Should You Build Your Own Rollup? with Hilmar Orth

[Music] we got three more talks to go and next up I'm excited for this actual talk we're going to talk about how and if you should build your own rollups there's a lot of stuff going on in the world of rollups it's got getting easier and easier to launch your own rollup chain but there's a lot of nuances to that question and talking about that will be Hillmar from gelato telling you all about whether you should build rollups please give hmar a big round of applause thank you thank you thank you how's everyone doing awesome so maybe before we begin with the presentation maybe uh if you can raise your hand if you know what a roll up or Layer Two is okay most of you um who has heard about the fud that is going on on Twitter about new l2s sorry and no basically why why should you launch another L2 what's the reason is is it actually great for ethereum or not right so a bunch of you uh maybe just briefly who's heard about gelato before okay cool nice then let's Jump Right In uh the talk is should you build build your own layer too and the motivation for this talk is actually there has been a lot of discussion discussions online on Twitter recently about new layer tws launching and people criticizing them of hey should you actually do it and who should actually do it and how much does it actually cost and what are actually the benefits of doing it and is it just all narratives and um stories that people tell you to dump a token on you um and I want to shine some more light on the reality of actually doing doing this and launching al2 and if you should actually do it as a developer yourself and I guess there are a lot of developers in the audience today by the way I'm HMA one of the founders of gal and this is roughly the agenda uh before we jump into um into it I want to talk quickly about what we do at gelato to give you some context why I give this presentation um gelato um is a rollup as a service platform and we help projects launch their own rollups so of course I'm a bit biased because I help projects launch a new l2s and their new rollups right but roughly how you can think about it is that you can come to gelato you can go to Alli and literally in a click you can launch your own Layer Two on theorum on base with the op stack with the AUM orbit stack with any cool stack that you like and the cool thing about it really is it's that you don't only get like the the rollup and then you're kind of like lost and you can't really build applications that require maybe um pay masters bundas some automation some verifiable Randomness or or whatnot you get all of these bundled in serviced by us our first part uh Native modules that we have and then we also have a Marketplace of all the good Bridges wallets block explorers everything you really need that we uh provide to you in a single click so you have literally the same environment that you're used to on like Bas or ethereum on your own rollup and you can build scalable applications but of course the question is should you actually do it in the first place um these are some of the examples of chains Al on Jato yeah you are seeing right there's fox in there so Fox News has a chain on gelato there are some of the some one of the Enterprises that are working together with us there are a bunch of arst Lisk Raya Ever Clear LF zero so you name them so if you want to launch a rollup you're not alone and this is by the way how it looks like you can literally just go to the UI click your stack click your da layer click your native gas token you want to have click the services you want click deploy and then you have your own rollup so it's it's pretty simple um let's quickly talk about the state of layer twos um is is it a hype what's actually happening so what we have seen in the last couple of months especially since the beginning of the year is that there is a explosion of new layer twos being deployed um and you can see this big uptake since 2023 and they are probably even more now since we are going into that Q3 just ended and we are going to the end of 2024 um and I think what is always interesting to look at is what are who are actually the applications that are deploying the on layer tools right we hear a lot about um the general purpose layer tools building their own ecosystems we hear a lot about per deck and so on but I think what people don't talk about a lot are they all the games that are launching and I think especially for games layer twos is a great use case because what games really want is games want High throughput games might not care about having the exact security properties and decentralization properties of ethereum right they just want to have a really scalable virtual machine that provides them with verifiable computation that all of their customers can follow along if they like and so actually what we are seeing is the biggest sector of rollups is actually games which I think uh super interesting and I think we are now seeing the first couple of games launching it actually have real users and not just like play to earn um type of um incentivized playing um and yeah if you look at all these icons of projects that you probably know we see that a lot of the alternative layer ones that have been launched in the previous cycle they now pivoting to uh layer twos like Lisk ASA zero uh applications big applications such as Unis swap worldcoin fra Ray they're all going to their own chains Enterprises coinbase has Bas Fox and there actually some new ones uh that we excited to announce like in a week or so that will also launch with us and then of course the games you know like a proof of play like League of Kingdoms they're all deploying their own chain so a lot of things happening in the layer two space but does it actually make sense to launch your own Layer Two and these are just like screenshots that I uh made Lester day um and there are some some fat going on on Twitter where someone is claiming that it costs you $14 million per year to launch your on L2 and should you actually do it um which I think uh I want to try uh to argue against today and uh yeah can you and people are making fun of launching l2s and I think uh this sort of like framing is a bit flawed but uh let's try to go through the argumentation here together and see if it actually makes sense for you or or anyone else to to launch an L2 and so let's start with the the drawbacks right why shouldn't you launch n to like why why is that maybe not the right thing for you as an application developer as an EOS developer right so first of all the costs we saw $40 million it cost $40 million to launch an L2 well that's it doesn't cost $40 million to launch an L2 I we launched literally a bunch of l2s with a lot of top projects um and if you really need all the infrastructure that most of most of the application available require which are like the chains the rpcs the block explorers maybe if you do defi you need some oracles to get offchain data in you need maybe like account direction or relay services to abstract gas and have a greater ux right uh you want maybe multi support in order to manage upgradability of smart contracts you might uh need external bridging which is not with the canonical Bridge you might need indexing for displaying data smart contract Information Services these are sort of some rough numbers but roundabout you get to like 18.5k let's call a 20K per month right which is definitely not $40 million right it's not cheap it's not if you're just starting out and you haven't raised around as a project you probably don't want to pay this right but if you actually have users and you if you actually have some money and you actually need scalability then it might actually be worth it and my prediction is what I'm seeing in the market costs are coming down significantly so I expect like like latest than a year from now this will be $10,000 and two years from now this will be $5,000 and at some point it is literally just everyone can afford to deploy their own L2 but I'm getting ahead of myself what are some other drawbacks that exist for launching your own L2 that you should be considering so you have these fix cost that you have to pay infrastructure providers gelato and other people right and but they're also variable costs because launching your own chain means that you have to pay for security for your chain and that usually means that you have to pay a data availability provider and potentially another chain where you want to settle to and these transaction costs first and foremost have to come from you you have to fund like a sequencer account or like another account that then pays for the gas of these networks this could be ethereum if you have an ethereum R up this can be Celestia if you use Celestia or whatnot right um the good thing about this is that you don't have to like Finance all of this out of your own pocket you can charge your users right because there are these things that we all are aware of which are gas fees right and we pay a lot of them on ethereum slightly less on base or AUM or whatnot um and these actually go into your own pocket if you have your own chain so usually how projects should be configuring the chain is that um if users actually use it and they have to pay for gas then you should do it in such a way where it's actually at least break even so you cover your variable costs um by the way an interesting use case here um and this I think we just have to normalize I think we're just used to chains Never Dying especially l1's just keep on running and running but I think we have to normalize chains dying I think it's okay for chains and layer twos to die if no one uses them and and if it's too costly to operate them and I think this just has to be become the norm and there has been an example of a chain that uh sort of didn't make it or actually ran out of which is the public goods Network it was actually one of the first of PCH chains back in the day uh I know the team great bunch of people but they actually got wrecked by the variable cost that they had to pay um to ethereum back in the day uh because they paid tens of thousands of dollars per month in data availability in settlement cost to ethereum because I think was even before blobs got shipped so they had to just like post the transactions to ethereum and was still in a very sort of high gas environment and so they literally got broke by just posting uh empty blocks like no one used them and so empty they posted empty blocks to ethereum and then they run out of money bad timing I guess uh the solution would have been to actually have some users before you launch a chain right that then actually help you pay for it and potentially use a more effect cost effective data availability layer which to their like uh to to to be fair to them wasn't available at the point so they just had bad Market timing so you have to cover variable cost you will lack Atomic composability with other applications right so if you are building an application that is for example relying on being composable with a Unis swap or with a landing Market because you're building like a structured product on top of Ava that shorts AA and goes long somewhere else right there and you want that all to be atomically composable then of course you should probably build on a chain where you have all of these things together on the single chain right you can probably build the same use case in an asynchronous composable way but usually applications are not built in that way today so this is definitely a trade-off that you have to be fine with you you're trading of composability you're also requiring your users to bridge into your chain themselves right and I think we all still used to these old bridge uis where you have to like oh how do I get from this chain to that chain I think that's the problem today I think in the future there these B wordss like chain raction stuff I think it will be much easier I think the there will be tools that help find your chain for customers without them having to go through your eyes and and and go to the list of a millions of chains to select yours um then of course we just had a call by our friends from circle circle unfortunately doesn't support native USC for all these new chains that are being deployed uh because it's very difficult to do so and it requires a bunch of work I try to push them to always do it hey please do it for all of our chains they say no we can't um uh hopefully eventually they can uh but of course your binance right now doesn't accept deposits and withdrawals from all these new chains that are deploying so you have to hop over like Bas or hop over arbitrum if you want to do it and I think most importantly or like one of the biggest reasons I hear when I talk to projects a lot of them get GRS from layer ones and they're like hey do I get a grand for building my Layer Two no then I probably want to build on the layer one right uh but as a fun fact they're actually also grounds from layer tw's uh like framework developers so you're not but let's go to the benefits they are much more fun anyways and I'm more bullish chain so I I like to focus on the the benefits here um and first of all is it's actually a way to generate revenue and why because all this Revenue that there is a lot of if you have an application with actual users then there's a lot of untapped Revenue that you are forgoing to the network that you are deploying on which might be ethereum or which might be base for example and and this especially let's say if you Deploy on bass or any other layer too then what you are basically doing is you are using a reseller of data availability services so Bas is paying data availability on your behalf and you are paying base which is then paying ethereum um and they are taking a cut right and that cut is their margin and if you look on the coinbase annual report I don't know how much it was four or 5% of the entire annual revenue came from base sequencing that you guys are paying as users right so that's a good business model obviously you need as as many users as base right um so you can internalize this there's this chart of actually rollup that that is on gelato and you see the curve they're making like 25k a month in profits for sequencing free that's nice why not uh you can do the same with me uni chain that's the whole point of like unisoft launching on M $800 million got leaked in M since the update to POS of ethereum this can be internalized and like most of that was probably e Unis swap trash toxic flow Etc um unisoft Can internalize this and give it back to their LPS that's a pretty big number so that's a good reason um um I want to speed up a bit so block building is another thing like um it's not only about internalizing M but you can do all sorts of stuff if you control the sequencing of your chain you can say every time on the top of the block I want to update Oracle prices for my per decks in order to minimize me I want to run all the transactions that get processed on my chain through some sort of kyc or uh analytics service you can you can do all sorts of things that you can't do by Outsourcing sequencing a block building to external Pary so it just gives you much more uh control you can do as which is application specific sequencing which gives you a lot of powers you can customize your rollup for like maximum security or minimize cost like the games can say I don't need ethereum security I just need scale and they can really sort of optimize their own chain for this and uh you can you can customize the bridging configurations the security you can do upgrades like right now if you Deploy on a chain you are beholden to what their I don't know token holders or whatever decide of how they're going to upgrade their chain right you have no control over it and I always think like if you're a big Enterprise and you want to launch a chain do you want to be Holden to some I don't know network of people that are making decisions on your behalf what these upgrades will be potentially I don't know but sometimes probably not and so having your change just gives you much more control over all of this um and most importantly I think for projects that IAL to that need scale is like they avoid the Noisy Neighbor problem you're not sharing your chain and I I see chains or l2s as like verifiable servers and you're not sharing if you need to scale you're not sharing your server with someone else like with some other application that has nothing to do with it you need your own dedicated environment where you can actually scale scale into and so having your own Chang just gives you that scale and eventually this is something we're working on we actually want to enable horizontally elastically scalable chain so if you are deploying application you shouldn't Deploy on one chain you shouldn't Deploy on a share chain you shouldn't deploy just on one chain you should deploy in a system that can elastically scale um the environment for you as your usage grows and so you might have four different chains at some point just for your application so it's a summarize this is very simplified model right if you have if you expect to have a lot of users if you expect to have some me I think it makes a ton of sense to have your own chain if you have no users right now and you don't expect to have users you're just starting out you don't have product Market fit you're just testing don't deploy a chain go to on existing it's like a co-working space right go to like a co-working space if you are preed and the co-working space is like arbitrum you can go there arbitrum one you can mingle a bit you don't pay a lot but you share the toilets and everything with some other people if you want to have your dedicated restroom then uh do your own chain and of course it's it's still a bit expensive right so I I showed you the numbers it's doesn't come for free so you need to have a bit of money to pay for this yeah so how does the future of all of this look like from our perspective talking to a bunch of customers and stuff and not saying this will be the future right but I I think there's a high likelihood that maybe web 3 will actually look much a bit more like what we're used to in web 2 and so first of all I believe there will be millions of rollups I think everyone will have their own chain eventually I think you come to platforms to deploy your chain and it doesn't cost you anything it costs you as much as deploying like your own server on like a versal platform or something today um and I think everyone in the industry is working towards making this happen um I think it will be from a x point of view that you actually don't deploy the chain you're just saying hey I don't I want to deploy this app here's my GitHub import GitHub import codebase click deploy like on verel and in the background it spins up a chain for you but then as more ATT traction comes to your application it spins up more chains and it sort of shards the state between them and Rec consolidates it if usage goes down and up again so it's a much more elastic system um but you don't even see it you don't care about it like you don't care about how Google Cloud does it and right now I think there's a big debate of like liquidity fragmentation and like hey we have million chains do a million chains all need to split the liquidity between each other I don't think so I think what we are seeing is you will have a small number of chains own most of the liquidity and then the rest of the applications um accessing the liquidity be on their own dedicated application specific blockchain and who are the best sort of I think people or organizations to become these sort of I call them kind of like Bank chains that's a it's a dumb name really but um I think they are like the central exchanges and wallet providers that really have a lot of power here because they have they're like the entry point to the users right users come to centralist exchange users come to a wallet and they can and this is what you see with like coinbase and Bas Unis swap with the Unis swap wallet and all the other people at all of wallets I predicted metam Mass chain like two years ago I think it will still come um and they will all try to redirect the liquidity into their own chain and then just make it accessible to other chains that run application specific wallets uh logic um yeah that's what I just said applications are built on app chains yeah that's it so um if you're a builder if you're interested in building your own rollup or if you're early on the journey and just want some feedback on how what I actually what does it mean to actually launch a chain in production feel free to talk to me I will be around also on the weekend uh until Sunday and yeah I hope you learned a thing or two today thank you [Applause] [Music]