Are L2s Still Scaling Ethereum? | Ethereum Korea One: Genesis

Transcript

Opening and panelist introductions

I'm Harry, engineer at Sunnyside Labs, and I'm stoked to moderate this session called Are L2s Still Scaling Ethereum? We have three people here who are living the frontier of L2s. So we'll be discussing several interesting points around L2s today with builders out there. So let's start with a brief intro. Ilia, could you start?

I'm product lead at StarkWare. I used to be in charge of the protocol, then I went to work on Bitcoin, now I'm working on the financial. Hey everybody, I'm Ed. I'm the co-founder and chief scientist at Offchain. and that means that I lead the research team at Offchain, which does a lot of research to advance our company. And I'm Chris Andreola, and I support growth at Optimism's development. Yeah, thanks for the intro. So we'll start with a kind of like spicy point that Vitalik has said earlier this year.

Vitalik's critique: are L2s still scaling Ethereum?

So he argued that the original picture of L2s as the branded shards of Ethereum is not working in reality. Since L1 scaled faster than expected, L2 decentralization and interoperability have progressed more slowly. Let's talk about each of our take on this. So, I think Starknet matches the view direction suggested by.

So, what's your take on that point? What's your feeling? Yeah, so I assume the reference is to Vitalik's choice of adopting a new VM that is very efficiently provable and verifiable. And we get this in Starknet from the get-go. I think it's a good choice, unfortunately the Ethereum Foundation has elected not to use Cairo, which is the VM we built. Personally I think it's not the correct choice but it's understandable. And as far as the broader question, I think it makes sense for Ethereum to move to a newer architecture that's efficient to prove because if Ethereum wants to scale then we think proofs are necessary.

How about Ed? So I think if you look at the recent history and you look at the state right now it's clear that L2s provide much more scale than L1 Ethereum if you just look at things like gas count or transaction count. L2s collectively do a lot more traffic than Ethereum, and I think you can fall into a trap of comparing the current state of one thing against the hoped-for future state of another thing. The hoped-for future state of Ethereum is to do more traffic than L2s are doing now. But if you compare now, or you compare L2s' hoped-for future state against Ethereum's hoped-for future state, either way, L2s are

a big source of scaling. So I think L2s are, these days, a necessary part of the Ethereum stack, and in fact support most of the activity on the Ethereum stack. And that's as it should be, because I think layering is a great way to design protocols. That's why the internet succeeded. And I think that Ethereum is succeeding in part because it's using this layering approach. I don't think that's going to change. Okay. Awesome.

Let me start by seconding the comment. Yes, it's true today, L2s have done far more than what the scale of Ethereum would. But second to that, I think as you look forward, I think the L2s will continue to scale Ethereum far more than the base layer itself. And third, when you think about the decentralization for the L2s today, as you look forward over the next year or two years, I think that we get to see not only L2s' progress to stage 2 but second to that have the capabilities of decentralized sequencers today.

Is Stage 2 still the goal?

Okay, thanks for the answers. So I think as Chris mentioned about stage 2, we can talk more about that. I think Vitalik's diagnosis had two halves, so there was like L1 is getting faster, but we have another branch of L2 decentralization. I want to ask, as an L2 user, do you still have plans to move to stage 2?

Or is it becoming optional or something? So, how do you think about it? We've had decentralization plans for a while. I'm not sure that being stage 2 is at the core of it. We just want good properties. They can be divergent with L2BEAT things. I don't even think decentralization is necessary in the frame of being an L2, it's just in the frame of being a network in itself. It's good to have censorship resistance and extra security, it's good to be able to depend on one entity for a sequencer, and we don't think of the framing as just something on top of Ethereum.

Just a proper way of approaching it. I would second that. I don't think that the L1 versus L2 distinction captures what's actually important about decentralization or appropriate governance of an L2. If you look carefully at the definition, you can see, for example, that Ethereum would not meet the standard for L1 today. And it's hard to see how it could, because the stage two, what it requires is that the protocol itself prevents governance-based changes that happen faster than 30 days, and there's nothing in Ethereum that does that.

What is important is how governance works to avoid the centralized points of control and have governance that is accountable and is fair to users. Those are the goals that are important. Ethereum does that well. And I think all of the L2s are moving toward those goals. So I think like StarkWare, we don't get too hung up on L2BEAT-specific opinions about this. We think it's more important to look to the merits. And I think we see all of the L2s striving in that direction.

Okay, that's an interesting take. The interview is not. Stage two. What are the definitions? It's not. How about you, Chris? I'll try to answer your question directly. I'll give a few anecdotal examples. We have several partners today that are pursuing stage two decentralization under the L2BEAT definitions. Why are they doing that? They're actually doing it not for the pure sake of decentralization. They're doing it out of fear that the government might come and ask them to make certain actions on chain, for example, upgrade the chain, and they need the defensible position that they purely cannot.

My second one really is regarding the core principles of decentralization not necessarily aligning to the definitions that L2BEAT shares with the world. That also makes sense. So it's becoming more practical, as I feel. So let me change the question a bit then. So stage two is, I feel like it's related to the security of the rollups.

Rethinking the stage framework and governance sovereignty

So and I also saw that Arbitrum was also doing multiproofs for better security. Do you guys feel like the stage framework has, like, evaluating the role of chains, or if so, how should we change it? I mean we can have a whole one-hour panel about this one. But I think there'd actually be a lot of agreement among us about what is truly important.

Right? Decentralization is about who holds the power. It's about how governance works. Who has the power in the governance. And that is ultimately the key. I think one of the real advantages of L2s is that each L2 can have governance sovereignty, meaning that the community of that L2 can decide for themselves what they want on their L2. And because there are many L2s, and in some cases many L2s even built on the same technology stack, you can have customization and you can see each L2 doing what it needs to do.

That governance sovereignty, I think, is a really important feature of L2s, which you would not get with this sort of more monolithic design. Okay, that makes sense. Like, Ilia, do you have anything to say? Yeah, slightly more cynical statement. The fantasy of distilling governance of any group into code is just a fantasy. It's never been achieved anywhere in any human society, in any group. It's not going to start now. I think we should stop at our focusing on the technology.

One point to acknowledge is that we're speaking about L2s generally here today. There's many L2s that are out in the market now that would be disadvantaged by moving towards any definition. Just as one example, many teams want to customize their chains, roll out new features, test things in production, and those projects should be pushed forward if they have merit. And I don't think any of us would think negatively about those changes because they cannot pursue pure decentralization today.

I think customization is one of the important arguments for governance, that you can have the customizations you need. That said, testing in production, that is not a best practice. Just saying. I think we can talk about another subject this time. So like, it's been a there's a there's always been an argument about, criticism I think about L2s

Blob fees and what Ethereum gets from L2s

that people, some people always say that L2s capture most of the value from their activity but pay Ethereum relatively little in blob fees. So, Base has been heavily criticized about this, and now Robinhood is being criticized by some guys on crypto Twitter. I think that's by design, but it still also means that the L2 to Ethereum relationship has almost no economic content at the protocol level.

I want to ask about the relationship of L2s and Ethereum. For example, let's suppose that your chain, in five years, is hugely successful. You have a booming economy and billions of dollars in fees. In that world, what will Ethereum get from your work? Is it blob revenue or demand for ETH as an asset? Ilia, do you have any take on this?

I think many, many things could be done to couple the demand on L2s to the success of Ethereum as an L1. But I want to say a few things before. First of all, I think the current discrepancy is completely by design. Ethereum was completely aware that L2s want to mitigate their fixed costs and also their users' marginal costs. That was the whole problem that blobs were aiming to solve. So I think it's sort of holding the rope for both sides to expect it to return value and also to try and minimize it. That's the first thing.

And the second thing is that I don't think financially Ethereum prioritized for this. I think people wanted to create demand for crypto, Ethereum had incredibly high cost for transacting. The solution was L2s. Now, looking at the state as it is now, I understand Ethereum wants to maybe derive more value. And a couple of things, one of the ways would be to couple the demand for ETH by some sort of buyback mechanism. Many applications are doing such things, probably Hyperliquid is the most famous example. So one could drive such things. Another thing would be to couple the staking in the upstairs and downstairs networks. So I think plenty can be done, but I think it's unfair to criticize

L2s for not having repaid any of their value yet. Yeah, I think one of the things that L2s have done to the economics of this whole space is to decrease the fees that users pay, like a huge amount. I think it's a fallacy that major L2s are enormously profitable or bring in enormous fees compared to Ethereum. That said, Ethereum should charge a fair price for the services that L2s are using. And I think it has been the case at some times that Ethereum was underpricing the data

availability that L2s were using. I think we're closer to a better price as far as that goes. But I think pricing services is a good way to do it. That's as you suggested, the demand for ETH from L2s is significant. Arbitrum stack chains have locked in their escrow contracts now close to 1% of the ETH supply, over a million ETH.

And so if you imagine the world where L2s collectively are 20x bigger, that means 20% of ETH being locked in these chains. That's a really significant impact on the Ethereum ecosystem. So those things do grow together and I think that will continue. I think the most important thing that L2s have done in this respect is to use ETH as their native token and gas token, because that is a big part of what drives demand for ETH as a core asset in those chains.

I second what you shared earlier, and to take a slightly different angle and kind of piggybacking off what you said before, Ed. If you were to scale to billions of dollars, billions of transactions and billions of dollars in revenue that you shared, you're really leading the distribution of the corporates and teams such as Coinbase, Robinhood, and many, many others, and their ability to capture client share users at their distribution. Distribution.

ETH gains a huge advantage by the go-to-market muscle of all of the teams building on L2s. And you can see that Ed mentioned the value of it today in the context of Arbitrum. But as you scale that across many L2s, an increasing number over that time period, I think it's safe to say that ETH will benefit by the number of holders, the number of users, and in a timely sense, the number of people aligned to the Ethereum vision. That makes sense.

What will look wrong five years from now

So at this point I feel like things have changed a lot. Personally I've been to EthCC two years ago, I think, and I watched a session by Cartesi, I think, I think there was Ed and Mark from Optimism and there was Lil No and Justin Drake, and we talked about proof systems and the stage 2 and everything, but I think in two years it changes a bit. It's not a bad point but things have changed, literally, literally.

So I want to ask that in five years, when we look back at 2026, what will be the one thing that the current L2 industry treats as obvious today but will turn out to have been wrong? It could be technical, economic or anything. Yeah, I think the market is, the industry is beginning to realize this, but the emphasis at the beginning was much more security-focused and maybe even ideological.

It was much more adjacent to the cypherpunk thesis, and I think what we see, or at least what I feel, is that users care about products. The chains themselves are almost inconsequential. So I don't think when people use L2s they dive into the security workings. I don't think many people know where most of the live. I think they care about products and the user experience. So I think if we were to have gone back and rebuilt things, we would have probably focused on that first.

And then people would have matured to want security for their assets. That's my take. I think looking back five years from now we will say that people back then in 2026 didn't understand the importance of user experience in the design of these technologies. The way that the design from top to bottom needs to reason backward from user experience.

Things like security and scale, those need to be there. But beyond that, in order to really achieve success and reach all those users, we need to figure out what it is we need to build at the hard technical level to enable the user experience that people need to see. One of the things, for example, that L2s have done well so far is to reduce latency compared to L1. Right, and that's been a huge user experience unlock and has pushed L1 to try to get some of that latency reduction as well. I think that's just the tip of the iceberg in terms of the ways that user experience will increasingly be driving everything we do up and down through the stack, not that we're ignoring the other important goals, but that once

we have locked down the security, once we have the governance as well as we can do it with humans, then I think user experience will become the driver even at the deep tech level. Go ahead. Great answers, though, really. Thinking for five years in terms of what we look back and say, potentially was undervalued, which should have been obvious today. If you zoom out to the broader industry, we look around, we look at teams like Canton, Tempo, Arc, all of the others, and many teams, if you're on the

front lines speaking with the world's institutions, are lured in by the attractiveness of their net new technological models or their incentives or the list of institutional names that are adopting their stack. I think they'll look back and say it's pretty obvious that Ethereum persists through this. Me personally, I've traveled around the world speaking with the largest institutions, whether they're banks or fintechs or payments companies. They're struggling with that question today, but really what aligns us all at the core is our belief in Ethereum, and as you watch many of Ethereum's competitors in some way stumble, make missteps, overreach, I continue to see the merits

of Ethereum through, and I think it's becoming increasingly obvious now as we see some of these teams make those missteps. Yeah, just a minor anecdote. So the previous panel discussed account abstraction from a technical point of view. And in Starknet, we went for it from the get-go. And it was not this unbelievable unlock that caused floods of users. It was a really good example of wanting to build something for the right technical reasons.

And it made some things really nice, like having multi-sig accounts natively. But it did not bring a flood of users who do a Face ID and start using crypto. That did not happen. Products were missing. Thanks for the answer. And on that part, we can also talk about different aspects of L2s. So we've talked about, talked that like, L2s are like, the scalability of L2s, like, and it will be a good UX and it can drive more usage of ETH as the Ethereum

What each L2 uniquely offers

ecosystem, but I think there are several points that each L2 can provide to the users that Ethereum L1 and other L2s cannot. I think there will be different points of each of you. So could you introduce one single point that L1 and others cannot provide but yours can provide to the users?

I'm happy to kick off and take my moment to. One thing we're trying to do at Optimism is meet the institutions where they are today. Over the longest horizon, we, Optimism wants decentralization. However, as you're engaging with teams globally, they don't necessarily seem to make a merit of it. We are trying to meet teams where they are, help carry, bring them on board, get them on into the Ethereum ecosystem broadly, but help them graduate to our ideas over time rather than forcing them on them today.

Amen. I guess as I said before, I think L2s have been designed from the beginning to try to drive down latency and drive up scale compared to L1 designs. And I think with a lot of success. And both of those things are pretty significant unlocks. It does mean that L2s use fundamentally different design approaches than L1. And that's one of the challenges that we see. We saw it a little bit in the account abstraction discussion before, because we're trying to remain within a standards-based system, and yet using, in some respects, fundamentally different designs.

I think we're going to continue to work with that. But I think one of the things that L2s can do is simply provide customization, different solutions, different approaches, different governance for different use cases. And that's really important, right? And again, I go back to this analogy of the internet. You have the basic TCP/IP level, and on top of that, you know, a million flowers have bloomed of different sorts, different applications, different kinds of services. The internet did not try to force everyone into the same model, but instead had this idea of a solid core at the base that's in common, and some really important standards and practices, but that at the same time,

different applications, different designs for different uses. That's something that L2s do well, not only because there are multiple L2 stacks but also because individual stacks can do a lot to meet the needs of particular use cases, particular customers. We've really focused a lot on that in Arbitrum, and I think it's a core advantage that all of us have. Well said. Nothing to add. Thanks for all the answers and sharing your views on L2s and the relationship with Ethereum.

Q&A: corporate L1s and privacy

So I think we have some time for the questions. So if any, please show your hands. So my question is, three years, or four years, during this Layer 2, this presentation, now we're seeing, on the one hand, there's a lot of really big companies like Coinbase, but also we have Circle, and Stripe.

They chose to launch their own alternative L1. So I have two questions. What does that tell us? Are they just confused? Or they're not our ideal customer to begin with because they don't align with the decentralization value? That we're not doing it right? Or there is a movement? The second question is, what makes us confident that the L2 ecosystem will persist in the end? And I agree, and it's not here, but what's the answer?

Is it network effects or decentralization or something else? You want to start? In my opinion, we're seeing that crypto is broader than people who believe in the cypherpunk ethos. And I think a lot of these big companies that are going in the L1 direction are using crypto as a technology that has properties that are unrelated to decentralization. Maybe it's building on top of the existing platform easier, maybe it's easier for people to create derivatives on existing assets, etc. And so I'm just not convinced everything has to boil down to decentralization and holding our own assets.

I think we just see people who use it for different purposes. I think even if we set aside concerns of decentralization and progress, which we should not do, even if we were to set that aside, I think there are really important economic advantages to building in terms of L2s on top of a common L1 layer. And the reason is that L1s rely either on tightly permissioned validation, which has real issues with respect to power and reliability, both in the political sense and reliability, or they rely like Ethereum on a very large stake

consensus, and very large stake consensus system is the gold standard for security and decentralization, but it's incredibly expensive. You have to pay out huge rewards to people to put those stakes. That's a very expensive thing to do, and so it makes sense to do that once, and to do it in a way that is as decentralized and large-scale and well -governed as you can. That's Ethereum. And then on top of that, you can build any number of other blockchains.

So, I mean, my challenge to those companies, if they were to come to me, is to say, why are you paying this cost when you could be an Ethereum L2? L2s are now proven technologies. And, you know, I think it is because of a desire to have more control than they would have if they were in the Ethereum ecosystem. You know, which is a choice. But I don't think in the long run that that approach wins. I think the open, standards-based and more diverse approach of Ethereum is what's going to win in the long run, as has happened in most other important areas where people need to interoperate with technology.

I'll keep it very brief, but I think in short to me, whether it's the talks post- or February of last year or the continuing funding of these new L1s, I think it's fair to say the world sees an opportunity to change the playbook here in Ethos, and it's a call to action for all L2s to get back to what they were doing at its core and not get distracted. Thanks for the answer, and do you have any other questions?

Thank you for the good talk. I'd like to ask you all about your native privacy integration schedules. If you have any rough schedule for including, retail users can use the privacy natively on your network. If you have any schedule, please explain. Thing. So on Starknet, private Bitcoin is live and we'll open it to a few more assets. You can see it for yourself. These days I think these sorts of private applications exist mostly at the application level, right, and they're being deployed across a range of different blockchains.

It is doing privacy on a chain that is open and public and has the degree of transparency that people expect from technologies like ours is difficult to do, right? And so the way to reconcile that tends to be that you have a base layer that is transparent, transparent, it has transparent proving, and then people can build private things on top of that. It's a core part of the value proposition of these chains that you can inspect those parts of the state that are meant to be inspected, and that you can know more than just, oh yeah,

we're in a valid state, but we won't tell you what that state is. When you start to look at the use cases and ask what needs to be revealed, what needs to be kept private, I think this gets very complex, and these problems are much more difficult to solve in practice than in theory. That said, there are some really exciting use cases where people have found the way to sort of slice what's private from what is revealed in a way, and these are exciting to see, and I think they're going to continue to develop.

Optimism's general stance on privacy is to let the users or our partners decide what they believe the best construct is for them on top of our baseline. We are incredibly excited about many use cases and applications. Many of the privacy technologies we have arranged across our many chains. I'll give a call out to the Sunnyside team today. Their privacy solution, Privacy Boost, is live on three OP Stack chains today for very specific use cases.

But our general view is we've built the base layer, the permissionless, transparent, let the users and the applications decide what the appropriate privacy solution is for them. Yeah, thanks. And I think we have time for just one more question. Thanks a lot for the talk. It's actually very refreshing to hear all this. I want to bring this to a more practical level. So I think for the crypto itself right now, even after all these years, they were niche.

Q&A: adoption strategies

So compared to when we talked about L2s right now in Korea, as well as Arc, Tempo, etc. And so for L2s right now, each of your projects here, what's the overall adoption strategy and focus right now for you guys to come to the board? Or more whether it's an institution or it's more retail or? I guess I'll go first on this one. We think it's really important to have two things at the same time. One is to have a really active and robust public and open chain that, for us, is Arbitrum One,

that has liquidity, that has activity, and that is governed by a DAO. But at the same time, to have adoption by enterprises. Because enterprises can bring the distribution and the user interaction, the ability to integrate mainstream financial activities into the crypto space. Space. And we've seen this with Robinhood, we're seeing a lot of progress in the pipeline to do that.

I think in the long run though, I think you need to have both of those things, and that it's valuable to have them in a way that is compelling, so that people can do the same kinds of activities across both types of chains. But I think long run, the financial companies that have users will be key to unlocking, bringing the great masses of users into our technology. Two things. One of them is the open public chain that is DAO-governed, right, the other one is increasingly robust set of enterprise-driven chains that

have more integration with the consumer-facing TradFi ecosystem. That's where the users are, and to bring the front ends where the users are together with the back-end technologies that are crypto-based, and that are built in the style that we all build. That I think is the big unlock that can bring users onto blockchain rails and ease them into the future of finance. At Optimism, one of our core principles is do whatever it takes to make our partners successful.

We recognize we have core strengths and core weaknesses. We also recognize we have many large partners with huge customer bases globally. Globally. We seek to leverage our technology, the merits of our technology, to help our partners reach that distribution, whether it's an application on OP Mainnet, whether it's an import, whether it's a net new blockchain for a large global institution. Us on our own probably won't be able to capture the world's zeitgeist to bring all securities on chain, but potentially some of those partners can.

Yeah, I feel the same way. We need to leverage other people's distribution. We thought if we build it, they'll come, but they don't, so we need to bring them. Great answer. So yeah, I think we can conclude this discussion, and please give a round of applause to our panelists.