Vitalik Buterin Proposes Binary State Tree and RISC-V to Replace EVM on Ethereum

Two Deep Changes to Ethereum's Execution Layer

Ethereum co-founder Vitalik Buterin on March 1, 2026, laid out a plan for two sweeping changes to the network's execution layer: replacing the current state tree with a binary structure and phasing out the Ethereum Virtual Machine (EVM) in favor of the RISC-V instruction set architecture.

«Now, execution layer changes. I've already talked about account abstraction, multidimensional gas, BALs, and ZK-EVMs. I've also talked here about a short-term EVM upgrade that I think will be super-valuable: a vectorized math precompile (basically, do 32-bit or potentially…» — vitalik.eth (@VitalikButerin), original post

According to Buterin, the state tree and virtual machine account for over 80% of the network's technical limitations. He acknowledged that many in the community shy away from these "deep" changes, preferring an incremental approach they view as more "pragmatic." Nevertheless, he argued these transformations are "practically mandatory" for client implementations that rely on proofs.

Why This Matters

The proposed overhaul targets Ethereum's two most fundamental components — how data is stored and how smart contracts are executed. Successful implementation would dramatically improve light client performance, slash computational overhead, and deepen Ethereum's compatibility with zero-knowledge proof technology. The binary tree transition also addresses growing concerns about quantum computing threats to existing cryptographic methods — fears that reignited interest in binary structures over Verkle Trees in mid-2024.

Binary State Tree: A Four-Fold Data Reduction

The state tree reform is anchored in proposal EIP-7864, which envisions replacing the current hexary Merkle Patricia Trie with a binary tree powered by a more efficient hash function.

The binary structure would shrink Merkle branches by a factor of four, significantly reducing the data footprint for light clients like Helios. Swapping to Blake3 or a variant of Poseidon as the hash function would further boost proof efficiency by 3x to 100x. Buterin noted, however, that Poseidon still requires additional security auditing.

This proposal evolves from earlier plans to implement Verkle Trees. The pivot toward binary trees gained momentum in mid-2024 amid concerns that current cryptographic approaches could prove vulnerable to quantum computers.

Replacing EVM with RISC-V: A Three-Phase Strategy

Buterin revisited the case for replacing EVM with RISC-V — an open instruction set architecture already used by most ZK provers. He outlined a three-step roadmap:

• Deploy RISC-V exclusively for precompiled contracts.
• Allow developers to deploy their own contracts on the new architecture.
• Fully retire the legacy EVM by converting it into a smart contract running on top of the new virtual machine.

Buterin stressed that Ethereum's core value lies in its universality, and if the EVM falls short of delivering on that promise, the network should address the problem directly by building a superior virtual machine.

The proposal has not gone unchallenged. In November 2025, researchers at Offchain Labs — the team behind Arbitrum — argued that WebAssembly (WASM) is better suited for Ethereum's long-term evolution. Their central contention: while RISC-V excels at ZK proofs, the "interface architecture" and the "proving architecture" need not be the same.

AI and the Pace of Ethereum Development

Beyond the technical proposals, Buterin addressed how artificial intelligence could accelerate the Ethereum roadmap. He was responding to an experiment by developer YQ, who used AI-assisted "vibe coding" to produce a reference implementation of the Ethereum 2030 roadmap — 702,000 lines of Go covering 65 roadmap items — in just a few weeks.

«Two weeks ago I made a bet with @VitalikButerin that one person could agentic-code an @ethereum client targeting 2030+ roadmap. So I built ETH2030. 702K lines of Go. 65 roadmap items. Syncs with mainnet.» — YQ (@yq_acc), original post

«This is quite an impressive experiment. Vibe-coding the entire 2030 roadmap within weeks. Obviously such a thing built in two weeks without even having the EIPs has massive caveats: almost certainly lots of critical bugs, and probably in some cases "stub" versions of a thing…» — vitalik.eth (@VitalikButerin), original post

Buterin flagged "massive caveats" — the rapid coding speed "almost certainly" means critical bugs exist, and some components may be mere stubs rather than fully functional implementations. Yet he emphasized that such results would have been unthinkable six months earlier, and the trajectory of progress matters more than the current state.

The Ethereum co-founder advocated for using AI not just to write code faster, but to strengthen security. Specific applications he highlighted include generating additional test cases, employing formal verification, and building alternative implementations of the same components to catch errors through redundancy.