Buterin says that Ethereum's biggest problems are the state tree and VM, suggests a deep fix

Buterin offers binary state trees and the ultimate RISC-V VM transformation to improve Ethereum’s proven performance and simplicity of implementation.

Vitalik Buterin proposed implementation layer changes that could fundamentally change Ethereum’s core architecture. The founder of the project stated that deep changes to the network state tree and virtual machine were necessary to overcome what he described as the biggest proof-of-chain obstacles.

In a detailed post at X, Buterin said that the state tree and VM together account for more than 80% of the constraints that affect proof efficiency, calling them “essentially mandatory” goals if Ethereum wants to enable client-scale and zero-knowledge proof use cases.

Ethereum repair

He pointed to EIP-7864, a proposal by Guillaume Ballet and others that replaces Ethereum’s current hex tree based on the Merkle Patricia Tree with a binary tree built on a more efficient hash function. According to Buterin, the change will shorten Merkle’s branches by about four times, reducing bandwidth requirements and making customer branch checks significantly cheaper.

Buterin added that this could reduce the data costs for tools like Helios and information retrieval systems by 4 times. Efficiency can be improved by 3-4 times only from shorter branches. He expects that if Ethereum switches to hash functions like BLAKE3, which is three times more efficient than Keccak. Meanwhile, a variant of Poseidon could offer up to a 100-fold improvement, although he noted that further security work would be required.

The proposed binary design also groups storage slots into “pages” of 64-256, allowing for more efficient loading and editing of contiguous storage, potentially saving more than 10,000 gas per transaction for programs that access storage slots early. Buterin explained that the prover-friendly state tree also allows zero-knowledge applications to build directly with Ethereum state instead of building independent trees, while simplifying the structure and allowing for metadata additions for future state completion mechanisms.

Besides the state tree overhaul, Buterin made a case for replacing the Ethereum virtual machine with a RISC-V VM, as he described the idea as long-term and non-consensus. But he expressed high confidence that it would be “an obvious thing” once the road map is finalized.

Map of possible expansion

The Ethereum founder said that a RISC-V VM would be more efficient to run, easier and simpler for providers, while noting that many existing provers are already written in RISC-V and interpreters can be implemented in just a few hundred lines of code. He started with a phased transition plan by using the new VM for preemption, then allowing developers to deploy contracts directly to the new VM, and eventually retiring the EVM to the compatibility layer written as smart contracts in the new system.