At the recent EDCON conference, ZK became a hot topic and was an important part of Vitalik's personal speech. During the discussion about ZK, the zkEVM field was repeatedly mentioned. In the current encryption industry, several top players in zkEVM, such as Scroll, Polygon, and Consensys, are competing fiercely, making zkEVM the most popular track in the entire industry. Everyone is rushing to launch testnets and mainnets, and striving to attract more ecological developers.
During Vitalik's speech, he mentioned an interesting project called Kakarot. This zkEVM, built on Starknet using the Cairo language, has received support and favor from Vitalik and a group of top crypto VCs, although it has not yet been launched on the testnet. In the recently announced Kakarot financing information, Vitalik's name appeared rarely on the list of investors.
Kakarot is actually a meaningful name. In the classic anime "Dragon Ball" produced by Toei Animation in Japan, the protagonist Son Goku's name is "Kakarot". As a representative of Saiyans, he has the ability to transform into a super warrior. For Ethereum, which is currently in the expansion phase, what the ecosystem needs most is a super Saiyan who can solve the problem of expansion.
Of course, a well-chosen name alone cannot bring such high attention to a "rookie" in the encryption industry known as the "most exciting track in the encryption circle". So what is the charm of Kakarot? BlockBeats interviewed Elias Tazartes, co-founder of Kakarot, to gain insight into the unique perspectives of this 25-year-old French youth on Starknet, zkEVM, and the future of blockchain scalability.
It's hard to imagine that Elias, who has become a leader in the highly demanding field of ZK technology, was still a newcomer to the industry two or three years ago. Like many developers, Elias has developed on many L1 ecosystems, including Solana, Cosmos, and other non-EVM compatible public chains. It was this experience in the non-EVM ecosystem that made Elias a firm "Ethereum Maxi". And it was the encounter with the concept of ZK that truly showed him the direction of his development in the industry.
Elias: Hello everyone, I'm Elias, a 25-year-old Frenchman born in Paris. My major in college was mathematics and humanities. I almost became a trader in traditional finance, but I gave up later, or more precisely, I didn't pursue it further. Later, I switched to learning programming. I started investing in Ethereum in 2019, but I didn't know how to code at that time.
In France, I found that there are so many cool applications emerging. I don't know if you know Sorare, they are the company that first launched NFT football games, and their team is in France. I don't even watch football, but they are famous in France and have a valuation of $1 billion. This may be the first large-scale, consumer-facing retail cryptocurrency application. Cryptocurrency is a brand new business, and everything needs to be built, and there are no barriers. For someone like me who doesn't have an engineering background, only a mathematical background, it's easy to get into. So I started developing something on Solidity and joined this company through a college friend.
Elias: This is indeed a big span. However, I like new things and exploring everywhere to see what's fresh. I started with NFT, Token, and ERC-20 like everyone else. I found that they are the best foundation layer for blockchain, but they are not the most exciting technology, right? This is good for consumer goods, as we have all reached a standard, but next we need to build new things.
I came across ZK by chance through a Circom training camp. At that time, I didn't have much confidence in EVM, but through this training camp, I realized that L2 and Rollup are using ZK. Not for privacy or applications, but directly for L2, such as the StarkWare team. There are many Starknet developers in France, and I got to know them and started developing on Starknet. Now, I can be considered an "Ethereum Maxi".
Elias: I think it makes sense to diversify and try everything during the "L1 War" period. In 2021 and 2022, I tried building on the Algorand, Solana, and Cosmos ecosystems, but later I realized one thing: all the teams developing on these ecosystems eventually returned to the EVM. So I also "came back".
The cost of deploying a contract on L1 is $100. You can choose Optimistic Rollup, but I don't like them. I believe that within three years, there will be no more Optimistic Rollup, only zkRollup. Because they are cheaper and faster to execute, from the user's perspective, this is very good, but the security assumptions of Optimistic Rollup are lower than those of zkRollup. For example, when you want to withdraw funds from L2, Optimistic Rollup takes seven days, while zkRollup only takes two hours.
Elias: In my opinion, Optimistic Rollup is very interesting because they are really smart. In an Optimistic Rollup, after a transaction is confirmed, if you think someone is cheating, you can challenge them and go through a process called fraud proof to prove that someone is cheating. The great thing about Optimistic Rollup is that the team can launch an EVM, which is L2, without fraud proof because they control the only sequencer. It's like saying, "We don't have fraud proof technology yet, but we are a trustworthy company because we have sorted the blocks."
As for zkRollup, it is more complex because security must be ensured first. Starknet must invent a secure chain, which is now being produced on the mainnet, but it took a lot of time and the performance is not even as good as Optimism. Simply put, we cannot make ZK Rollup without mathematical principles, so more time is needed for technical implementation, as well as to improve performance.
Elias: It's not entirely true. The interesting thing about zkRollup is that they are based on a proof system. One of the main features of a proof system is that you cannot cheat. If you submit a proof to L1, the verifier is a state contract and will be executed by Ethereum L1. If the Starknet sorter says that Elias gave all the tokens to Jack, but I didn't sign it myself, then he can't create this transaction. He can't lie because he needs to submit a proof to L1 that proves that the transaction is valid, and integrity means that you cannot prove false transactions.
Due to the need to fully use the Cairo programming language, many developers who value efficiency and cost are discouraged from using the Starknet ecosystem. Therefore, it is indeed very reasonable to build an EVM-compatible platform based on Starknet. However, many people see this as a very difficult and challenging task in itself. In order to use EVM, a new language must be used to build an engine that runs old languages, and there are suddenly three stacks added: Ethereum mainnet, Starknet Layer 2, and Kakarot zkEVM. So how did Kakarot design to solve these problems?
Elias: EVM uses a block format called Keccak, which is a very expensive proof that requires a lot of computation and needs to be constantly used in the blockchain. For example, Keccak will appear in various parts of your code library. Every time you encounter a Keccak, it's like losing three minutes or losing $100. Now, to prove a block, if there are 100 Keccaks, you lose that much money. And if there aren't any, then you don't lose any money.
When designing a new ZK Rollup, your dream is to eliminate Keccak, which is what Starknet does. However, zkEVM cannot do this because it must be compatible with Ethereum and must use similar primitives. Technically, there isn't much complexity. When people say that combining the two is very complex, it's mainly because as a zkEVM, what the zero-knowledge system is trying to prove is basically the same as what they take over all transactions, just like in a block, and they prove them one by one. When you prove a transaction, you are saying that this is how it happened, and I have no possibility of lying. This may be what is meant by ZK and EVM not being very compatible.
Elias: We need more zkEVM. After competing with each other, most of them will disappear, and only the best will remain, just like the "L1 War". Everyone is a killer of Ethereum, and Ethereum has withstood the test of time. I believe L2 will be the same.
There are currently about 20 L2s, and in a few years there may only be 5 left. I like Starknet and I think people like EVM, so combining the two is natural. Starknet is undoubtedly the best zk-Rollup, and Solidity is a great toolbox. EVM-compatible tools like MetaMask, WalletConnect, and Gnosis Safe are all good tools, but you can't use them on Starknet. Some very cool builders are building new things that are very different, which is great, but most users want to use what already exists.
Elias: Maybe we will encounter some issues regarding block format, and the team is researching this area. However, we have already implemented EVM in Cairo. This was quite difficult, but it has been completed. This means that our EVM is like our code library, or our product, and our project can deploy Solidity code and run it in real-time. This is feasible, we just don't have a chain yet, but the engine can work. Next, we are thinking about how to integrate it into many ecosystems or products, but it is technically feasible.
I want to say fairly that it will also be very difficult to build zkEVM based on EVM, because EVM is not designed for zero-knowledge technology and mathematical design. When you try to combine the two, you will indeed encounter many problems.
Elias: I think it's actually just an L2. It's built on top of Starknet, but it's not a separate layer because Kakarot is currently a smart contract on Starknet. This may sound a bit strange, but you can try to think of it in a simple way through some examples.
For example, if you click on some applications on your Windows, it opens a Linux inside your Windows. EVM is just a computer that can perform calculations. You give it some compiled Solidity code, and it outputs something for you. EVM is not a blockchain, it is an engine. If you have a car, EVM is the engine, but the seats are something else.
Elias: Yes, you can think of the structure as changeable. We can consider Kakarot as a "Rollup as a Service". If we provide Rollup service, Kakarot can act as L3, which will be a comprehensive chain. A chain may have its own blocks, its own sorters, its own verifiers, and so on.
For example, Starknet verifies many blocks and submits them to L1. On Starknet, there is Kakarot, which does not settle or submit blocks, but inserts EVM transactions into Starknet blocks. If Kakarot is L3, the process will be: Kakarot batch processes blocks, submits them to Starknet, and then Starknet submits them to Ethereum. However, currently, a developer who deploys Solidity on Kakarot actually calls the Kakarot contract with their EVM code. We deploy a Starknet contract with EVM code for them and remember the address. The dApp actually calls Kakarot, which is a bit like a router.
Elias: Yes. To me, the concept of "Rollup as a Service" is like Coinbase forking OP Stack. After they have their own chain, they package it into a nice service and call it a Rollup service.
This is a bit like a B2B experience, where a large company wants its own chain. It doesn't have to be private or licensed, but it must be their own. For example, if there is a lot of congestion on Starknet, your transactions will be delayed. If your company is doing high-frequency trading, you may not want to be congested. So they can find someone who provides Rollup services and create their own chain, so they can manage congestion themselves.
For example, dYdX has its own chain, but it is private and permissioned. However, in the future, it is likely that Rollup services will come into play. Those who like zero-knowledge ecosystems and their security can come to Kakarot and say, "I don't want to be on L2, I don't want you to be my transaction router, I don't want to deploy on Starknet, I want to deploy on a higher level and settle on Starknet every three hours." At this point, they can come to Kakarot and deploy a Kakarot chain.
Elias: It is an EVM for Starknet or an instance of a Starknet chain, and Kakarot is the only possible entry point. What you get is a real EVM, and the experience for users and developers will be exactly the same as Polygon.
It's similar to a sidechain, but without its own consensus and security. Proof is submitted to Starknet, which then submits it to Ethereum. This is a good way to do things, but the finality is delayed. You submit to Starknet every four hours, while Starknet may submit to Ethereum every 30 minutes. It takes longer for your transaction to be finalized, but you may be able to achieve 5000 TPS because you only do your own transaction batching.
I believe that if someone wants to make an RPG or large-scale MMO game, such as Ubisoft, they need an environment without congestion. They can deploy their own chain and tell players that each transaction is submitted to Starknet once every four hours. You can get the same security as L1, at a very low price, and fully compatible with EVM.
Elias: I think the main difference in this situation is that through zero knowledge, even if you go to L5, you will get the same level of security as L1. This is because you are sharing the security and consensus of Ethereum. I think in this field, we have studied many ways to expand, including Avalanche's subnet, and perhaps Cosmos' IBC, etc. But in the end, we all come back to Ethereum.
Despite the fact that the development of Kakarot itself has not yet been completed, Vitalik's support and the pursuit of VCs have already made Kakarot a dark horse among the crowd of zkEVM. In an interview with Elias, BlockBeats asked about the current development progress of Kakarot, the specific time for the launch of the test network, and Elias's views on the future of the zkEVM ecosystem.
Elias: I think so. Ironically, 80% of the teams entering this field choose EVM, but they realize they can't build on L1 because it's too expensive. Going to L2 will have good security, and there's Arbitrum and Optimism. In a few months, it might be Polygon zkEVM, zkSync, and Starknet. And except for Aztec, there's basically no privacy. No chain is primarily concerned with privacy, but with performance. Privacy may appear in the L2 space in two years.
Elias: It depends on the actual situation. We don't need sorters or verifiers now. If we only build on L2 in the future, we won't need them. But if we do L3, we will need them. Because if it's L3, we need our own blocks, and if we need blocks, someone needs to sort them and someone needs to verify them. In L3, a sorter set and a verifier set may be needed. You may need a token to achieve decentralization and have incentive mechanisms. Therefore, verifiers may need to stake, users will pay gas fees, and then have the balance of this token economy.
Elias: We should have a public testnet before the end of summer. But the mainnet is still pending, and I cannot make any promises. It may take another six months or longer.
Our main goal is to make you not perceive any differences. Tools like Metamask will run on L1 EVM just like any contract, and everything will be compatible. As you said, the main goal is for composability, so you don't have to reinvent the wheel. When you come to the ecosystem, you can still use any tools in your toolbox that you have been using for years.
Imagine a DeFi protocol that has its own DeFi protocol, for example, multi-chain operations can be carried out on Arbitrum and Optimism. They create a code repository for each ecosystem. For them, the easiest option is to come to Kakarot and deploy the existing code repository, because as a multi-chain protocol, you may never want to maintain two code repositories, one on Cairo and one on Solidity.
Elias: There is something very interesting about Kakarot, and that is the team behind it is currently working on Cairo because EVM was written in Cairo, but our developers need to write Solidity. We have received a lot of attention from developers in the Starknet ecosystem, but they are all Cairo developers. They helped build EVM, and we need to guide the Solidity developer community. Our current idea is to have the co-founders and ecosystem leaders participate in the Autonomous World and ZK communities, and we hope to attract these people. On the StarkWare side, we are reaching out to those who want to deploy EVM code to Starknet.
I believe our true goal is for people to be able to use our tools. This means our ultimate goal is for you to not be able to distinguish between Polygon and Kakarot, Ethereum and Kakarot, or Arbitrum and Kakarot. Even with new technologies and architectures, this remains the primary objective of any EVM chain.
If you copy and paste Ethereum or fork Ethereum now, everything will work fine, right? But if you invent a zkEVM, that's quite complex, and you've changed a lot in the architecture, you need to make sure it doesn't harm users. Our main goal and main user experience goal is that developers and users won't notice that they're on Kakarot, it will be abstracted from them. Excitingly, they will be in the zero-knowledge field, like zkRollup, and they will benefit from all the innovations at the forefront of technology and lead the future efforts.
Elias: This is definitely a collaboration, because it did not exist five years ago. Even three years ago, it had never existed. In Jordi Baylina's EDCON speech, he kindly said that we need to work together, stop marketing and mutual slander, etc. There should be cooperation between Polygon, Kakarot, and Linea, and even people from Optimism, Arbitrum, and other zkRollup. Because the more zkEVM projects there are, the more thinking and research will be put into its development, which will produce many positive external effects.
Finally, what I want to say to everyone is that we need more developers from China to join this field. I know there are many Solidity developers in China, maybe not so many Cairo developers, but if they want to join, that would be great. We are all working towards the same goal, which is to enable people to use cryptocurrencies, use Ethereum, use applications, and execute transactions in less than a second, with fees lower than one cent. This is the only important thing.