A Revenue Model for Based Rollups by Conor McMenamin | Devcon SEA

Hey everyone. Uh there's a little box here just in case I I need to see down on top of the front row. Okay, cool. So, yeah, my name is Connor. That's not my slide.

Perfect. Okay. So, yeah, I'm Connor. I'm from Nethermind. And uh today we're going to talk about uh revenue model for base rollups.

Uh uh Now, I wrote this title before I really sort of had an idea what I was going to talk about. And then I realized that revenue probably just refers to income. But base rollups are not just about When you're thinking about rollups in general, it's not just about income, it's about income and expenditure. And comparing those two is is how you decide if you want to uh be a part of the base rollup. And I guess not only are we going to talk about like the base rollup as a sort of an entity, but also the entities like the real life people that are that are participating in base rollups.

Um So, a quick primer on base rollups for anybody who isn't aware what they're lost uh is a rollup is said to be based or L1 sequence when the sequencing is driven by the base L1. Um So, basically the L1 proposers are in charge of of of producing the proposing the blocks for the for this rollup. And the rollup derives its state from the from the the sequence of transactions that appear on the L1. Now, a lot of rollups might actually you could you could squint and say that most rollups are like that already. Uh but I guess the key thing is that the L1 proposers aren't supposed to be sequencing the chain.

Uh So, the goal of today's talk uh is introduce the players in base rollups. So, we're going to look at each of the players and and like think about the income and expenditure that they have and should they be playing Should they be playing this game? Uh identify identify their income and expenditures. Uh Discuss how likely they are to have these income and expenditures are are going to change on the road map that's like upcoming uh the the big change that we're expected to see and then motivate you all to I guess participate in base rollups and be thinking about these income and expenditure problems that exist uh and then yeah I mentioned fast stacks for all but really what that means is everybody has positive P&L when when they're playing these games. Um okay so who's who do we have here?

So we have the users who are like the most important right? We we're always trying to sort of make although I was on a panel yesterday where we're talking about uh where we're talking about who should be gaining the revenue in a system and users were the were the bottom of the list but users should be the ones that are gaining something from playing the game so they're the going to be the ones that are going to be like driving the revenue for a base rollup. Uh proposers and sequencers I'm putting them in the same bracket. Uh ZK provers uh so we have this proposer over here and this base with a this base uh blockchain that's accepting the blocks. Uh we have ZK provers uh because I guess a stage two rollup should have ZK proofs.

Uh And the DAO so I guess when maybe some of you are here and you're thinking about the revenue model for base rollup and you're thinking that it's the DAO that we're trying to like create revenue for but I think there's just too many players in the ecosystem uh to like fully focus on just the DAO or like whatever this the treasury that exists on a DAO on a on a base rollup. We're going to be thinking about them along with everybody else. And then also we're going to have to think about these other chains and rollups because they are actually a big driver of the revenue the income and expenditure for a rollup. So I have this the shark from Finding Nemo. Uh so depending on the way how you look at it it could be a friend or it could be a foe and there will there will be cases where they are they are both.

Uh Okay so setting the table uh so I guess maybe one of the reasons why why I got accepted today was because I basically when I when I submitted this talk we were in this we're around June or July and we were in a situation where Taiko was losing money by posting blocks to L1. Uh which people were focusing on and people were getting really scared like base rollups might not have a future. Uh what's happening here? Uh but there was a good reason for this. Taiko wanted to make sure that uh users were getting a reasonable UX that they were getting updates every 12 seconds because base rollups can only confirm blocks when they get posted to L1.

Now again, there's this concept of pre-confirmations, but they aren't live yet, but at the time, Taiko was force forcing uh L1 blocks onto the uh oh sorry, L2 blocks onto the L1 uh while they were only half full or less. So, they weren't actually generating enough fees to to sort of uh compensate for the cost of posting to L1. Uh so, but again, this was this was sort of the the P&L that grow the pie grow the grow the pie was generating. Right? And the the P&L that they were looking at was L2 gas fees collected minus the L1 fees that were paid to post data and proofs to the L1.

Uh so, that's a relatively simplified model uh of the the revenue the income minus expenditure for a base rollup. Uh there's actually a lot more components that we need to consider when we're thinking about is a is this base rollup uh viable to run. Uh so, if you want to get a bit bit deeper, Nethermind research did get a little bit deeper at one point and I guess we're slowly we're getting deeper uh uh but by the day. Um we had this sort of maybe a little more complex model where we're looking beyond the on-chain cost and looking at the sort of the the running cost of running a prover, for example. Uh like running a prover is not cheap.

Uh not only do you have to actually pay the electricity cost for running the prover, you actually have to buy the prover in the first place. Uh so, most people today aren't actually buying provers, they're outsourcing to AWS or or any of these internet cloud providers. Um so, there's there's a there's an additional cost that needs to be considered there. But there's also additional incomes, so I'm I'm going to get to those in a second. Um, so again, like I guess we're we're we're in a situation where maybe, okay, someone says your income your your income minus expenditure is bad.

So someone else says it's not actually that bad. And really everything that's everyone is looking at this in sort of a quite imprecise model. And we're not actually able to consider all of the all of the incomes and all of the the expenditures. And I guess you could say that we're shooting from the hip a little bit. Um, but these metrics do do provide us with some indication of how things are changing, right?

Uh, if they're getting bad, if they're getting really bad, or if this really is something unsustainable. And in this in that sort of early stages for Taiko, that wasn't an unsustainable model, but it is getting better. And that that's that's the key. Um, so again, setting the table, this is the fourth slide that's setting the table. I was struggling for titles.

Uh, so thanks to all the competition, uh, users can base users can migrate easily between rollups that have similar network effects. So network effects are a key part of of why why users will stay on a rollup uh, today or base rollup today. But base rollups sort of by the warm fuzzy feeling the base rollups give us is that they're actually very connected with the L1 and other base rollups. So this network effect is actually potentially strongest uh, or like shared between all base rollups. So the individual rollups don't actually have much network effect.

So base rollups So users on base rollups actually have ba- basically not much switching cost between base rollups. So in this setting, the the provider profits, so the providers I mentioned were those the the provers, the sequencers, and the DAO. In this setting, the hypo- the hypothesis that I'm that I'm claiming is that these profits of all these people must tend to zero because it's easy for another base rollup to come along, replicate all of the functionality that exists in that base rollup. Um, and and basically just remove some of those overheads that the previous one was that the profit side of the overheads that that base roll up was trying to generate. Um, but the zero profit paradigm is far away.

Uh, there's still a lot because of because of all this haziness, because of this wild west that we're in, uh, it's still possible that provers and sequencers can actually be making money. Um, but again, we this is going to be a sort of a long a long road and we're going to get there, uh, incrementally. Uh, and again, depending on how you look at the road, uh, like income income is tending towards expenditure. You can see that as a bad thing, um, but you can also see that as a good thing, right? Especially in the case of Taiko, uh, Taiko was when they initially launched where the expenditure was really, uh, greater than income.

Getting it close to income and and maybe even getting it through past the income, uh, is is definitely a good thing. Uh, so how you look at these things is is very important. Okay, so on this path to the end game, this zero profit end game, uh, there is going to be increases in income and there's going to be decreases in expenditure. The format is supposed to be decreasing in expenditure. Uh, it'll be on a theme with Kevin stock later on, uh, but not like HD video, so, uh, with music.

So, anyway, we have as we as we shed some of the the technical debt or shed some of the sort of inefficiencies of early stage base roll ups, we're going to decrease the expenditure, um, and we're also going to come up with new ways to increase the income. Um, and yeah, basically that that that's the summary of what the sentence at the bottom. Okay, so, uh, here's the list, right? So, there's no real model as such, right? It's just just a list.

Uh, you have to basically sum up all of these peak components that you have in your base roll up and check is the left hand side greater than the right hand side. And again, with there's like multiple players in the system, so when we're thinking about base roll up revenue, we we have to look at it by player by player. So, the users have this warm fuzzy UX or utility that's hard to quantify, but they pay like very quantifiable fees. Uh, so we just need to make sure that there is still some utility that these users are getting. And if they stop paying fees, then we probably have to question, are we charging too much or are we actually generating any benefits to the users?

Uh proposers. The proposers have relatively straightforward costs as well. Uh fee well, sorry, as well. They have they collect fees, uh they generate MEV or they collect it they execute MEV. Uh they also potentially collect issuance.

Issuance is a contentious debate uh like in terms of who's paying and who's receiving. Uh but L2s can or base rollups can create a token and uh and issue it to the to proposers. And they the costs are L1 posting and running costs. Provers, again, they can collect fees as well, issuance as well. Um but they have another cost which is the setup, which is quite a little bit more significant than a proposer uh because again, ZK provers are pretty significant pieces of machinery.

Uh the DAO uh the DAO sort of actually just have have a sort of unique income source, which is investors. Uh so when base rollup launches, uh you'll see that there might be a seed round or multiple seed rounds. Uh and and that revenue that that investment does pay for a lot of the upfront costs, which again, you see with the provers and the and potentially the the proposers. When you don't when you don't know what the fees are going to be, you do need some some money in the bank to sort of get things rolling. And and that's what sort of capital we're doing at the start, right?

And that makes sense. You need to get the users, make create the utility, make sure that the UX isn't too bad, and then you start start to generate the fees. And potentially the DAO can recuperate those for the those losses for the uh investors. Uh but the the DAO also has significant uh I I also mentioned proposal fee. Proposal fee is sort of a more more of a complex topic, but uh it is possible that the DAO can be collecting money from actually charging proposers on L1 for the right to be a proposer, but uh that's not going to be a focus today.

Uh they also have to pay it, the DAO is important though because the doubt does have to pay for things like R&D. We can't keep depending on Ethereum Foundation for for funding. We have to sort of go beyond that, especially when we're trying to create base rollups sort of parallel to the Ethereum ecosystem. So, some R&D money does need to be generated, and that comes out of the doubt. And there's also governance running costs of the doubt.

And then we have this all these all chains and rollups. So, what do they actually mean? So, when the these all rollups can uh compose with the base rollup, that could be seen as income, right? You're creating UX you're creating improved utility for users. Um and that can allow you to charge more fees.

But, similarly, they also create competition, and and with competition comes sort of tighter profit margins. Uh So, is income greater than expenditure? I already sort of touched on what why it should be, but in general, if it's not, you can pack your suitcase, and you can go to the some other ecosystem, some other rollup. And the proposers is an expectancy, yes, it should be, right? But, uh we can see that there is new technology coming along, and there's probably going to be more where proposers may not actually have be profitable in the short term, because they may commit to offering pre-confirmations, for example.

And if you offer a pre-confirmation too cheaply, it may be significantly more expensive to renege on that pre-confirmation than just posting it at a small loss on the L1. The prover in expectancy in long term, yes, but again, it may be a substantial cost to to pay for proving to uh proving hardware. So, the prover must invest and see that there is a substantial chance of of return. Uh and the doubt, again, I mentioned this investment model, long term is not sustainable, so they need to be to be investing in I guess sustainable infrastructure for for for these fee mechanisms, these proposer fees, these sequencer fees, and we're yet to see what that's going to look like. But, it's It's going to be very interesting if if the if the token can actually recuperate those costs.

Uh okay, so the the costs right now are relatively straightforward. Uh but our hope I could be considering the sort of the the type of profit margins for the Liko is that these will change, right? So I I quote this uh sort of ancient philosopher or alter ego philosopher uh who says base income and expenditure is always in flux. You can't step into the same base revenue model twice. Uh what that means is that depending on the day, like as obviously today everyone I when I go to the toilet I hear people talking about how what the what the markets are doing.

Like revenue is changing all the time, right? The fees are changing all the time, but also the technology changing all the time. So when we think about the base revenue model, we need to have a dynamic way of framework for thinking about these things. And that's what sort of be the focus next. Uh so here are some of the big big changes to base revenue that I expect to see in the future in in the near future.

We're going to see cheaper L1 costs. we're going to see proving becoming more efficient. We're going to see prover and builder markets maturing. So a lot of L2's base rollups will have run their own prover, but we are there's there's whispers about prover markets and and potentially also builder markets, right? So builders and L1 are generating generating revenue for the uh for the proposers, and that's probably going to happen as well on on base rollups as well.

So that's going to be an income source for proposers. Uh and shared everything, okay? So I mentioned this composition, but it's good it's definitely going to be a significant area for income and reduced expenditure. So with a share with with shared uh resources or shared like shared block space, shared proving, we're going to get bigger network effects. We're going to get better utilization of of of the resources that we're using.

Um but we may also have higher costs, right? So how are we going to balance these things is going to be very interesting. So let's go through each one in a little bit of detail and I'm running out of time. Okay, so cheaper L1 costs, right? So EIP-4844 was relatively straightforward.

You just have this sort of uh sort of temporary data that's being posted to L1. And it makes it cheaper for L2s to post uh data to L data to the L1. And then we're seeing proposals like PeerDAS. I don't really know what it is, uh but I hear it's going to reduce cost significantly. Uh but it's uses some pretty significant sampling or it's pretty pretty high-tech sampling uh procedures.

You can check out Vitalik's really recent blog post for more information. Uh Zero-knowledge proofs becoming more efficient. That is the best. Uh I I can't speak for for from personal experience with zero-knowledge proofs. I don't have any.

Uh but again, we're constantly seeing cheaper improved cheaper proof generation and cheaper proof verification. Uh we're also seeing increasingly bad naming terminology, but again, it looks like they're getting better. So, the the worse the names get, the better we're going to see these proofs uh becoming. And we're seeing also more and more investment more and more teams at the the deploying people to ZK. So, I I expect this to to continue.

Uh a market a market that I saw this morning, but again, it's it's been taken down. No one's trading on this thing. Uh um So, the prover and builder market maturing. I mentioned this already. Um so, Uncle Ben mentions that with great competition comes efficient pricing and innovation.

And again, that's just what we're seeing on L1 and we expect that to see that in L2 as well. Uh so, prover competition, users and sequencers have to pay less less fees. Um And builder competition, we're going to get better UX through more efficient block building and sequencers are going to get higher fees. Um and then shared everything, right? So, with shared sequencing comes more MEV, more users, and we're going to get better higher fees.

We're going to be able to charge higher fees, right? So, this is again increasing the income of of the the providers in our system. Uh Shared blobs. So, right now Taiko is is creating their own blobs. If If you can share your blobs, you can get approximately 100% blob blob utilization.

So, you're no longer paying a You're paying a fixed cost, but you're not getting 100% utilization. You're able to claw back some of the cost that you're paying. Um share proving. We're going to get economies of scale and cheaper L1 verification. So, again, these are all sort of things that you're trying to do as a provider to increase your bottom line.

That is to improve your bottom line. And composability. Uh So, less list complexity for proposers. You don't You no longer have to use someone like a cross. You can just basically uh sign a smart or sign a transaction that interacts with a a shared smart contract.

And pre-comps, that's that's a bit It's a big area of of efficiency and reduce the cost that we'd expect to see and improve UX that will will definitely improve the revenue model for base rollups. Uh but it's also going to increase expenses, right? Uh so, we're going to get There's There's definitely a great greater cost for sequencing two rollups together than sequencing them in the both of them independently. Uh just because of the increased search space. Uh the the more the greater amount of MEV MEV opportunities that exist.

Um share proving, again, similar similar concept. As far as I understand, if proving two states uh together uh is more is more expensive than proving two individually. Uh but the big question will be are there some of the benefits greater than the increase in proving cost? We're only going to see that uh when we see what the results of the of of the R&D are. Uh but we definitely are optimistic that we can get these ZK proving costs low and the the utility from the shared sequencing high.

And centralization, I guess that's sort of a that's that's sort of a tougher one to to uh quantify, but it's definitely a significant risk. And risks are affect Risks are costs in our revenue model. So, centralization poses a risk uh and we need to be conscious that all of these shared things will add to centralization risks. Uh So, conclusion, when you're thinking about the revenue model for your base roll-up, you want to you want to basically get total up all of your incomes, take away all of your expenditures, and ask yourself, "Why am I playing this if I'm losing? And how can I play more of this if I'm winning?"

Uh This is here as a sample. So, again, do your own calculations. Uh and innovate on increasing income and reducing expenses. So, remember Heraclitus' trick. You can't step into the same base revenue model twice.

So, we want to see less trickles uh I'm I'm I'm more waves crashing, all right? So, thanks very much. And uh yeah, pre- appreciate your time. All right, great talk, Connor. I'm just going to direct you to stand over here for the Q&A.

Yeah, this one will be good. All right, so we do have a few questions. Let's go from the most voted question. How can we reconcile revenue models with the decentralization of ZKP roll-ups, where multiple sequencers and validators will need to be compensated? Mhm.

Uh Did the person that wrote this question, could they explain what they mean by reconcile? Yeah. Oh, I I sorry. I I yeah. Can we get a Can we get a mic to that person?

Or So, okay. Uh I think maybe we go to the next question while we're waiting. Uh Well, I maybe maybe I can I can touch on my interpretation of the of the of the question. So, maybe when in the case of current current roll-ups where we have a centralized sequencer and centralized prover, it's it's still not clear if the incomes are greater than the expenditures. But okay, but yeah, we have the mic now.

Let's go. Uh essentially, how do we like is as there's a risk of um fees increasing with um higher number of uh validators uh in L2 ZK um rollups? How what what what what are the risks of that making the situation like more unsustainable? And what are ways in which we can address that? Well, the good thing about like Best Rollup specifically uh is that we we don't have any validators to pay, right?

It's just the L1 proposer. Uh now, in terms of the multiple provers that may exist, uh we we're basically going to be creating a market, right? And it's it's a bit like Bitcoin and the some of the old arguments for Bitcoin where you're like, "Okay, well, I have uh my hash rate and I have the expected Bitcoin rewards. And if it's if it's uh if I'm losing money there, I can go to Bitcoin Cash or I can go to the alternative uh blockchain. That's the same the same applies to any of these prover markets.

So, if it becomes too expensive, users user will stop uh user will stop paying for it. And basically, you're going to have to spread out your spread out your costs or just wait until uh things become cheap again. Uh so, there's definitely a risk. Um in a fully decentralized system, yes, we're going to be exposed to these things being too expensive. Uh so, the only thing we can do is try and make sure that we have uh fallbacks, like maybe cheaper fallbacks in the case that that those things we are exposed very exposed to fluctuations.

Uh but I think pre-confirmations and things like this that allow us to wait maybe 16 blocks or or 32 32 blocks L1 blocks uh where we get to sort of pick and choose where we include things and maybe pick and choose which provers to pay. Uh that can maybe help to reduce the risk to maybe spikes, let's say. Um but I'm not sure if that answered your question. Sorry. Great.

Uh we do have around 2 minutes for another question. Um, which one would you like to take, Connor? Um, well, I'll go for the shorter one first. Uh, if base rollups have a sustainable and positive revenue, can it become a de facto design for the layer one? Uh, well, I I guess so, right?

Like at some point if if there is one base rollup that is taking up all of the all of the L1 block space, uh, like isn't that a that that is a theoretical possibility with uh, base rollups? Uh, that all the computation is being done like off-chain in on the base rollup. Uh, so in that sense, it could be doing all of the execution for the for the L1, but I'm not sure if it can actually replace the L1 because it is it is base still being sequenced by the L1. Uh, so, yeah. All right, do you want to answer another another one?

Okay, uh, I'll go for the short one. Uh, just uh, what is the plan to increase the income but with the scale of the and expenditure uh, with scale with scale the expenditure also increases. Uh, yeah, that is a tricky one. So, it's like again, we have to take a bet, right? Um, and we're lucky in the fact that a lot of people like flashbots, uh, like all of the ZK teams that I mentioned, they're plowing ahead with improved mechanism design or improved proving systems, improved mechanism design regardless of, let's say, like any specific end game that that may that may be likely.

Uh, they're they're trying to get make things cheaper. Um, and if things do become incredibly expensive with scale, we can always scale back, right? We I mean, there will there will definitely be a limit to the amount of things that we can merge, the amount of state that we can merge. Uh, but right now, we can see that single rollups can survive. And then it's the case of how can we actually just scale up that the merging of all this state, and that's definitely going to be a really interesting question.

But, we can always scale back. That's that's the key. All right, cool. So, we're right on time. Thank you so much, Connor.