
Vitalik's Speech at Wanxiang Blockchain Week: Blockchain is Ending the Era of "Everyone Having to Repeat Everything"
Host: Distinguished guests and friends, good morning everyone!
Welcome to the 2026 Shanghai Blockchain International Week and the 12th Global Blockchain Summit, hosted by Wanxiang Blockchain Lab.
Twelve years ago, when the first summit launched here, blockchain was still an obscure technical term. Over these twelve years, we have witnessed its journey from a niche geeks' community to the mainstream spotlight, from experimental exploration to industrial implementation, and from a singular technology evolving into the foundational infrastructure driving the digital economy. Twelve summits have not only documented every leap in this industry but have also become a crucial global stage for intellectual exchange in blockchain, collaborative standard-setting, and ecosystem connectivity.
Today, we gather once again in Shanghai under the theme "Intelligent Chain Symbiosis" to explore the future landscape of the deep integration between blockchain and artificial intelligence. "Intelligence" represents the cognitive revolution brought by AI, while "Chain" symbolizes the trust foundation built by blockchain. When intelligence meets trust, and algorithms meet consensus, we believe this is not merely a simple technological addition, but a brand-new paradigm shift that will redefine the boundaries of finance, the value of data, and the modes of collaboration.
On behalf of the organizing committee, I would like to extend our warmest welcome and sincerest gratitude to every distinguished guest present today!
At the same time, we extend our special thanks to the Diamond Sponsor of this summit, Qtum; Strategic Partner, Wanxiang Blockchain City; Gold Sponsors Arkreen, Boundary Intelligence, Blue Elephant Intelligent Link, Frontier Technology Research Institute, and Yuanbi Technology; as well as the Exclusive Dinner Sponsor, Sui Foundation, for their unwavering support of this event. Twelve years of persistence, twelve years of iteration. Today, let us turn the page together to begin the next exciting chapter.
Ladies and gentlemen, today's first keynote speaker for the summit is about to take the stage.
It is Ethereum co-founder Vitalik Buterin. Over the past twelve years, he has grown from a teenage geek into a spiritual icon of the industry. The Ethereum platform he created inaugurated the smart contract era and defined a series of narratives that have guided the development of this sector.
He will join us remotely to share his latest thoughts on blockchain. Let's give him a warm round of applause as we welcome Vitalik!
The full speech content is as follows:
Vitalik Buterin: Good morning everyone. I am going to discuss a relatively complex technical topic today, so I will be speaking in English.
Today I want to talk about a macro-level topic. We have been discussing blockchain and cryptocurrency for over 15 years. For instance, blockchain-related activities across Shanghai alone have continued for more than a decade. Since Bitcoin emerged in 2009, superficially, the tools and technologies we are building may still appear similar. However, in reality, there have been significant differences over the past 15 years, which I believe are crucial.
One reason is that as users, this means the types of assets you receive and the types of costs you incur are vastly different, diverging from our previous expectations. Even our perspective on goals and the objectives we aim to achieve have changed significantly compared to 10 to 15 years ago.
Fifteen years ago, blockchain was our primary tool. Bitcoin combined blockchain, signatures, and proof-of-work. Ethereum later added smart contracts and transitioned to proof-of-stake in 2022. These were the tools we used at the time. Thus, the characteristics you obtain stem directly from using blockchain itself.
First, blockchain is open. Anyone can freely participate. It is global, reliable, secure, and neutral. No centralized entity can intervene in your applications and stop them from running, which could lead to incorrect outputs.
Zooko, the creator of Z-Cash, often uses a phrase comparing public blockchains to Twitter for bank accounts. Back in the 2009 whitepaper release, I believed the fundamental answers to privacy concerns—such as personal names not being directly linked to transactions—would be acceptable given strong privacy safeguards. However, with today's incredibly powerful data analytics and AI capabilities, that answer is obviously outdated. Meanwhile, people want to do more with smartwatches and expect greater performance from them. People are not only interested in payment functions; we also want payments to possess stronger, superior features. Therefore, both sides desire further advancement. In many areas, we indeed need to push further.
Thus, the biggest changes have occurred in the technical realm. The transformation we've experienced over the past decade is the rise of zero-knowledge proofs, including SNARKs, followed by the quantum-resistant SNARK variant—STARKs. The role of these constructs is to allow you to verify a computation. One person runs the computation, generates a proof, and everyone else verifies it. This approach makes the verification process significantly more efficient.
Meanwhile, it does not even require disclosing all information—that is, all the data used for the computation. Therefore, a core component of blockchain technology relates to this issue of verification. We want a system where everyone can participate, and where everyone is assured that everything happening adheres to established rules.
In the beginning, the technology required everyone to re-execute all operations, thereby sacrificing scalability and privacy. SNARKs and STARKs changed these two aspects. They have been discussed for a decade, but they are gradually becoming reality year by year. I believe the biggest change this year is that all roadmaps are shifting toward these technologies. They are no longer distant future concepts or mere roadmap items. There are now EIPs, such as EIP-8288, which we hope will enter I-STARK (phonetic approximation) within two years, possibly integrating with AI or earlier. They act as a "scarecrow map," where every component will utilize STARK after the fork over the next two years, making them ubiquitous. This represents a major shift.
Another major change is that more advanced technologies are emerging. I believe these are also approaching maturity, such as Fully Homomorphic Encryption (FHE), which many are already utilizing. Additionally, obfuscation techniques and IO (indistinguishability obfuscation) technologies, long thought to belong to a distant future, are actually becoming increasingly viable in shorter timeframes, much faster than many imagined. Thus, we now possess more powerful tools. These advanced technologies bring us many benefits, one of which is parallelizing blockchain processing. Parallelization begins even before transactions are included in blocks.
FOCIL is an EIP transitioning into H-START (phonetic approximation). This marks the first practical instance of multi-party block construction on Ethereum. Consequently, blocks are no longer created by a single participant but co-created by 16 participants, with a final sequencer completing it. The goal is to provide stronger guarantees: if you send a transaction, it will be included in the next timeslot. Thus, no single entity has the power to delay it.
In addition to this, alongside EIP-8288 and EIP-828H (phonetic approximation), these indicate you deal with extremely large proofs. Rather than placing these massive, costly objects directly on-chain, we can instead handle them through large-scale parallel processing to prove all proofs. This process occurs off-chain, within the mempool and intermediate gaps. When you send a transaction and it goes on-chain, in that space, it replaces all other objects, proving their existence with just a single proof. Thus, numerous changes have occurred in the cryptographic technology field.
But I believe the true importance of these changes lies in fundamentally shifting our mindset regarding the objectives of such systems. First, a critical question is: what are we protecting? Bitcoin indicates that we are protecting a data asset—the "Bitcoin currency." If you can issue your own digital assets, we can protect them too by leveraging modern cryptographic assets and information. In Bitcoin and Ethereum, cryptography and code answer the question: "Who can send what?" Which assets belong to whom? Who can send which assets and when? We are now entering the era of programmable cryptography, where code determines who can see what. Thus, it is not only about who can send, but also who can view which information? Visibility is what is becoming programmable. Where does computation occur? Fifteen years ago, blockchain was merely blockchain.
In the next five years, I believe this acceleration will largely continue. We are accelerating specifically in this regard: rather than simply possessing a blockchain, we embrace the entire pipeline—a vast cryptographic network comprising the complete chain. Users perform substantial computation locally, completing most work before data leaves their devices. Subsequently, transactions pass through a pool. Pools come in various types with distinct privacy attributes. Then, it gets aggregated involving multiple types of participants, followed by block construction, and finally recorded on-chain. Currently, under this mechanism, identical transactions on Ethereum are reprocessed by each node. On Starknet, a proof is verified many times.
Finally, what are we sacrificing? I think it's a valid point: if we sacrifice scalability, we also sacrifice latency. Centralized systems can offer latencies of 100 milliseconds, sometimes up to 300 milliseconds globally. Bitcoin's latency is around 600 milliseconds—or roughly 590 milliseconds—with confirmation taking about 10 minutes. Meanwhile, Starknet currently requires 12 seconds to generate a block and takes 16 minutes for final confirmation. However, in the future, even theoretically, the C language layer is expected to reduce this to 4-8 seconds per block, with 8-32 seconds for final settlement. And scalability—this is a revolutionary shift. Blockchain architecture will no longer always demand such heavy compromises. In terms of scalability, it sacrifices not general-purpose scalability, but highly scalable computational capacity. If you impose no limits on this system, provide no guarantees to regulate its computational behavior, yet possess computing power that you can bundle, split into two parts, execute separately, verify independently, and run—before anything happens, even before entering the supply chain—then extreme scalability effectively becomes feasible, right? Therefore,
If you consider blockchain as a complete system, it includes L1, L2, and human resource pools. Human resource pools, L2s, and tools. All these systems have evolved into highly complex middleware layers that you or your AI agents can navigate. AI, specifically models like Q3.8, helps refine and optimize them. Essentially, every step in the workflow is being accelerated by AI. Prior to AI, we had zero-knowledge proofs but only applicable to very simple cryptographic operations. That was the limit. One reason is that if code exceeds roughly 100 or 1000 lines, bugs are almost guaranteed.
We believe we can leverage these methods for verification. By employing different types of redundancy, testing, specifications, and analysis, we can build larger and more complex systems. In the post-AI era, we can place virtually everything we do—including our algorithms and their associated overheads—into cryptography. Now, overheads of 100x or even lower are typically achievable. Therefore, for ZK-SNARKs applied to AI inference, we have achieved overheads of less than 10x. For certain hash computations, we also achieved less than 10x overhead. Then we have FLoC, which I estimate has roughly 250x overhead, but this number is rapidly decreasing. Previously, verifying all of this was entirely manual. Formal verification is now another major operational bottleneck, which I believe is one of the primary reasons L2s failed to meet expectations so far. This is because of inherent limitations: the degree to which you can protect them is constrained. Thus, you require redundancy, multiple verification systems, and security committees.
The new technology is formal verification. If necessary, you can even apply multiple specifications to a single implementation. You can express the intended work in different ways, using different individuals, agents, and teams, and then automatically and mathematically prove they all convey the exact same meaning. If they ultimately convey different meanings, the baseline shrinks drastically, allowing everyone to reference the divergence and better understand which side is correct. Therefore, tremendous progress is occurring across the Ethereum world, spanning both protocol and application layers. I encourage everyone to visit our website at drawmap.org to review the Ethereum protocol. Upon close inspection, you will see that every bit of hard work will contribute within roughly a year, integrating highly advanced cryptographic features, encapsulating everything within the state and virtual machine, aggregating results, formally verifying all components, and deploying quantum-safe signature algorithms. Collectively, these factors are transforming the theory from a simple blockchain into a broader network encompassing diverse layers. One of its advantages, I believe, is that AI not only helps us build all these components faster, but also builds them more securely.
Of course, security is a core focus, which we certainly prioritize on Starknet. However, I believe AI also makes it easier for everyone here to comprehend these complexities. Therefore, for developers or researchers, I strongly encourage trying to understand the different modules being constructed, grasping their distinct components, identifying which properties each component possesses, and determining if there are specific areas where you wish to or are able to contribute.
I believe contributing to such frontier fields, including technology and ecosystems, is rapidly becoming more viable and accessible to anyone. Specifically, those willing to learn, think deeply, invest effort, provide assistance, aid research, and help build are finding unprecedented opportunities. There are even ongoing public competitions aimed at refining and formally validating improvements to numerous algorithms.
Furthermore, this is an ideal moment to truly rethink the applications we've been building and develop entirely new ones. These are applications built on Ethereum. Five, ten, or fifteen years ago, many of them were not conceived with this premise: that these novel technologies are viable, that privacy is programmable, and that even complex code security can be guaranteed. Now that these possibilities exist, I encourage everyone to dive into development, build applications, integrate them with Ethereum and other decentralized networks, as well as real-world elements, precisely as you wish to create impact and drive improvement.
Thank you!















