From "Ledger" to "Cryptographic World Computer": Vitalik Buterin's Vision for Ethereum's Transformation and Three Shifts Financial Institutions Must Prepare For

Key Points of the Talk

On September 23, Buterin spoke online at the 12th Global Blockchain Summit in Shanghai. In his view, blockchains are open to anyone, neutral and tamper-resistant, but because transactions are public, privacy is a weak point. When Bitcoin launched, the fact that transactions were not directly tied to real names was considered a reasonable form of privacy, but with advanced data analytics and AI now widespread, he said, that is no longer sufficient. Traditional blockchains also face throughput limits, because every node repeats the same computation to confirm it is correct.

What Zero-Knowledge Proofs Change

As the technology that resolves both constraints at once, he named zero-knowledge proofs (SNARKs, and their quantum-resistant counterpart STARKs) as the biggest change of the past decade. If one participant runs a computation and produces a proof that it was done correctly, others only need to check the proof rather than redo the computation. Nor is it necessary to disclose all the information used in the computation. He stressed that bringing this technology into Ethereum has moved beyond the concept stage into concrete proposals. The example he cited, EIP-8288, aggregates large numbers of signatures and proofs into a single proof in the mempool, the waiting area where transactions sit before entering a block. He created it with Thomas Coratger on June 3. It is still in draft, however, and the target is the fork that follows the next major upgrade, "Hegotá."

From "Who Can Send" to "Who Can See"

The core of the talk was the expansion of what cryptography protects. In Bitcoin and Ethereum to date, code has decided "who can send which asset, and when." Going forward, he said, we are entering an era of "programmable cryptography," in which code also decides "who can see which information." Fully homomorphic encryption (FHE), which allows computation on encrypted data, and obfuscation techniques that run programs while hiding their internals are also approaching practical use faster than expected, he said. On cost, he explained that the processing overhead of zero-knowledge proofs has already fallen below 10 times that of ordinary computation for some AI inference and certain hash computations, while FHE is at about 250 times but falling quickly.

The Chain's Role Becomes "Minimized"

Where computation happens is also changing. Much processing happens on the user's device; transactions pass through the mempool, are aggregated, go through block building by multiple participants, and are finally recorded on-chain. The whole system, including not just Layer 1 but Layer 2s, the mempool and user devices, becomes one large cryptographic network, and the portion every node must process shrinks. He described this as a shift from a "ledger" to a "cryptographic world computer." On speed, he said Ethereum currently takes 12 seconds per block and about 16 minutes to finality, but block times are expected to fall to 4–8 seconds and finality to 8–32 seconds.

AI Changes the Premise of Security

He named AI as a factor that made these changes possible. Until now, bugs became unavoidable as cryptographic code grew larger, limiting what could be built safely to relatively simple systems. Combining AI with "formal verification," a technique that mathematically proves a program behaves according to its specification, makes it possible to automatically prove whether specifications written separately by different people or AI agents describe the same thing, sharply narrowing what humans need to check. The Ethereum Foundation's protocol team aims to make all three layers, execution, consensus and data, quantum-resistant by December 2029, and every upgrade after Hegotá is expected to rely on advanced cryptographic features.

[Business Development Insights]

1. The choice around privacy will shift from "public chain or private chain" to "designing in code who sees what." If selective disclosure, such as showing transaction details only to the parties involved plus regulators and auditors, becomes technically feasible, the main reason financial institutions have chosen closed chains weakens. The ability to design information access rights in advance as business rules will become a competitive edge for on-chain financial services.
2. The move to quantum resistance has become a practical management issue for long-dated financial products. With Ethereum targeting December 2029, tokenized bonds maturing after that date and custody built for long-term holding will need plans for migrating signature schemes. This creates demand for overhauling account structures and wallets so authentication schemes can be switched flexibly, a clear business opportunity for key-management and custody providers.
3. Smart-contract security assurance will move from manual audits to "proving specifications" by combining AI and formal verification. At that point, the "specification" defining what counts as correct behavior becomes an important governance document for financial institutions. Demand will grow for people and services that turn business requirements into verifiable specifications. That said, the cost forecasts are Buterin's personal predictions, and EIP-8288 is still in draft. Investment decisions should rest on continuous tracking of how implementation progresses.

[Sources]

Atarashii Keizai, "Vitalik says blockchains are shifting from 'ledgers' to cryptographic computation platforms" (Sept 25, 2026) https://www.neweconomy.jp/posts/611038

Wu Blockchain, "Vitalik Shanghai Speech: Blockchain Needs New Cryptographic Tools" (Sept 24, 2026) https://wublock.substack.com/p/vitalik-shanghai-speech-blockchain

YouTube, Vitalik Buterin's speech at the 12th Global Blockchain Summit https://www.youtube.com/watch?v=Vsa6elxxwGg

Atarashii Keizai, "Vitalik explains EIP-8288 using recursive STARKs to cut costs of quantum-resistant signatures on Ethereum" (Sept 10, 2026) https://www.neweconomy.jp/posts/606960

EIP-8288 https://eips.ethereum.org/EIPS/eip-8288

Vitalik Buterin, X post on EIP-8288 (Sept 9, 2026) https://x.com/VitalikButerin/status/2097711433073172837

Atarashii Keizai, "Vitalik: 60% chance the compute overhead of SNARKs, FHE and iO falls below 10x" (Sept 7, 2026) https://www.neweconomy.jp/posts/605662

Vitalik Buterin, X post on cryptography cost forecasts (Sept 6, 2026) https://x.com/VitalikButerin/status/2096589479305961591