Organized & Compiled: Deep Tide TechFlow

Guest: Vitalik Buterin (Co-founder of Ethereum)

Host: Qiu Dagen

Podcast Source: Wu Says No Crypto

Broadcast Date: April 27, 2026

Editor's Introduction

Vitalik spoke at the opening event of ETH Hong Kong Community Hub, revealing that the goal of Ethereum L1's computation scalability is to increase it by 1000 times, and the team has now entered the actual tackling phase. He also stated that L2 should not be a copy of L1, but should provide capabilities that L1 cannot achieve, including privacy, low latency, and high throughput. In terms of post-quantum computing, he provided a specific technical path: replacing elliptic curve signatures with hash-based or lattice-based signatures, and then compressing the on-chain volume through STARK aggregation to an acceptable range, expected to be completed in 3-5 years. Vitalik also specifically warned that AI could automatically discover all DoS vulnerabilities in clients within 2-3 years, replaying the attack events before the 2016 Shanghai DevCon, which the Ethereum Foundation is currently prioritizing as a high-security issue.

Highlights

Positioning and Value of L2

“A good L2 cannot be a copy of L1. A good L2 should be complementary to L1, doing things that L1 cannot achieve.”

"Every application will have its own computation, there will be many combinatorial problems, and it will be difficult for every application to communicate with each other."

L1 Scalability: Data and Computation

"Ethereum L1 can scale a thousand times, but there are many challenges."

"A scalable ERC-20 token will have a significant difference from the current ERC-20."

"Many of the underlying philosophies of Ethereum cannot be completely detached from L1, so L1 needs to scale more."

Post-Quantum and Security

"An elliptic curve signature is 64 bytes, whereas a post-quantum one is about 2000 to 3000 bytes, so new technologies are needed."

"We need to have the nodes creating blocks take all the signatures and do a STARK to prove the existence of those signatures."

"If it's not secure, AI three years from now will surely find every vulnerability, causing us great pain."

The Chinese Community and Technological Generations

"When a new technological trend occurs, it gives a great opportunity to the new generation. Because there are no ZK developers with ten years of experience, everyone is starting from scratch."

"Think back to what made you first interested in Ethereum and what this world needs now. It’s possible that the answer may not even include Ethereum technology."

The Chinese Community and Ethereum

Host: It's very touching to meet again after four and a half years. Let’s talk about your journey from founding Ethereum to now, and your personal feelings regarding your interaction with our Chinese-speaking community.

Vitalik:

I first heard about the Chinese community in 2013, before Ethereum existed, only Bitcoin was around. In 2013, I heard others say that the mining farms and exchanges in China were larger than those in the U.S., and I found it to be a very interesting world, but at that time the English media had not covered it at all, so I was very curious.

In May 2014, I came to China for the first time, visiting Beijing, Shanghai, Hangzhou, and Shenzhen, exploring many communities, and saw many miners and exchanges. I remember Huobi’s office had hundreds of employees, while Coinbase and Kraken in the U.S. had not yet exceeded one hundred. Many people were thinking about Bitcoin and starting to contemplate smart contracts.

When I came in 2015, BitShares was quite popular. Their method was somewhat different; each application had its own blockchain, and these blockchains could interact, but each application remained relatively independent. Some very smart individuals based their work on BitShares' code.

The first academic contributor to our community was a developer named Qian Youcai, who published a white paper introducing a decentralized random number generation method called RANDAO. RANDAO later became our current method for selecting the next block node on the chain.

The community was particularly active, with everyone working on different applications, and there were many ICOs. After around 2020, during the pandemic, there were many changes. I found that many people completely forgot the feelings from that period, as if it didn't exist. A significant shift since then could be ZK (Zero-Knowledge). I believe when a new technological trend occurs, it offers tremendous opportunities for the new generation since there are no ZK developers with ten years of experience; everyone is starting from scratch. Additionally, now with AI, writing any code has become incredibly easy. If you have not written any Solidity code until now, you can start writing it now.

Since 2022, with the advent of L2, more people began to participate in DeFi, and gradually some started engaging in core development. Core development might be more challenging because L2 resembles a free market; if you are an outsider, not part of the core community, but you can create good things, you can still succeed. But core development isn't like that. They need to discuss which EIPs (Ethereum Improvement Proposals) to add next, when to increase the Gas Limit, when to add ZKEVM. Participating in this process requires coordination with current developers, making it more difficult. However, it is slowly expanding.

Overall, my observation is that it started with mining and exchanges, and then shifted to more profound involvement in technology and every layer of the community. This process has been very good, but there is still a long way to go.

L2 should not be a copy of L1

Host: Thank you for quickly reviewing the developments of the past eleven to twelve years. Since you launched Ethereum, you've spent a considerable amount of time traversing various places in the Chinese community, in China, in Hong Kong, etc. Recently, you shared your views on L2, stating that L2 cannot merely be about scalability. How do you view the current development of L2?

Vitalik:

L2 is still very important now, but I believe a good L2 cannot be a copy of L1; it should be complementary to L1, doing things that L1 cannot achieve. There are many applications that need blockchains, but beyond the blockchain, additional factors are needed, such as privacy, high scalability, low latency, or oracles.

The ideal approach to L2 is to look at applications you think are very necessary and worth pursuing, consider where those applications need L1, and what common needs L1 cannot meet. If you think this way, you will find that creating a copy of EVM is not ideal. I find that the most successful L2s started with many EVM copies, but now the most successful ones have their unique, non-EVM parts. Creating these must have purpose; it should not be done just to create an L2.

L1 Scalability Pushing Forward in Two Lines

Host: You recently released content regarding the roadmap, and I particularly want to ask you about data scalability and computation scalability—both of which are of great concern to everyone. How do we scale? Are there any directions?

Vitalik:

First, why is L1 scalability important? Because we find there are many applications that, in theory, would need to place more parts in L2 if there is no scalability for L1. However, doing this would require your application to rely on more intermediaries, leading to more possibilities for failure and problems. Thus, the underlying philosophy of Ethereum still cannot completely detach from L1; L1 needs to scale more.

L1 has two parts: data and computation, both of which need to be scaled. The last upgrade we did was Fusaka, which included PeerDAS, a method for data scaling. Currently, only 25% of the data that the Ethereum chain can support is being used, but we can scale when needed—potentially ten times or even more.

But data is not the only consideration. If there is only data without computation, it’s entirely possible to create an application where everyone uploads their data, but how do you interpret this data? How do your users know the meaning of this data? If all computation is off-chain, then each application will have its computations, leading to many compatibility and combinatorial issues, making it difficult for applications to communicate with one another, so Ethereum's built-in computation is still required.

We are also recently concerned about computation. Scaling a thousand times is feasible, but there are many challenges. Developers often need to change their approach. A scalable ERC-20 token will differ significantly from the current ERC-20. We are currently considering this, and we have already started working on it. This includes using ZK to prove the execution of each block’s EVM; this is possible but takes a long time and must be done securely.

Since Ethereum is a rather complex system, we need to ensure security. If it isn’t secure, AI three years from now will certainly find every vulnerability, causing us greater pain. You might remember a complex incident that happened ten years ago in Shanghai. Four hours before our DevCon began, while I was sleeping, someone called my room saying there was an issue I needed to resolve. After coming down, I discovered that the Ethereum network had suffered a DoS attack.

We spent a long time understanding what a DoS attack was and how to resolve it. About three hours later, we resolved the issue, and announced some client PRs before the conference started. The conference began on time, and we felt we had won. However, two days later, there was a second attack, and five days later, a third attack. We still do not know who the attackers were, but one person spent almost an entire month discovering every DoS vulnerability in Ethereum's two clients. During that month, the Ethereum chain was almost unusable.

Some people still used it. I remember Augur's (an early decentralized prediction market project) ICO took place during the DoS attack. Because we identified all the vulnerabilities at that time, there were no more issues afterward. But it’s possible that AI could also find all the attack methods in our clients and cause us such pain again in two to three years. Therefore, we hope to create more secure clients and use new ways to verify that our code has no such problems before such incidents occur.

AI and Quantum Threats

Host: Since you founded Ethereum, I believe you and your team have always focused on two areas: scalability and security. In terms of scenarios, you’ve allowed the community to handle it, while your primary focus has been on scalability and security. However, you have also shared many thoughts on the explosion of AI and ideas about quantum computing. There are many different opinions in the media; could you share with everyone today how you see the future intersection of AI, quantum, blockchain, and Ethereum?

Vitalik:

In terms of Ethereum, these two issues can be handled but will require considerable effort. I can provide a completely non-technical analogy to explain. For example, imagine a country that has never seen rain, and no one there has ever heard of the concept of rain; houses in this country are completely unprepared to handle rainfall and will have many leaks. If it rains for the first time, maybe 5% will be completely ruined. But they feel there’s no problem since it never rains.

Then one day a scientist says: I have discovered the concept of rain, and it might start raining in 5 years or 10 years. I know how to build houses that can avoid this issue. Theoretically, we all know, but it still needs to be done; we must check every house, every school, every office, and every piece of infrastructure to identify potential problems and solve them everywhere. This process is very difficult.

Post-quantum and anti-AI will likely be similar in the Ethereum ecosystem. In the next 3 to 5 years, we know how to proceed. In fact, I first wrote an anti-quantum signature algorithm five years ago, in a rather obscure place on my GitHub. In 2017, I also wrote a smart contract, which was a hash-based signature algorithm. We know these things.

Why haven't they been used? Because their efficiency is lower. An elliptic curve signature is 64 bytes, whereas a post-quantum one may be around 2000 to 3000 bytes. So new technologies and approaches are needed. For instance, we cannot directly place every signature into the block; it would be too inefficient. We needed to use aggregation methods, meaning that nodes creating blocks need to take all the signatures and create a STARK to prove the existence of those signatures, but the signatures themselves can be removed, and only the STARK can be placed directly in the block.

In terms of AI, our current strategy is to use formal verification to ensure that the ZK code we create meets security needs. These are all achievable. At the L1 level, it may be somewhat easier since there is only one L1 and it is a limited entity. But at the application layer, there are many different applications, each with numerous dependencies, making the application layer more complex. Yet, this is also something we all need to address. Within our foundation, this is a topic of significant focus.

Host: If we incorporate these post-quantum solutions, will it affect gas fees?

Vitalik:

By default, it will be higher. If our only action now is to replace elliptic curve signatures with hash-based or lattice-based signatures, then the gas for each transaction would increase from 20,000 to possibly 200,000, and we would lose about ten times the transaction capacity. Thus, we need to use aggregation methods, meaning that the block-creating node should convert all signatures issued by users into a single STARK.

For example, there could be a thousand users, each sending one transaction, and each transaction containing a signature, with each signature possibly being three thousand bytes, totaling three MB. A node can create a single STARK that can prove the existence of those three MB of signatures, but if the STARK is well done, the size could be just 256KB or 128KB, regardless of how many signatures there are, it would still be 256KB. Theoretically, we all know how to solve these issues, but there is a difference between knowing in theory and executing. So the next step is to implement.

Message to Hong Kong

Host: Everyone is also very concerned about your return. Especially the Chinese community is very happy to see you coming back often. Finally, we would like to ask you to send a message to our Chinese community regarding the construction of Ethereum, in encouraging builders of the entire community, and any thoughts on future development, especially since today marks the official opening of the ETH Hong Kong Community Hub.

Vitalik:

There are many things to do in the new Ethereum ecosystem, and the past two years have also been an opportunity to rethink many matters, as we have mature ZK technology and AI. AI has reduced the cost of writing any code by ten times, now ordinary people can develop some of their own projects.

However, even with more tools available to meet the demands of a new world, the demands of the new world will also increase. The world is becoming more complex, and everything is accelerating. I hope everyone can participate in this process and start to rethink from scratch. For example, in the financial field, do not think from the current Ethereum ecosystem or the current Ethereum protocols, but think from the core philosophy behind Ethereum, from the reasons that made you first interested in Ethereum, about what this world currently needs.

It's possible that the answer might not necessarily include Ethereum technology; it could involve a part of Ethereum, alongside other elements related to AI, or may relate to ZK-SNARKs (a type of efficient zero-knowledge proof technology) or secure hardware. Because next door in Shenzhen, there are many hardware developments, including open-source hardware and many open-source AI projects. We can start combining different fields to create many new things.

I hope everyone can think more and start doing things that are completely different from what Ethereum was three years ago.

Host: Returning to the original intention, every good project is meant to solve problems for humanity. Thank you, Vitalik.