TLDR
- Buterin says state tree and VM drive over 80% of proving cost
-
EIP 7864 would replace hexary tree with binary structure
-
Shorter Merkle proofs may reduce bandwidth and gas use
-
RISC V VM proposed as long term EVM replacement
Ethereum’s next scaling phase may focus on its foundation rather than Layer 2. Vitalik Buterin has proposed deep execution layer changes to address proving efficiency. He identified the state tree and virtual machine as primary bottlenecks.
In recent posts, Vitalik Buterin said these components account for over 80 percent of proving costs. He argued that efficient zero knowledge proving requires structural reform. His roadmap includes state tree redesign and long term VM changes.
Vitalik Buterin Binary Tree Proposal Under EIP 7864
At the center of the state overhaul is EIP 7864. The proposal would replace Ethereum’s current hexary Merkle Patricia Tree with a binary tree. The change aims to reduce proof size and improve verification efficiency.
Binary trees would produce Merkle branches about four times shorter. This reduces bandwidth needs for light clients and privacy tools. It may also lower costs for systems such as Helios and private information retrieval.
Buterin said the new design would group storage slots into pages. Each page could hold 64 to 256 slots, or about 2 to 8 kilobytes. This structure allows related storage to load and update more efficiently.
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) March 1, 2026
He noted that many decentralized applications access adjacent storage slots. Under the new model, such patterns could save over 10,000 gas per transaction. The redesign also reduces variance in access depth between small and large contracts.
The proposal includes replacing keccak with more efficient hash functions. Options mentioned include blake3 or Poseidon variants. These changes aim to improve proving speed and make the state tree more prover friendly.
Vectorized Math Precompile for Short Term Gains
Before larger VM changes, Buterin proposed a short term upgrade. He suggested adding a vectorized math precompile to the Ethereum Virtual Machine. He described it as “the GPU for the EVM.”
This precompile would allow 32 bit or 64 bit operations on lists of numbers simultaneously. It could accelerate cryptographic workloads such as hashes and STARK validation. Buterin said this may improve performance by 8 to 64 times.
He also linked this upgrade to future quantum resistant signatures and fully homomorphic encryption. The proposal seeks to reduce reliance on complex workarounds. It aims to improve raw execution within the current framework.
Long Term Vision to Move Beyond the EVM
Buterin also outlined a long term plan to replace the Ethereum Virtual Machine. He suggested a transition toward a RISC V based architecture. He argued that many provers already operate in RISC V environments.
He said Ethereum has grown more complex due to special case precompiles. He noted that developers often avoid using the EVM directly. According to Buterin, a simpler and more efficient VM would better serve Ethereum’s general purpose design.
The roadmap proposes phased deployment. RISC V could initially power precompiles. Later, users could deploy contracts in the new VM. Eventually, the EVM could operate as a compatibility layer written in the new architecture.
Buterin said Ethereum would function adequately with incremental upgrades. However, he expressed strong conviction that deeper changes will become clear once state reforms mature. His proposal shifts attention from rollup capacity to Ethereum’s core execution design.
