Vitalik's Hong Kong Speech: Looking Ahead to the Next Five Years of Ethereum, Scaling, Quantum Resistance, and zk Verification Moving Forward

Author | Vitalik

Translation | Wu Talks Blockchain

Vitalik Buterin systematically elaborated on the future evolution directions of Ethereum over the next few years during his keynote speech at the Hong Kong Web3 Carnival. He positions Ethereum as a global infrastructure that combines a “public data publication layer” and a “shared computation layer,” emphasizing that its core value does not lie in pursuing maximum speed, but in security, decentralization, verifiability, and long-term robustness. The speech focused on Ethereum’s short-term scaling paths, including increasing gas limit, advancing zkEVM, optimizing data availability and account abstraction; it also discussed medium- and long-term goals such as quantum-resistant signatures, formal verification, leveraging AI to enhance protocol proof capabilities, and enabling mobile phones and IoT devices to perform chain verification.

Below is the full transcript of Vitalik’s speech, optimized with GPT for better reading experience:

Good morning everyone!

Where is Ethereum headed? I believe that over the past few years, we have seen many significant changes at the theoretical and ecosystem levels. At the same time, many changes have also emerged outside the Ethereum ecosystem, including the possibilities brought by artificial intelligence, the potential rapid realization of quantum computing, and advances in formal verification, cryptography, and zero-knowledge proofs.

I think one important thing we have been doing is rethinking what truly matters: Why use Ethereum? Why Ethereum? What features make it worth using? And how can we incorporate these features into the Ethereum protocol we have built, while planning the development path for the next five years?

I believe Ethereum has two main functions.

First, Ethereum is like a public bulletin board. It is a place where all applications can publish messages, and everyone can see the content and order of these messages. These messages can be transactions, hashes, encrypted data, or many other things. In fact, many applications see Ethereum as a place to publish data, which can then be interpreted, decrypted, and computed by other types of protocols.

Second, it is computational capability. Essentially, Ethereum allows you to have shared digital objects controlled by code. These digital objects can be many things: assets, ERC-20 tokens, NFTs, and their significance is not limited to finance. ENS is an example, and even control rights over organizations are digital objects, like DAOs. You can do many things with them, and both points are highly valuable. For decentralized applications, they ensure autonomous security, verifiability, fair participation, and bring together all users.

Self-sovereignty essentially means that as a user, you can participate, verify, and ensure your own security, all built on your own infrastructure. You don’t need to trust any third party to run Ethereum, nor do you need to trust third parties outside Ethereum unless you choose to.

Therefore, verifiability and the ability to verify mean you can ensure the chain operates correctly and verify everything that happens. At the same time, it guarantees that anyone has the right to publish information on this public bulletin board. This is the core. We should see Ethereum as a technical module and think about all applications that can be built upon this technical foundation.

The most interesting applications are often products combining on-chain and off-chain components. This includes ENS and prediction markets. Prediction markets have on-chain components, such as assets created for each event and the trading of these assets; they also have off-chain components, such as oracles. In some cases, the design or order matching of prediction markets occurs off-chain. Additionally, privacy-related aspects are involved.

For example, for decades, people have been researching cryptographic protocols aimed at simplifying or achieving secure electronic voting. Many such protocols rely on a public bulletin board where information can be posted. In this case, encrypted votes are published. This approach ensures everyone can participate. Anything related to privacy must include both an on-chain component for publishing data and an off-chain component for interpreting data.

If data needs to be interpreted, this process usually must be done through some off-chain private protocol. So, we talk a lot about L2. My idea of a “meaningful L2” is: copying a chain and enlarging it 100 times, making it more centralized, is not meaningful L2. Truly meaningful L2 is when, when examining an application, you ask: what is its off-chain component? Besides the framework, what else does it need? Then you build around these needs.

What does this mean for Ethereum? We need to expand data capacity, enabling more data to be published on-chain, and also improve peer-to-peer network capabilities. This has already appeared in recent hard forks, which included some improvements last year, but we still need to push further. Expanding computational capacity is also crucial, as it is part of Ethereum’s chain, helping different applications to compose and communicate with each other without intermediaries.

If you visit the roadmap page, you can see a plan for the next five years. The protocol’s short-term core goals are: first, short-term expansion, continuing to actively increase gas limit; second, beginning to advance zkEVM. zkEVM allows Ethereum to scale to larger sizes, perform more complex computations, while still making on-chain information easy to verify. Also, early preparations for the post-quantum era. We have been thinking about quantum computing issues for years and recognized early on that this is a real challenge. We already have some measures, and in the near term, we will further improve quantum-resistant designs and refine the entire roadmap.

Ultimately, we want Ethereum to remain secure and efficient even in a fully post-quantum era; at the same time, further improve the “building block” approach and strengthen privacy support.

Regarding short-term scaling, many EIPs will enter the next phase. For example, protocol-level hard forks and access list improvements can help achieve parallelization; gas re-pricing can improve efficiency, enhance system security, and make increasing gas limit safer.

ePBS will also make block verification and protocol validation safer and more controllable. There will also be improvements in protocol capabilities, such as enabling state downloads. EIP-7701 is an account abstraction proposal that is simple yet powerful. Essentially, a transaction is a series of calls, where one call can be verification, and another can be execution. This makes it easier for Ethereum to natively support smart contract wallets, including paymaster transactions, quantum-resistant signatures, and privacy protocols.

This will broaden Ethereum’s use cases, supporting more functionalities. Quantum-resistant signature algorithms already exist and have been around for 20 years. We know what they are and how to build them. The issue is their efficiency. A post-quantum signature might occupy 2000 to 3000 bytes, while current signatures are only 64 bytes; it might also consume 200k gas, whereas verifying signatures on-chain now only costs about 3,000 gas.

We can use two types of signature schemes: one based on hashes, and the other based on lattices. Our approach is to introduce vectorization and incorporate it into the EVM. The core idea is similar: just as computers can efficiently run AI logic, often processing large amounts of data in parallel, the same approach can be applied to many post-quantum signatures. We are actively working to make signatures resistant to quantum attacks and more efficient.

Storage at the state level, account balances, and smart contract execution are relatively easier to scale; expanding storage is more difficult. There is still much work to be done in this area.

This is the overall direction for both short-term and long-term development, and it is the true path we want Ethereum to follow. Ethereum is not competing with high-frequency trading platforms, nor is it aiming to be the fastest chain. Its goal is to be the safest, most decentralized, and most reliable chain. This chain will stay online and be dependable.

One goal is to maximize secure consensus. That is, if the network itself is secure, it can withstand 49% node failures, and even in extreme cases where almost all nodes go offline, it remains as secure as possible, with features similar to Bitcoin.

If the network truly encounters issues, you should still have 33% security certainty—that’s the first part. The second part is formal verification of everything. We have begun actively using AI to generate proofs, verifying that the software running the protocol’s long-term version indeed has the properties it should. We have made some progress, which was impossible two years ago. AI is advancing rapidly, and we are making full use of this, while insisting on extreme simplicity to keep long-term protocols as straightforward as possible and to prepare for the future.

Therefore, a protocol must pass “offline testing.” If a protocol is worth using, even in extreme scenarios, you should be able to rely on it. This aligns with Bitcoin’s pursuit and is a goal that any system aiming for long-term existence must achieve. If you want to protect your digital assets long-term, you need to rely on something that can provide security over the long run—security that does not depend on a team’s continuous existence or constant work.

Simplified consensus combines the advantages of two approaches: one is Bitcoin-style, the so-called global chain; the other is BFT-style, which provides finality. It possesses optimal security features, quantum resistance, and fast finality.

Finality can be achieved within one to three slots. The entire chain is expected to reach finality in about 10 to 20 seconds, or even less. zkVM will enable you to verify the entire chain without relying on large computers running all operations yourself. Everyone should verify the chain before trusting it—your phone should verify it, and even IoT devices should verify it.

The zero-knowledge virtual machine zkVM is already fast enough to make “real-time VM execution” feasible. The goal for this year is to make it sufficiently secure; also, to start using zkVM with a small proportion of the network and gradually increase that proportion. One of its main methods is to verify the entire chain. By 2028, this will allow Ethereum to scale up without centralization, handling more transactions.

What is the vision behind these technologies? Ethereum is the “world computer.” It is both a global shared layer for making commitments, publishing data, recording actions, and documenting user-initiated activities; it is a platform for publishing data, where you can prove data has been published or not, and it is open to everyone.

It is also a global shared layer for ensuring the execution of high-value rules. The Ethereum protocol needs to be as robust as possible, while also being extremely easy to verify. I believe that in the future, with the help of AI, ensuring software security will become easier than we imagine.

But if people are unwilling to seriously ensure security, software vulnerabilities will increase tenfold, and attacks will become ten times more frequent. Therefore, Ethereum, as a blockchain, must first ensure security; second, ensure decentralization; and only after these conditions are met, provide this security to users as much as possible.

If you want to build decentralized applications, you need to guarantee users’ autonomous control, security, verifiability, and real participation. This includes financial applications, decentralized social platforms, identity systems, and some applications that are partly financial and partly non-financial, such as ENS, prediction markets, and more, covering many directions. Ethereum should make application development simple and default to having these capabilities—that is the core goal.

The roadmap for the next four years is designed around this goal.

Thank you all.