has to pay I don't know $10 a month uh for using that SAS service there could be models where you pay them you know in in based on some usage or some tiers so you could say uh if you're using it for let's say I don't

know this this many transactions or these many this much of data then you're allowed uh you you pay would you you you would have to pay let say $100 a month if you using you know certain different tier you would have to pay slightly higher there are different hybrid pricing models as well um and on the other side here there are models which are much more metric based you would say depending on how many users are coming in uh depending on how much uh traffic you're driving to the SAS platform they will charge you based on those metrics as you can imagine not all of these models are ideal for Ras products so if you're a Ras Builder or if you want to use a Ras service you could imagine not many of them are actually going to be a good fit this is where uh

Things become interesting because a Ras provider has to consider certain cost and that cost basically goes back to the end clients right the first cost is the cost of uh let's say if you're building either building let's say optimistic rollup or a zq rollup you have to consider the execution cost so basically the cost of running and processing your transactions and because that's basically run and managed by the RAS provider he has to Ras provider has to basically take that cost into account the second thing is the proof cost so basically the cost of if let's say the RAS is is provisioning a z roller for you in that case the RAS might actually be running a Prov and so there's a certain cost that is associated with running the pr itself because provs at

the moment are quite expensive and it takes quite a bit of time to generate approves and so approves are quite beefy and so the rest has to take uh you know in account the the the proof generation cost as well um the second cost is the da cost because eventually you might be actually posting data back on certain da layer right so it could be ethereum it could be Celestia it could be some other networks uh and so um so there's a cost that these raspers have to pay uh for the da cost in some cases and depending on which role of Stack you choose in some cases you do not have to pay for da itself because da in the sense that you don't have to pay for the transaction publishing itself but you have to pay

for the state Delta for example there's a rup that does not put the full transaction raw transaction data on on ethereum it only puts the difference between uh the previous state and the and the current state um the RAS also has to in some cases has to post um has to post the vality proofs and these vality proofs basically gets verified on chain and so there's a cost associated with that as well and so um in this course I'm I'm I'm removing and not taking into account any cost that comes from let's say Manpower so because Ras pro has to go and build a team to be able to help you establish and help you run these rollups uh I'm not taking any of that into

account I'm mostly looking at the cost that's let's say the bare minimum that the RAS has to pay because provider cannot escape the proof generation cost in case of ZK it cannot escape the da cost it cannot escape the cost of uh proof verification on chain so that's the bare minimum cost that the RAS prod obviously has to pay and then of course there's additional cost of let's say non- infra cost which is the Manpower cost for example um so um let's take a look at the cost depending on which rollup Ras provider Provisions for you so imagine if you're using a ZK rollup uh Z rollups today are actually quite more expensive than optimistic rollups simply because they require proof

generation and proof generation requires very beef infrastructure and so it's it is quite expensive and I'm not talking about the the gas fee cost I'm talking about the gas fees can be much cheaper so if you are let's say using a zq up uh from end user perspective the cost of running a gap the cost of sending a transaction is actually could be much cheaper compared to optimistic rollup but from INF providers someone has to go and run that Pro and that Prov is actually expensive um this second part is the da cost we basically the cost of putting the data so imagine if you're running a ZK rup there are certain ZK rops as I said earlier they do not put put the raw transaction data on the on ethereum so the cost is actually quite cheaper

compared to if you had put all the raw transaction data on ethereum um and so if you're putting just the state Delta on ethereum then your uh ZK rollup can actually have a lower da footprint compared to an optimistic R because an optimistic in general they actually put the whole transaction data um there is a cost of as I said of verifying the proof that you put on ethereum that obviously doesn't exist in optimistic rollups but so it only applies to Z rollups and so there's a cost cost for that as well um and similarly the cost of proof verification that does not exist in op optimistic rup but there that does exist in in ZK rups there's one thing that I haven't

mentioned on this slide which is the fraud proof part um so fraud proofs at the moment do not get triggered simply because um most of the rollups or if not all of the optimistic rollups today in the market they operate with centralized sequences and so the chances that a centralized sequencer will make you know malicious um let's say State commitments is actually quite low so which means that you know fraud proofs actually never get triggered practically speaking real estate speaking given the current uh setting things of many optimistic rollups and so now going back to the business model that I mentioned earlier you could have um on everything that's on the on the right on the on in the box so premium model Fiat pricing tier

pricing for US user pricing hybrid pricing these are not quite ideal for um you know Ras models because they are very predictable right so you know how much if if you're looking at Fiat pricing a flat pricing model you kind of know that okay I have my user will have to pay $10 that's not really the case with blockchain systems as you can imagine because in most cases you're running with the price auction kind of model where basically the network and the demand basically dictates how much uh it will be how much how costly it will be to actually run some of these systems right and so um the ideal SAS model would be something that captures uh

usage as well so basically if let's say a network if a rollup that has been spun by a Ras provider if that rollup basically gets used a lot which means that there's a lot of usage for that rollup then the rollup service provider will actually charge you slightly more compared to if the rollup was completely ideal idle so um now this is towards the end of my my my talk so let me let me tell you um some of the things that you know we are thinking as a Ras provider I think generally speaking the space has to see as well as space has to believe R rasers have to do something differently as well because so far what we have done mostly is that we have basically taken an SDK so we have taken let's say or

roll up SDK like op stack orbit orbit and be basically putting hosting them in 18s but if if you go back to if you remember my previous or my my initial argument is that you we should do something differently it has to be we should as a Ras provider we should build something that is much more transparent much more decentralized much more permissionless so it should not be an opaque system like the way ad BL operates but sadly this is what we have end up becoming right we have mostly end up becoming an ad BL rapper and that's not ideal so what's what's the future uh and how how should a better Ras look like um so if you wanted to move towards decentralization right so imagine today when you as a roller Builder you go to a

ASAS Prov and say hey I don't want to run any of the node infrastructure I do not want to go into this uh understanding how sequences can you go and manage and run my notes uh you know for my infrastructure and prod basically does that for you and that's ideal in some cases but what if at some point you wanted a flexibility you say you wanted a situation where you'd say hey I now work very closely with Ubisoft I now very closely work with some other gaming companies and I want those gaming companies to become my node operators my sequences can you do that and so this is kind of the model where I believe the RAS provider should eventually go where it becom a much more open system rather than a closed and opaque system where uh if if where basically you riant on on a Ras provider

to basically run everything for you and that's not ideal because it's kind of opposite to what what we have been what basically opposite of the blockchain ethos in general right so here's here's a view of things right but basically Ras uh would become more like a network instead of being a centralized party which runs you know it will become a network where basically you as a developer you send a request to the rollup and say hey I want to have a rollup um and uh I have a certain request certain requirements for this rup so you could say I want to have a r up that settle theum that uses celesti as a DA has a certain runtime it is optimistic in nature it is uh you know app specific and maybe You' specify for example how many sequences you want and

you could say hey anyone that can come and uh that wants to become a sequencer has to deposit some collaterals so you could say okay if you are let's say staking eth and if you could stti that eth through icon layer then you're allowed to become my node operator so it becomes a much more open system where you're not relying on one single node provider but you're you're tapping into a much more decentralized Network to run your run your infrastructure and then uh this this R network instead of becoming a centralized n centralized party that offers you service now it becomes much more like a marketplace where um it basically matches uh people who want to have rollups and people who actually want to provide let's say sequencing

services to that rollup so imagine you said oh here I want I'm looking for two sequencers and anyone that wants to my to be my operate my rollup can deposit a certain amount of collateral and if you do so then you can become my operator and this model sequencer will come and deposit collateral and if they deposit collateral in the network then the RAS network will basically select let's say some some of the sequence randomly and then assign them to operate that so this way Ras moves from a Ras operator to a much more Network and protocol level setting so it becomes much more open decentralized and permissionless and not like AWS in a close system the other benefit of this is um that now you can tap into the network effect of all the

sequences so imagine for example um let's say you have block demon figment and other very wellknown validators in the market today if they start to support your network you can tap into their expertise and their Network effect their marketing ability uh to support your your your rollup that's something that you cannot do if you're just using a Ras providers the ideal world will be where some where where these Ras providers will basically move away from a central from being a centralized entity to a much more open system well that's it for for this session thank you very much for listening and if you want to learn more about uh rollups uh and PR Services you could head over to our docs page docs.

allio thank you very much for this thanks em for the comprehensive presentation I really like the uh the comparison with the traditional software as a service model uh I've done a lot of marketing in the past with different models and uh it's been it's it's interesting how how they uh kind of form and if you don't mind leaving the slides up I I'd actually like to pinpoint a couple uh specific things on a couple of the slides so I'm going to ask you actually to to go back uh towards the middle for one of the first um slides where you compare to ZK and optimistic rollups and the cost of involved uh for rollup as a service providers uh and

that one was super interesting to me because I think a lot of teams here we go um was it this one it'll be no it'll be towards the middle actually um it was a uh side by-side comparison of uh there it is so uh this is something that that teams have to think about right like let's say we are a a Dap that is looking to deploy our own app specific rollup because we want to capture uh you know sequencer fees or we want to be able to have more control over our environment uh so you know when it comes to a a making this Choice with a role as a service provider like uh like allir um is there a is is there a kind of

framework uh that that that you guys on the BD side or that uh would be suggested for okay maybe you guys should go look to do a ZK role like is there specific use case application types or team desires customiz customizations there versus okay guys should do an optimistic rup um yeah that's my first question Yeah so basically um the reason why some people go for optimistic um is mostly because it's easier to understand it's easier to deal with so people feel like okay you're just basically running a running a system where um we're executing transaction that's mostly it but as you can imagine ZK is where ethereum is

going ethereum is very very hard pushing towards zups so I think um many people actually want to go uh want to be future proof in some sense right so basically they feel like okay ziki is the future and therefore they want to go with you know zi RS the end from um from a uh let's say client perspective um we have seen that at the moment at least stly and by the way this is going to change over the next six months and by the way this is changing almost every single day rapidly but basically at the moment it can be a little bit cost heavy if you're running a zqa unless the underlying uh let's say rra provider or underlying let's say rooll up as DK provider

subsidizing the cost of proofing so at the moment many people do go to Z s so if you are thinking about Zs you have to make sure that either you find a Ras provider that's going to give you let's say subsidized uh proving uh facility or you can tap into the underlying SDK provider that gives you um you know um subsidized uh you know proving service by the way there's some people who do that at the moment um and so that that that is actually the main consider the second thing is to consider is basically around uh let's say VM changes so for example example let's say at the moment most Z rollups that are out there are evm there's I think one or two that are building was some Bas rollups as well

but if if let's say if you want to have Solana VM based rollup or move VM based ZK rollup that's not quite out there so in that model you most likely will probably go with optimistic rups so I would say these are two main criteria one is of course the cost of proving um and the second thing is around um you know the the ability to customize ability to pick a certain VM that's something that's not available uh for I that's available for evm but not available for other other run times gotcha and then furthermore on this slide before we move off of it um because a because the role up as a service provider in this case alt layer

has a network um and and an infrastructure set up from sequencers to explorers and imprs and uh you know that there's different kind of aspects that you can just jump in and start to use compared to having to do this all on your own are the costs associated with whether it be ZK or optimistic uh are the costs of the provs in the ZK uh uh sense and are the cost of the sequencers in the optimistic sense are those shared amongst each participant or each chain within the alt layer ecosystem is there is there some sort of uh I I don't want to say preferential treatment but is there a a way in which it is itemized

based on block space used or gas used or how is how is the the these costs that we're describing in your first point here specifically on the ZK but broader how are those uh shared amongst uh different chains Within These rup AOS service ecosystems uh and kind of how are they like prioritized very good question so this is something that it depends on certain parts of the system so and also depends on clients PR so there are there are scenarios where the client for example wants to have a zq rollup but they feel like um The Proving cost could be quite expensive for them and they're happy to share that with other Prov other other rup Builders as well so there models where we actually offer a shared prover um to

across that basically is shared across different rollups there are certain scenarios where people there are clients who believe like look I want to have my own dedicated infrastructure that does not get share with anyone in that case we offer them a dedicated you know system um so it it depends on mostly on client's preference and how much client is willing to pay um however when we we do give people and by the way this is one that I think most rasters do as well uh where basically sometimes they give you a common ground playground like a Sandbox where it's shared with everyone but this is where you get your hands dirty this is where you get a feel of how much it might cost you eventually and then once you graduate once you feel

like okay I know I have a I get a sense of how much it's going to cost me or or then I I'm going to graduate to either a shared model or a much more individualistic model where you basically pay for your services no one is allowed to share so we basically give people preferences and you basically can go and pick and choose by the way this applies to almost every rups so you can actually Ras provider so you can actually go and pick say a Ras pick a rasp and say hey can I use a sequencer that is currently sequencing someone else and also sequence my and by the way this is something that I also this is where I ALS I should also mention something like espresso as well so basically espresso is a network uh which is which offers sequencing ability so basically you can

send your transactions to Espresso as a network and aspr will basically order your transaction send them back and they order that they offer that in a shared model where basically a network is shared across all the rollups out there so you don't have a dedicated sequencer for everything but in Ras model you can basically choose and you could say hey I want to have my dedicated sequencer or I I'm happy to have a sequencer which basically shared across different rollups and by the way there pros and cons to this model right one is that if you have let's say a sequencer that is shared across two rollups then you can actually have you know cross rollup composability cross rollup communication quite easily if let's say you have two different sequences that are serving two different rollups and then if you want to have cross roll messaging that

becomes a slightly more complicated business so also that is something that you should also take into account so for example you could say hey I have a rollup that is being um served by a sequence or a and I want to spin up a new rollup and I want to make sure that this rup actually can easily communicate with my other rollup I want to make sure that it's actually under the same shared sequenc so it actually gives you some benefits as well so these are out say some of the criteria that you know you have to consider if you want to pick a shared provider versus individual dedicated one yeah that's where was kind of going with that which I appreciate you bringing that up because um I remember I did an interview in Paris with Jordy from polygon zkm with regards to Shared

ZK liquidity um you know and I know that they're doing different types of stacks and ZK Sync has their hyper chains and and it's it's quite an interesting uh model where we can we can see different uh ZK rollups with sharing approver or like you say uh sharing a a sequencer uh within a rollup as a service uh provider lower the cost and then also get some of that shared liquidity and composability which I think is something that a lot of teams are are likely after uh if they are a a defi app or or some sort of nons siloed use case however in the same sense I think that perhaps you can kind

of fill us in on this is if that a lot of the clients that you guys speak to they're actually perhaps okay with being a bit more siloed from the rest of the ecosystem perhaps it's a game or something like a dydx like a perpetuals deck is this the case is are are the rups that are coming to to out layer to build their own sorry are the daps that are coming to all lat to build their own rollups app specific mostly looking for Silo use cases or are they looking for like more of a general purpose ecosystem or a bit of both I would say mostly app specifics um um there are some some scenarios where they want to have a general purpose L2 uh but even that when they say it's a general purpose L2 it's actually kind of

a sector specific L2 that they're looking for so basically they want to build a D5 uh ruup where they starting it's kind of like I me maybe it's not the right right name to use but it's kind of like the teral Luna thing where basically you spin up a a rollup where you start with a stable coin and then you add let's say an a protocol you a learning protocol you build something else so people are looking at that kind of scenarios as well where it's not just for single application but it's for a set of applications I would say that you know very closely can work with each other um but yeah but most in most cases it's app specific where basically they are looking for individual uh block

space where they can customize to suit their needs and do not have to worry about okay what about if I bring someone else and then they may have their own demands so I'll give an example of let's say gaming right so there are a lot of fully onch games that actually want to have their own rollup simply because um their contract is actually quite heavy but what I mean by heavy is that it's actually quite gas consuming in some cases you cannot even deploy them on general purpose see like arbit optimism today simply because they qu you know it takes in some cases like what 10 million 20 million gas limit directly just one single transaction and so they want to make sure that their blog gas limit is customized to be able to accommodate that much of gas limit limit and so the

question is yes you know there might be some other games that might want this kind of gas limit but maybe not all so in most cases I would say people are looking for app specific in some cases people are looking for sector specific not necessarily general purpose gotcha yeah that makes total sense if you could just uh go back a couple more slides there was another one that I was quite interested in as far as the question goes a little further keep going going one more two more I think it was like the Modular One um yeah sorry sorry I think it's number three there not this one the Fallen one yeah okay great either one of these is is fine so um

since we're or since we're focused on uh the ZK side of things here uh with the with the uh validity proofs um kind of being the the important aspect the differentiator between ZK and optimism really there's other other factors but that's kind of what what what we are seeing now alt layer can can support uh based on this graph is this a is this a general uh rough outline of what is existence today or does ALT layer have the capability to support uh you know a validity proof with a different VM for an ABS specific for Layer Two meaning a layer three is this very much a plug-in play based off the the current state of

the modularity ecosystem that exists and then after kind of that is this something is alt layer and as a service something that as that modular ecosystem grows that you guys can just keep kind of adapting and plug in playing yeah so let me start with this last last question yes the ideal would scenario would be this where basically let's say I don't know tomorrow someone comes with a new VM or tomorrow someone comes with a new kind of ZK proof by the way just talking about Varity proofs there's not just one style right there are many kinds of proof systems out there in the market today right and so almost every single um you know CK rup that you see they have their own uh you know proving mechanism uh in many cases some of them are type one type two type three type four right and so basically

um the idea is to basically make sure that we are model enough to adapt to any new thing that comes in the market and by the way this is not an exhaustive thing you could imagine I don't Sovereign rollups have not I didn't include in The Sovereign rollup model right so you could have a ZK Sovereign rup model where basically there's a peer-to-peer Network which verifies the proofs there's no single contract on chain that verifies the CK proof but so so the answer is yes um that would be the ideal model but unfortunately as a Ras Builder our we do not build many of the underlying SDK stuff so for example we do not build uh let's say polygon let's say we tap into polygon cdk but we do not go and build uh polygon cdk style systems we

sometimes do contribute to their ecosystem do contribute to their code base so so there what what so from that perspective from that angle you can't build everything at the moment today for example you can go and build let's say a ZK proof system that uses that let's say has evm but you can't have ZK with Solana at the moment there's some people attempting it uh but there's no for example ZK with mov V for example there's nothing that exists in the market right so our goal is not to basically go and invent all of that our goal is to so if someone builds let's say ZK move VM rollup then we would support that that's our goal but at the moment you know we like this is this is something that does not exist today in

in entirety but this is what I would like to present this is how the space would look like and this is what modity means where you can basically take a cross product of the whole Space so you could take Varity proof and you take any runtime here you take any purpose and you could pay take any L1 L2 right as the base chain um but yeah um not all VMS are supporting the ZK model today but you can take any zkm today and deploy them as uh an L3 or an L2 quite easily and that's something that we obviously do today yeah that's so exciting that um that you know as as the boundaries are pushed on the Innovation side because if you remember polygon was working on zkm since I mean the like 2020 maybe early

2021 and then it was just earlier this year late 2022 that it even kind of came out I remember the uh ECC speech there again from Jordy actually uh kind of announcing that they were open sourcing the code um so these things take time uh but because you can kind of position yourself there as as a as a I mean I I I hate to say selling shovels during a gold rush but effectively being a infr provider for for people who are trying to you know that that allows you to have flexibility and a a agility so um to add to this basically at some time point of time you may also want to switch to some something so for example

you could say hey I'm using ethereum as my settlement my D today can I switch from ethereum to Celestia right or can I switch from let's say um let's say I'm running a ZK Ru up today and I'm putting let's say all my transaction data on ethereum today can I switch from posting all all transaction data on etherum to let's say putting only state Delta right and these are certain things that we should also be able to adapt so we should make we as a Ras provider have to give flexibility to people where they could change some of these things at run time so while the system is running you can basically change some these things as well of course assuming and and making sure that that does not you know create security issues but yeah this is some these are the certain things that

you would be able to customize hopefully in the future as well with zups yeah that's really really really cool um I know we just have a couple minutes I just want to ask one last question uh specifically about the runtime uh are are you pretty bullish on the Move uh to different VMS for all different types of use cases and onboarding developers um yeah I would say so look I mean even ethereum right ethereum and evm is not super paralyzable right so someone has to go and build a parallel version of evm right Sol is a good example of it's not evm but it's a parallel instance of of around time so I think I think I think

you know it's there's room for building interesting VMS I if you look at the traditional space there's no single language right there's no just C and just Java and just go right there I don't know how many languages out there in the market today right and so I believe that people should get flexibility and and evm is not the best yes it's most popular but it's not necessarily the best run time out there both in terms of performance but also in terms of security right so I think there's a room to bring new innovation through new run times and you know for example wasam is a great runtime right because it gives you flexibility you can run different languages that can compile out to wasm it runs in web browsers and things like that this is something that you can't do with something like evm so

I think there's there's a there's room for other run times including move Solana and some of the others that might come in the future yeah thank you so much for uh coming on it's been an absolute pleasure uh to sharing what you guys are building and uh look forward to seeing more and more of it in the future my pleasure thank you very much for inviting me Che