Talks

Scaling Etherlink Without Compromise

July 12, 2025

Scaling Etherlink Without Compromise

Important

The content of this article comes directly from a talk I gave at TezDev 2025 (Catalyst) on July 3, 2025. This is a personal transcript, and any mistakes in it are my own.

Good morning everyone! Today I’ll be talking about Etherlink—and more specifically about how we are scaling Etherlink without making any compromise. And before I dive too deep into the details of this talk, I think it is always a good idea to start from the beginning. That is, what is Etherlink, really?

What is Etherlink?

First and foremost, Etherlink is the EVM-compatible layer of Tezos. We have implemented this layer as a Layer 2 blockchain powered by a Tezos smart rollup, but that’s an implementation detail. What matters is what we wanted to achieve with Etherlink: bringing EVM-compatibility to the Tezos ecosystem, at a time where it could make a difference.

Now, we still ended up implementing a Layer 2 blockchain no matter our main motivation—and one we want to be attractive. But as an EthCC speaker put it earlier this week, there is a new L2 chain being launched every day, which means we need to bring new things to the table that our competitors don’t, if we want to make Etherlink a success.

This is why we have built Etherlink according to three core values, summed up in a very simple motto: fast, fair, and (nearly) free.

Etherlink is fast. Its sequencer is creating new blocks at a competitive pace (every 500ms), providing low-latency to the chain’s users. You only have to wait 250ms on average for your transaction to be included, which enables exciting opportunities (typically in DeFi).

Etherlink is (nearly) free. As of today, it only costs around a tenth of a cent to make a tez transfer on Etherlink. This is necessary for Etherlink to be a welcoming platform for a wide range of use cases including some requiring the injection of hundreds of thousands of transactions every day. Paying 20 cents once is okay, but with high volumes, every cent counts.
Etherlink is fair. In this area, we probably are pioneers actually. What we mean by fair is that Etherlink treats all its users equally, by being a permissionless platform where no arbitrary set of third-party entities holds dedicated rights. There is no committee, no chain owner capable of doing executive decision. On the contrary, we have entrusted the governance of Etherlink to Tezos bakers, making it the first non-custodial rollup. Anybody can become a baker, which means anybody can engage with Etherlink governance .

Etherlink in July 2025

We have been building Etherlink for a while now, and it turns out that July is a pretty busy month for us. In 2023, we announced Etherlink publicly at EthCC and TezDev and launched the Testnet . One year later, we had launched our Mainnet and it had been non-custodial from day 1. It was of significant importance for us that as an Etherlink user, you didn’t need to trust us in gradually decentralizing a custodial solution. If you are familiar with the Ethereum rollup space, you already know that this approach is actually quite singular.

Now it’s July 2025, and we have once again exciting news to share. This is especially true if you are a Tezos baker, since July will be a voting month for you with two key votes for Etherlink’s futureOh, and we have also announced Seoul , Tezos’ 19th protocol upgrade proposal since then. !. One in particular—and I mention it because it is actually a good transition for the topic of this talk—is about a new proposal upgrade notably changing how the gas price is computed to align it better with Etherlink’s real capacityTurns out that we were a bit conservative in Dionysus. When testing more thoroughly, we noticed the chain was behaving better under load than what we previously thought. .

That being said, I would not say that we are happy with the capacity of Etherlink today. It’s enough for the current activity of the chain, as hinted by how stable the gas price is nowadays. But it’s not enough for the kind of ambitions we are having for Etherlink in particular and Tezos as a whole.

This is why we are committed to increase the capacity of Etherlink in the short-term (the next six months) at least tenfold.

Making Etherlink Blazing Fast

Software developers in the room are probably familiar with this expression 🦀.

But when facing such a claim, it’s only natural to ask: how do we achieve that? In my opinion, there are two aspects to consider.

Firstly, we need to execute transactions faster. For each transaction submitted to Etherlink, we want to spend as little time as possible executing it. For this endeavour, we have several plans that will allow us to increase the throughput of Etherlink.

To give a few examples:

We are not particularly happy with the EVM implementation we are currently using (SputnikVM ). We are making great progress in transitioning to REVM .
We want to introduce parallel execution where it matters. It is a hot topic in the industry, with already solid results that we can leverage.
We have identified for a while now the shortcomings of the current runtime our smart rollups are built upon—and we have a team tackling this challenge to deliver a new RISC-V-powered runtime that will come with its own storage backend designed, implemented and optimised for optimistic rollups.

After combining all these efforts together and throwing a few more optimizations to the mix, we are confident Etherlink will be able to process transactions significantly faster.

Which brings me to the second side of making Etherlink fast: volume. Processing the transactions that we already have will make Etherlink more responsive, but we really want is to increase the overall capacity of Etherlink so that when you bring more activity to the chain, the system is not congested. To do that, we need to consider the amount of transactions that you will bring. We already have hundreds of thousands of transactions sent every day, but what happens when they become several millions? And several dozens of millions?

As of July 2025, the sequencer publishes Etherlink transactions inside Tezos Layer 1 blocks, but sadly block space is a scarce resource. To give you a few numbers, Tezos Layer 1 bandwidth (provided by its block space) is around 500 KBytes every 8 seconds. We can increase the block size, we can reduce the time between blocks, but the order of magnitude will not change. This becomes a real concern when we want to scale Etherlink without compromising with its non-custodial nature.

Scaling Without Compromise

We need to keep in mind what it means to increase both the speed of your system and the amount of data that you are processing with it. On the one hand, processing transactions faster is first and foremost an optimization challenge. It is a hard one, but ultimately it’s about using the hardware at your disposal in a more clever way. Processing more transactions on the other hand raises a security challenge, because it requires to publish more transactions. If Tezos Layer 1 blocks are not an option, where should the sequencer publish those?

One popular alternative to Layer 1 block space for a Layer 2 to publish its transactions is to rely on a Data Availability Committee (DAC). In this model, a small group of trusted entities is tasked with ensuring the data is available to participants.

But would Etherlink have stayed a non-custodial rollup if we had done that? The answer is a loud and clear “no.” Why? Because optimistic rollups rely on at least one honest participant that is willing to protect its state. That participant must have access to the full transaction data. If that data is only available via a closed committee, we introduce trust assumptions. And the moment you need to trust a small group for security, you're no longer truly decentralized.

This is why I am saying processing more transactions raises a security challenge. We knew that in 2022, when we started building the Data Availability Layer (DAL) of Tezos..We nicknamed the DAL the Rollup Booster , because the goal of the DAL is to increase the available bandwidth for Tezos’ smart rollups, and it is now live on Tezos Layer 1 Mainnet.

In a nutshell, the DAL is a proof of publication system. Anyone can send arbitrary data to the DAL’s peer-to-peer network and get back an attestation that this data has indeed been published. The same way Smart Rollups have been enshrined to the Tezos Layer 1 protocol, the DAL is enshrined to its consensus. That is, the body of the DAL’s attesters is the same one that is participating in the consensus of Tezos: the bakers.

Now is actually a good opportunity to celebrate that more than 85% of the baking power of Tezos is running a DAL node. DAL is already here today for Etherlink to use in order to increase its capacity. And since Dionysus (the current version of Etherlink), the sequencer can already decide to publish transactions to the DAL. It’s not doing it at the moment, because going through the DAL instead of Tezos Layer 1 blocks has a small impact on the latency of Etherlink’s bridges—but as soon as we need to, we’ll able to enjoy the additional bandwidth provided by the DALx10 compared to Tezos Layer 1 blocks today, x200 by the end of 2025. without any loss in decentralization.

It’s All Coming Together

If you have been following the work of core engineering teams from Nomadic Labs, Trili Tech and Functori working on Tezos and Etherlink, 2025 is a really exciting time. Everything we’ve been hard at work delivering our scaling roadmap for the past four years is now in production. Etherlink is live, bakers have enabled the DAL, and every piece of this technology stack remain true to the decentralization ethos of blockchains.

July 2024 to July 2025 was about bootstrapping this EVM-compatible layer of Tezos we call Etherlink. I want to emphasize how much work that entails. Launching the chain was actually the easy part. Having the chain being used, having an ecosystem in place, having the tooling, the partners, the projects landing: that was the hard part. It’s now done—as much as “being done” in the rapidly evolving landscape of blockchains means anything.

But we’re not done. We’re still preparing for the future to come. This talk was about discussing Etherlink’s capacity and how we intend to increase it tenfold. This will allow to create more room for more users to come, for more projects to be deployed. In the meantime, I do hope you find in Etherlink a welcoming platform for your projects!

Being a Stage 2 Rollup from Day 1: Etherlink’s Journey

July 11, 2024

Being a Stage 2 Rollup from Day 1: Etherlink’s Journey

Important

The content of this article comes directly from a talk I gave at EthCC[7] on July 11, 2024, as part of the “Rollups & Scaling Solutions” track. This is a personal transcript, and any mistakes in it are my own.

My name is Thomas Letan. I’m working for a company called Nomadic Labs Nomadic Labs is one of the largest R&D centers working on the Tezos ecosystem. , and today I am here to talk about Etherlink . More specifically, I will talk about our journey to be a stage 2 rollup from day 1.

What is Etherlink?

Before going any further, it is probably worth giving a bit of an introduction for Etherlink, because you might not already be aware of what it is already. Etherlink is a newcomer in the Layer 2 landscape, and has the particularity of being part of the Tezos ecosystem. That is, it is not an Ethereum rollup. Our Mainnet has been available in Beta since May 15.

Etherlink has been built with three core properties in mind: we wanted it to be fast (with a 500ms soft confirmation time delivered by a sequencer), fair (as we will discuss during this talk), and nearly free (your typical transaction costing one-tenth of a cent).

Our main focus during Etherlink’s beta is onboarding new partners. We have already onboarded several infrastructure partners, and our immediate focus now is on bringing dApps to the chain.

Etherlink is (nearly) a Stage 2 Rollup

Besides, what we think is interesting about Etherlink is that it is already (nearly) what L2beat calls a stage 2 rollup , and it has been since its genesis block.

To support this claim, let's go back to the definition of what L2beat stages are, and what a stage 2 rollup is about.

Etherlink naturally fully qualifies as a stage 0 rollup. Every transaction is posted on Layer 1 (Tezos in this case), along with L2 state root hashes. Besides, you can recompute the state of the chain locally using free and open source software (released as part of the Octez suite).
Etherlink fully qualifies as a stage 1 optimistic rollup as well. We have the full implementation of the optimistic rollup infrastructure, including a trustless bridge and a fraud-proof system.
Because this fraud proof system is permissionless (and has actually been permissionless since the genesis of Etherlink), we also meet a crucial requirement of stage 2 rollups. We are missing a 30-day exit window for users wanting to leave their funds out of Etherlink to fully qualify as a stage 2 rollup. This is because we favor a shorter feedback loop with our early adopters during the course of our betaWhile the exit window is currently 24 hours, it is already fully permissionless. Users do not need the help or authorization of a third-party to remove funds from Etherlink. .

Etherlink is a Non-Custodial Rollup

Meeting the main requirements of being a stage 2 rollup has been a direct consequence of our primary objective for Etherlink: to be a non-custodial rollup. That is, it was always clear to us that no closed set of third-parties should ever be granted exclusive, irrevocable rights on Etherlink. In other words, we wanted Etherlink to adhere to the same standards as Layer 1 blockchains, like Ethereum or Tezos.

To that end, we built Etherlink on top of two cornerstones: a permissionless fraud proof system, and a decentralized on-chain governance. My goal for the remainder of this talk is to give you a peek into the kind of solutions we came up in these two areas.

Etherlink is Powered by a Permissionless Optimistic Rollup

I was at EthCC last year to introduce Tezos optimistic rollup stack . We won’t revisit the details of our solution in depth, but let’s use this opportunity to give you a refresher.

Optimistic Rollup in a Nutshell

We are in the rollup track, so we can skip the 101 course on optimistic rollup. To me, the three points you need to keep in mind from now on is that an optimistic rollup

offloads computations away from a layer 1 blockchain in a new layer
enables bi-directional communications between the two layers with a trustless bridge
relies on an interactive fraud proof system to ensure the safety of your assets while they are locked in this second layer.

You cannot remove one of these properties without making an optimistic rollup either unsafe or useless.

Smart Rollups are Optimistic Rollups as a Common Good

Etherlink is powered by Smart Rollups, a one-year-old feature of the Tezos protocol. You can think of Smart Rollups as a general-purpose framework for building optimistic rollups. I like to think that we have implemented the full optimistic rollup infrastructure once (and —to the best of our knowledge— correctly), and now every member of the Tezos community can benefit from this effort.

This is not the first time I’m tasked to give an introduction to Smart Rollups, and I like to put forward the following properties:

You have understood it by now, but Smart Rollups are permissionless by default. By default, any XTZ holdersXTZ is the native token of Tezos, that is used to pay gas, validator fees, and participate in Tezos governance (more on that later). can post and defend L2 state root hashesIt is also possible to create permissioned Smart Rollups, for use cases that require it. This is not the case of Etherlink, obviously. .
Smart Rollups are not implemented by means of smart contracts. They are directly enshrined in Tezos protocol, making them a first-class citizen of the Layer 1 blockchain. This was instrumental in enabling several optimizations to increase performances and reduce end-users costs.
They are fully programmable. Smart Rollups are not about creating EVM-compatible Layer 2 blockchains, they are way more general than that. The same way you provide a script when creating a smart contract, Smart Rollups come with a (WASM) program that decides the semantics ruling the newly created Layer 2 rollup.

Etherlink is not the only use-case for Smart Rollups! To name a few examples,

We have been able to demonstrate Tezos can support one million transactions per second on a public testnet using 1,000 smart rollups last year.
We are working on a Javascript runtime (called jstz ) for Layer 2 blockchains powered by a Smart Rollup.
Engineers from Plenty , a DeFi platform deployed on Tezos, have been experimenting with Plenty Rollup to exchange assets with low-latency and high-throughput.

Etherlink Decentralized Governance

Smart Rollups are the technological foundation of Etherlink, but they are not enough to make it a non-custodial rollup. The same way you can create a centralized smart contract on top of a decentralized blockchain, you can definitely build a custodial Smart Rollup by embedding a public key in its code.

This brings us to decentralized governance.

If you were aware of Tezos before this talk, it is probably because of its on-chain governance system. If you don’t know, Tezos is a Layer 1 blockchain officially launched in 2018, with a track record of 16 forkless upgrades. This is possible because the governance of Tezos is embedded in its protocol. We definitely wanted to leverage this when we started building Etherlink.

Etherlink (Absence of) Governance Token

It was clear from the get-go that Etherlink would “inherit” from Tezos strong and vivid governance. One consequence of this choice was to XTZ as its governance token, instead of introducing a new token. Besides, Tezos governance is larger than XTZ itself, because it is built on top of a pioneer liquid proof of stake mechanism. For Tezos upgrades, only the validators vote, but their ballot’s weight is determined not only by their personal stake but also (and sometimes mostly) by the balances of individual holders who have “delegated” their XTZ to themIt is clearly out of the scope of this talk, but for your information, Tezos liquid proof of stake has been significantly amended with the Paris upgrade to include a new staking mechanism which complements the existing delegation approach. .

This should give Etherlink a head start, because the ecosystem of Tezos has a clear and well-established maturity when it comes to on-chain governance.

Handing over Etherlink to Tezos Validators

Similarly to Tezos, Etherlink governance happens on-chain, and once the vote is finished, there is nothing we can do to prevent its consequences from becoming a reality. Everything is embedded in Etherlink program.

The voting process is a lightweight revisit of Tezos’ one. The governance of Etherlink is carried through the Layer 1 blockchain, because this is the place where the voting power of the governance body (Tezos validators) is determined.

The vote proceeds as follows.

Firstly, any Tezos validator can propose a new upgrade during the proposal period. I believe this is a significant rule to highlight. Proposing a candidate is not a right limited to —let’s say— Etherlink core developers. On that, we follow the footsteps of Tezos, and what happens on Tezos during the proposal period is often quite interesting. We have seen time and time again the community using this opportunity to engage very actively with the governance process. For the Ithaca proposal for instance, a validator had submitted a counter-proposal disabling a feature they didn’t like. For the Oxford proposal, validators proposed a variant of the proposal tweaking some parameters of the key feature introduced at the time. This is “governance at work,” and the governance body can then select the proposal they find the best fit for the Tezos network.

Secondly, the promotion period starts with the proposal that has received the most upvotes. Obviously, winning an election among many participants does not ensure to the victor a majority of supportFor context, I gave this talk in the midst of France almost seeing a far-right government taking power, and the left “winning” a snap election with too few parliament members to apply their program. . The question asked to validators boils down to: “Do you approve this proposal?” The goal is to verify that the selected candidate can gather a supermajority of support. This additional step was designed to alleviate the risk to see unhappy validators fork Tezos.

Finally, the cooldown period gives an opportunity to (1) the ecosystem to adapt to the features and changes brought by the new software, and to (2) dissatisfied users to exit Etherlink.

The current parameters for the governance of Etherlink are as follows.

	Regular upgrade	Security upgrade	Sequencer
Voting period	~ 2.5 days	~ 7 hours	~ 5 days
Cooldown period	~ 24 hours	~ 24 hours	24 hours
Quorum (proposal)	1%	5%	1%
Quorum (promotion)	5%	15%	8%
Supermajority	75%	80%	75%

I will not spend much time discussing each number individually, and instead focus on the three kind of votes.

Firstly, and this is probably what you had in mind when I started to mention decentralized governance, is voting on the next version of the software powering the chain. We distinguish between “regular” upgrades (the default case), and “security” upgrade. The latter can be used exceptionally if a bug is found in Etherlink that can endanger the assets of Etherlink users. The security upgrade has shorter periods of vote, but requires a higher quorumDid you know? Etherlink has gone through a security upgrade already . The patch deployed fixed a critical issue in the withdrawal precompiled contract, which was found by Spearbit during their audit. . It is worth mentioning that we do not have a security committee, as it is often the case for other Layer 2 blockchains.

Secondly, and I think it is a unique thing in our industry, Etherlink’s sequencer is elected by Tezos validators. This brings a clear incentivization for the sequencer to behave correctly. Otherwise, Tezos validators can simply vote them out of Etherlink. It does without saying —so it’s better when reaffirmed— that the sequencer power is limited to proposing blocks, the validation of said blocks remains permissionless. In particular, in the context of Etherlink being a Smart Rollup, only valid transactions that are actually posted on Tezos can lead to funds transiting from Etherlink to Tezos through the permissionless bridge.

Another interesting fact about Etherlink and its relationship with the sequencer is how Tezos consensus algorithm and low block time enables Etherlink’s blocks to become final under less than a minute. Tenderbake (a variant of Tendermint) features a 2-block finality property that is incredibly rollup-friendly. Since the sequencer typically posts its block proposals directly after their creation, it is not unusual for an Etherlink block to become final in 30 seconds or so.

For partners and institutional entities that cannot rely on soft confirmation from a sequencer, we believe this can be a game changer.

Why Non-Custodial?

This talk is slowly coming to its end, and one way to conclude it is to come back to this idea that we wanted Etherlink to be non-custodial from the get-go. This should raise an interesting question: why? Why was it so important for us to make it a priority that early in the development process?

This is particularly relevant when we take a step back and have a look at our ecosystem. To this day, the leaders of our industry are still powered by stage 1 (Arbitrum, Optimism) and stage 0 (Base) rollups.

The origin of our decision lies in our long term goal for Etherlink. Etherlink as a Layer 2 blockchain is important, and we have invested a significant amount of effort and time to make it as successful as we can, and it is also a component of a larger scalability roadmap we coined Tezos X . Tezos X aims to bring the full Tezos ecosystem into a scalable, decentralized optimistic rollup compatible not only with Michelson (the home-grown smart contract language of Tezos) and EVM, but mainstream programming languages as well (as suggested by our efforts on jstz).

In the meantime, I can only encourage you to take interest in Etherlink today. As mentioned, its Mainnet is already there and will soon move out of the beta phase. Our fast, fair and (nearly) free Layer 2 blockchain awaits you.

Can Tezos Become EVM Compatible?

July 20, 2023

Can Tezos Become EVM Compatible?

Important

The content of this article comes directly from a talk I gave at EthCC[6] on July 20, 2023, as part of the “Core Ethereum & Scalling” track. This is a personal transcript, and any mistakes in it are my own.

My name is Thomas Letan. I work for a company called Nomadic Labs Nomadic Labs is one of the largest R&D centers working on the Tezos ecosystem. , and today I am here to talk about our work to bring EVM compatibility to the Tezos ecosystem.

What is Tezos?

Tezos is a Layer 1 blockchain that has been around for over five years. One of Tezos’ key differentiator —at least initially— is its on-chain governance system. The governance system is a voting procedure to propose, select, and automatically deploy core upgrades for the blockchain. This governance system enabled the Tezos network to seamlessly go through 14 upgrades since the beginning of our Mainnet. These upgrades allowed us to keep innovating at a strong pace, and today Tezos notably features its own smart contract language and runtime , a 2-block finality consensus algorithm called Tenderbake since April 2022Tezos was initially deployed with a different consensus algorithm, which means we have hotswapped Tezos consensus while it was live. , and a first-class citizen framework to deploy and operate so-called Smart Rollups .

Smart Rollups are optimistic rollups enshrined in their Layer 1 blockchain (Tezos). We are today at EthCC[6] because we believe they can be a game-changer for Tezos, and we are excited to share this with you.

The starting point of Smart Rollups was the scaling roadmap of Tezos. It has been our main focus for the past months, leading us to demonstrate on a public testnet how 1,000 rollups can process 1,000,000 transactions per second together. But Smart Rollups as they came to be also enable us to explore brand new use cases. One of them is EVM compatibility, something Tezos itself does not provide to our users.

During the remainder of this talk, I will provide you all the keys to get a grasp on what Smart Rollups are, and how we intend to bring EVM Compatibility to our ecosystem through one Smart Rollup.

A Primer on Smart Rollups

The selling pitch for Smart Rollups is to present them as new ways to implement decentralized applications. They offer a trade-off that is slightly different than smart contracts already propose. In a nutshell, Smart Rollups get more computational power compared to a smart contract, but since these computations are offloaded from the Layer 1 blockchain, you need to bring your own infrastructure to carry it.

Optimistic Rollups Done Once

Under the hood, Smart Rollups are an optimistic rollup framework with three critical properties.

They are permissionless by default, meaning that any Tez holderTez is the native token of Tezos. can create a rollup, and compute, publish and defend the state of any rollup created on Tezos. That is, we have permissionless interactive fraud proofs already!
They are enshrined. A Smart Rollup on Tezos does not rely on a collection of smart contracts, but is instead of first-class citizen of the Tezos protocol itself with a dedicated address. Optimistic rollup maintenance operations are dedicated operations of Tezos, optimized for performances and gas efficiency. Besides, they benefit from Tezos being able to evolve and improve over upgradesIt’s more than a promise already. Smart Rollups have been enabled on Tezos Mainnet on March 2023 with the Mumbai upgrade. Since then, Nairobi has been activated and has improved the developer experience for every Smart Rollup developers. .
They are programmable. The same way you create a smart contract by providing a script, Smart Rollup are provisioned with a so-called kernel (a WASM program) when they are created, and this program will dictate the semantics of the rollup. This is the key property that makes Smart Rollups a solid contendent for addressing a wild range of use cases, from generalized Layer 2 blockchains to specialized AppChains.

The Smart Rollup Runtime

But what is exactly a Smart Rollup, from a developer perspective? The best way to answer this question is to provide a high-level description of the runtime used to execute its kernel.

The first component a kernel can interact with is the shared inbox. The shared inbox contains messages coming from the Layer 1 blockchainI want to insist on that: the inbox is shared in the sense that all the smart rollups deployed on the Layer 1 blockchain have access to all messages sent at a given time. . These messages can be sent by wallet-owned accounts or smart contracts alike, but they they must compete for Layer 1 block space to do so. This is fine until it is not —in case of significant adoption for instance— so we equipped Smart Rollups with an additional way to import data from the outside world: the so-called reveal channel. At its core, the reveal channel is a primitive which allows a kernel to retrieve an arbitrary data, provided it knows the data’s hash. Although a smart rollup cannot guess a hash out of thin air, the competition for block space becomes a lot less fierce when the competitors only need 32 bytes per block each.

Similar to smart contracts’ storage, a kernel can read from and write to a “durable storage.” The key difference between the two is how a kernel does not pay for bytes allocated in the durable storage. The rationale is the same as for execution time: since everything happens offchain, the cost for allocating more Layer 2 state is higher requirement for the hardware infrastructure.

A kernel can send back messages to wallet-owned address and smart contracts of the Layer 1 blockchain through the outbox. Smart Rollups being optimistic rollups, the usual limitations apply for this communication channel. Messages of the outbox can only be triggered on the Layer 1 blockchain once the Layer 2 state advertising them has been published and cannot be contested with an interactive fraud proof anymore. For Smart Rollups, this means waiting for two weeks as of today.

Finally, Smart Rollups get free execution time for every block created by the Layer 1 blockchain. This is a very different model than gas-bounded computation of smart contracts, and enables new use cases. For instance, you can imagine having a kernel reorganizing its storage across several Tezos blocks in a transparent manner, without the need to inject any message to the shared inbox.

Smart Rollup’s Adopter Starter Kit

Implementing a kernel is only the beginning of the journey. You still need to execute it, which is more involved than deploying a smart contract.

If you work on Layer 2 solutions yourself, you may know that optimistic rollups can be tricky to operate, and it would not be reasonable for every Smart Rollups adopter to have to reimplement from scratch their own daemon. From this perspective, our pledge has always been to make that as easy as possible, and to that end, we have invested significant efforts to provide a software suite that lets you focus on your business logic. In particular, we provide

A high-level, type safe Rust framework to implement your kernel. This makes a very interesting alternative of the low-level, C-like API exposed by the Smart Rollup (because that is how WASM works).
A debugger to test, inspect, profile and benchmark the execution of your kernel.
A general-purpose daemon responsible to track the state of your rollup, publish it on the Layer 1 blockchain, and defend it if need be.

A Generalized, EVM-Compatible Layer 2 Blockchain for Tezos

The main reason why we have been eager to comme to EthCC this year is because we are about to bring EVM compatibility to the Tezos ecosystem, thanks in large part to Smart Rollups.

Building an EVM-compatible Blockchain

When we started working on this a few months ago, we have quickly realized that “building an EVM-compatible rollup” does not mean much... or rather, it can be interpreted as many different things. When you build a rollup, you are confronted to a lot of design decisions. The first of such decisions that comes to mind is your requirements in terms of latency: do you need a sub-second block time, or are you fine with sticking to your Layer 1 blockchain latency? The answer to this question has a direct impact on the decentralization of your rollup. As of now, the de facto solution to sub-second latency is the delegation of block production to a single entity (the sequencer). Another interesting question is about data-availibility. Are you fine with being limited by the block space of your Layer 1 blockchain, or do you know your requirements will far exceed them? And what about gas price in your Layer 2 blockchain? Smart Rollups do not have a notion of gas, but if you are building a generalized rollup emulating a full blockchain, you will need gas and a gas price to constraint what your users can do according to your infrastructure capacity. On the other hand, if you are building an AppChain with only a handful of Solidity contract whose execution can only be triggered by yourself, you probably don’t need that.

What was clear to us is that we did not want to take too many decisions like these too soon in the development process, but rather built a versatile solution which can be adapted to end-users need. As a consequence, what we are working on today is a twofold projectWe are providing a layered answer to a complicated question. , with:

A customizable “template” for generalized rollup, compatible with the Ethereum ecosystem at large.
A community instance of this template, focusing on decentralization and cost-efficiency.

The key idea of this approach is to leave the door open to potential adopters whose requirements are not addressed by the community instance we are working on. It should be straightforward to deploy a dedicated instance that is tailored to their needs.

To achieve this, we have drawn inspiration from the block proposer/block validator paradigm. Our rollup program is divided into two “stages.” The stage 1 interprets messages coming from the shared inbox to reconstruct block proposals. Then, the stage 2 validates these block proposals, and publishes the Layer 2 history. From this perspective, most of the work required to support a sequencer setup is focused on the stage 1, while manipulating gas fees is done by modifying the stage 2 for instance.

Ensuring Compatibility with the Ethereum Ecosystem

This architecture has proven itself to implement the semantics of an EVM-compatible Layer 2 blockchain, but this was only the beginning of the journey. For this new chain to be usable, we also need to be compatible with the galaxy of tools and solutions that is the Ethereum ecosystem. This is achieved by implementing the JSON RPC API . To be clear, this was unknown territory for us at the beginning of this project. The nodes participating in the Tezos network uses a different approachWe have settled on a REST API. . As a consequence, we are developing a new node to fill this gap, and we are now capable of serving the API to tools like Blockscout and MetaMask, or dApps from partners. To give you a concrete example, we had an improvised workshop with Arianee today to deploy their application on our brand new TestnetIt was deployed the day before the talk, and it is still used today . . As far as I know, it worked! And maybe your dApp is next ?

I mention we are building a community rollup, and I am happy to unveil today that its name will be Etherlink .

What’s coming next?

Our journey to scale the Tezos ecosystem and open it to new use cases and opportunities is far from over. In the meantime, if there are two things that you should remember from this talk, it is that:

EVM-compatibility is coming to Tezos this year, starting with EtherlinkThis promise was a tad optimist. The Mainnet Beta of Etherlink was eventually launched on May 15, 2024. .
You can already test to deploy your dApp on our brand new Testnet.

Thank you very much, and see you on Tezos!