SVM L2 on Ethereum: Bullish or Bearish SOL? with Neel Somani, Founder, Eclipse

Solana on Ethereum? That is the question we explore on today's episode with a new product release as well. This is an opportunity to pick into a topic we haven't yet touched on on Ethereum, which is a virtual machine that Solana has called the SVM. That's the Solana virtual machine. So proponents have said for a long time that it's much better than the Ethereum EVM. It has multi threaded execution, lots of bells and whistles that Ethereum doesn't have. And now on today's episode, we are finding out that it is coming to Ethereum in the form of a layer two.

The Eclipse mainnet is what it's called. That is the layer two that is launching today with the SVM embedded. And Eclipse is not only a new layer two on the scene, but it's an entire framework, kind of similar to the Optimism Super Chain network as well. So this poses a lot of interesting questions, as you might imagine, that we are going to dig into today, including who wins? Is this a is this a W for Ethereum? Is this an L for Solana or maybe the reverse? What does this mean for the future of our industry? A lot to unpack on today's episode.

I would even say that a lot of Ethereum people will say that yeah, the SVM is a fantastic piece of technology and it is better than the EVM, especially when it comes to the things that virtual machines do, which is execution. There are some massive questions that this brings to the table, and I'm about to ask them. But first, I want to talk about our friends and sponsors over at

There are there are some big questions on today's episodes. And so we have the uh the founder of the Eclipse Layer 2. His name is Neil Simani. Uh, and he's going to be jumping on the podcast in a minute here. But David, we're going to be unpacking four different protocols. All right. So it's not just Ethereum and Solana. There are a few others as well that are woven into this uh layer two. So what what are listeners in for? What are the big questions to prepare us?

Yeah, the question doesn't stop at is this Ethereum versus Solana, Ethereum or Solana? Celestia and Risk Zero are also relevant here.

And I think one of the questions is like the big question, if we are investors in the space trying to invest in the future, we want to ask the question who wins? What are these components doing together? And does some of these components win more than others? What does it mean to win? And or am I just using this lens of winning versus losing? Because that's my lens for all blockchain systems. And is that even the right lens? So the first question I think we're gonna get at is Does the Solana VM as an Ethereum layer two, is that a bigger win for Ether than it is for Seoul?

the Solana fanboys in my mentions are convinced that this is the Ethereum protocol moving closer to Solana, but the layer two centric take is that all good execution tech will eventually settle on Ethereum liquidity and security.

But then what the hell does Celestia have to do with this equation? And what's Risk Zero doing there? So this is not just a question about like the tug of war between these two cooperating competition competing protocols, but what does this mean for the modular thesis at large? So some very big questions. And I think I listened to Neil Samani over at the Modular Summit back in ECC and some of some of his talks around the space. And I think he's got some of the answers. So we're going to ask all of these questions to Neil here in a second. But first, a moment to talk about some of these fantastic sponsors that make the show possible, especially Bankless Nation. I would love to introduce you to Neil Samani, the founder of Eclipse, a project working to bring the Solana virtual machine, the SDM, to Ethereum.

Today, Eclipse has announced their SVM Mainnet, the first Eclipse layer two on Ethereum, that uses the SVM as its execution engine, but Ethereum for settlement and liquidity.

Is that is that a curveball? Well, it doesn't stop there because Eclipse is also using Celestia for data availability and risk zero for fraud proofs.

So, Solana's execution environment, settling on Ethereum,

with ETH as its gas, using Celestia for data availability, and then security offered by Risk Zero fraud proofs. My first big question to you, Neil.

Is what the hell is that?

What is this? Who even allowed you to put it on? What did you create? What did you create? What is this?

I love this image. This is amazing.

Uh for the both listeners out there, this is the uh what the hell is this family guy meme for the for the people listening on the podcast?

It's like Noah on the Ark and there's some kind of hybrid animal like a giraffe, elephant, like well, I don't know what that tail that is, but some kind of hybrid animal that's been created. Like the platypus.

Yeah. Uh huh.

That is awesome.

Uh, David, Ryan, thank you for having me.

Yeah. So, Neil, let's let's uh where where did this idea to uh put four different networks together come about? And we'll maybe talk about some of the motivations here.

So the original idea was just to put Solano on Ethereum. And that was the motivation behind Equipes. And what we ran into was just a ton of constraints and

things that you'd expect to be true for a virtual machine because that's how it is in the Ethereum world. So an example is like chain ID. When you switch your MetaMask wallet to another Ethereum chain or another EVM chain, then they actually have like a well-defined mechanism for doing that. Another example is there's no global Merkle tree for Solana.

So the lack of these primitives means that that

initial idea was not so easy to implement. So we basically implement we had to add in Celestia, risk zero out of necessity in order to make this possible.

Wait, so you had to add Celestia and and Rist Zero? Like it wasn't an option to just do a layer two

Just pure Ethereum. I think so. It depends on the amount of transaction volume that we end up running. But at our projected amount of transactions,

a Ethereum DA would just prove to be very expensive. And it would also lack a lot of the benefits of the Solana VM, which is that

you're going for scale, meaning that you want the transactions to be really cheap.

So right now it's like the base cost for writing 200 bytes to Ethereum is about 15 cents. So that would that would be like much more expensive transactions. And that opens us up to a dunk from the Solana community, which is like, oh, we're much cheaper than you. By doing it this way, then we can actually be competitive on price too.

I I want to get back to to why Solana, because that's the big question. Why the SVM, right? But before we do, just really quick, yes, data availability is very expensive in today's world. However, there is some hope. We just did release an episode uh earlier this week with um with Dom from ETH Research on um blob space and EIP4844, where the cost of uh data availability for roll ups is going to drop uh quite significantly. Or maybe let me just say rephrase that and say the availability of data availability is going to be uh increase massively through the in the form of these blobs. Does that change the calculus at all for you, or is it kind of the same? It's still more expensive than something like Celestia.

It definitely changes the calculus. Yeah, given that it's a fee market, we have to see where that fee ends up landing. So I want it to be live, but yeah, there's great research going on in the Ethereum community for scaling DA. We're watching it closely, and we have ambitions to eventually just be fully on Ethereum and just use that for DA as well.

Okay, well let's talk about the the big thing here. So the Solana virtual machine, the SVM. We've not done an episode comparing the SVM versus the EVM. Um we've been thinking about doing one. This maybe gives us a little taste of that. Can you tell us what is what is so great about the SVM? Why do people um some people you know sing its praises and seem to prefer it for certain things?

The way to think about it is uh by starting with the EVM and just understanding the failures of the EVM so far in scaling Ethereum. And I think that if you ask a lot of the researchers in the Ethereum community, scale is pretty much the biggest issue.

But and it feels kind of redundant for me to be saying that because everyone's talking about it.

Yet the way that they've tried to bring scale to Ethereum is just by taking the EVM and turning into a rollup effectively. But what's nice about a rollup is that once you've decoupled execution, consensus, settlement, and DA.

Is that you can make that execution layer whatever you want.

So the issue with the EVM is that it's single threaded as it exists right now, meaning that all the transactions get in a single file line. They're executed one at a time.

And that means that if there's a big NFT draw, for example, then that's gonna really spam the network. And there's no way to get your DeFi transaction in unless you're competing with all those other transactions that are in line.

But what's nice about a parallelized virtual machine is that you can execute those transactions concurrently, given that they're not touching the same piece of state.

So that's what the Solana virtual machine primarily brings. And it's the most battle-tested VM in the sense that Solana's been around, as opposed to Move or some other parallelized virtual machines, which are much newer. And that means that you also get the benefits of the existing Solana code base, any existing libraries that are written for the SVM. And we can take all of those and immediately port it to Eclipse. So that's the main reason to use the SVM, which is just scale. And scale is an inherently maximalist take for a roll-up to claim because we're basically saying that

if you have one SVM, you don't need all these other single-threaded VMs around it.

Yeah, so one of the um reasonings I've heard around the EVM is that while the EVM does execute, because that's what virtual machines do, it was not built for execution. It was built for the Ethereum layer one, uh, and that has prioritized other properties. Uh, and while the EVM has had a ton of just network effects, the open source network effects around the EVM are super, super strong. I think in this day and age, in the year 2023, uh, one of the things we're seeing is that the network effects around the SVM are growing sufficient tailwinds that, Neil, I think what you're saying is that, well, I can feel safe that there is sufficient momentum in the network effects of the SVM code base that is going to be an alive code base moving forward into the future. And so it is going to be the execution virtual machine that we can deploy to Ethereum because if we have a roll up centric roadmap, we aren't beholden to the EVM.

on layer twos. Is it this is my interpretation. Is this about right?

Yeah, that's right. And what's cool is that it's it actually goes beyond just the existence of the SVM per se. It's a bit of a misnomer because the SVM is actually the Berkeley packet filter virtual machine. And that's been in the Linux kernel for like decades. So this is a very old virtual machine, and it has RO support, Rust can compile to the bytecode. And there's a few surrounding extensions that are pretty battle-tested as well. So we're taking all this existing wealth of open source code and just repurposing it as an Ethereum L2.

How much better is it, Neil?

Like it ca how can you c

I would argue it's like orders of magnitude, like a hundred a hundred times or something, some ridiculous magnitude or faster. Yeah.

okay, so like if if we take a a roll up with the exact same stack you're talking about, but um rather than the S V M we have the E V M versus what you're doing with the S V M, how much better is yours? You're saying an order of magnitude? Like tens of the city.

Yeah, it's always tough to talk about throughput because the reality is it depends on what are the state access patterns. Like if everything, if let's say every transaction, the SVM was accessing the same piece of state, that then you actually can't beat sequential, because that that's just like a fact about databases, right? You'd have to impose if it was sequential rights, let's say. So then you have to lock the trait, you have to lock that piece of state to execute the transaction, then to do the next one, then do the next one. So you're you're not saving anything in the sequential case, but assuming they're all touching different pieces of state, which if the crypto thesis is right, we're gonna have all kinds of applications doing all sorts of different things. And that's effectively what these EVM rollups are mimicking. If you look at OP stack, the reason why people are deploying these is a lot of time they cite dedicated block space. But that's exactly what the SVM already solves for. So our thinking is why fragment liquidity, fragment the user experience, make them switch networks. It's looking like a mess already, but uh you can you can prevent all that by just keeping it all in the same chain.

Okay, so one of the I guess the bull case for the EVM is basically I don't think anyone really prefers it for its uh execution prowess and capabilities. They all prefer it because of the network effect. It's because we started with the EVM, a whole bunch of apps were built, and now we could just very easily port those apps to uh roll-ups. That's kind of why. And I guess with SVM, it is you're you're making the case that it is much uh like better for just kind of the execution layer type thing, and you're still taking advantage of some of the network effect that that Solana has has built as well. I I want to ask a question here because I'm not sure I fully understand this, but about like kind of different fee markets and and resourcing. Um how does how does that work? So is it the case inside of your roll up that if there is a big NFT drop or some sort of app that is consuming a uh a large quantity of the block space inside of uh Eclipse? Um, is that somehow isolated or segmented, or is there some kind of um bifurcation of resources between another layer?

Localized fee market is

Yeah, this is this is a term that I I've not fully understood, but I know um our friends uh on the Solana side really purport this as one of uh main virtue of actually uh Solana is like these localized fee markets. Tell us about that.

Yeah, it's critical. And if you've looked at Arbitrum lately,

Fees still spike substantially on rollups if they're single threaded. And that's because of that exact reason that you're mentioning the global fee market. If GMX gets a bunch of activity, then everyone suffers as a result.

But economically, that's not actually really the way that it should be because there's this negative externality to GMX being imposed on all of the other apps. But really, it should be constrained to that one app. So it's as if you took all these global fees and you just concentrated it on the one app that's actually causing on the hotspot or a state hotspot, as some of the folks on the Salana side will describe it as.

Uh and you just uh everyone else can just keep like merrily happily uh walking along and they can execute on other cores. So it's it's really a property of the scheduler within the validator, because in every validator they have to somehow decide what order am I gonna process transactions. And they're saying everything that's accessing this piece of state is gonna all go on a single core, and everyone else can use the other cores, and and there's no contention on this.

So Neil, this is really just a statement. Uh uh the design philosophy of Eclipse is just a statement that it's a bullish statement on the growth of the SVM network effects. So maybe in this day and age, if you are like Optimism or Arbitrum, you're going for this thing called Ethereum equivalence. Just and they are making a statement that, hey, we are bullish on the growth of the EVM network effects. And the, I mean, network effects of different virtual machines can grow independently. It's not an either or. But what your the philosophy of Eclipse is saying, like, hey, there's the SVM network effects are going to grow. They're going to grow a lot. We want to take that and couple into whatever the value is of being a part of the Ethereum ecosystem. That's my like synopsis synopsis of this.

Yeah, I'd even argue that EVM network effects have proven to be not so important.

And I think it made a lot of sense initially a few years ago when you had Polygon and BSC launching. But if you look at the types of apps that were deployed there, you get like SushiSwap, which is effectively a fork of Uniswap. And now I recognize they've deployed even to like Ethereum. But those are the types of apps you often get.