Ethereum co-founder Vitalik Buterin has shifted the scaling conversation away from Layer 2 solutions and back to the protocol's core architecture, proposing a bold overhaul to address fundamental bottlenecks. Buterin argues that Ethereum's biggest long-term constraints are not rollups or blob capacity, but deeper architectural issues within the network's state tree and virtual machine.
According to Buterin, these two components account for more than 80% of the network's proving costs, a critical issue as zero-knowledge (ZK) technology becomes central to Ethereum's roadmap. "Today I'll focus on two big things: state tree changes, and VM changes," Buterin wrote, adding that both are "the big bottlenecks that we have to address if we want efficient proving."
The centerpiece of the proposal is EIP-7864, which would replace Ethereum's current hexary Merkle Patricia tree with a binary tree design. This change would produce Merkle proofs roughly four times shorter than the current structure, dramatically reducing verification bandwidth requirements. The new structure would also group storage slots into "pages," allowing applications that load related data to do so more efficiently, potentially saving more than 10,000 gas per transaction in some cases.
Buterin also suggested pairing the tree change with more efficient hash functions for further gains in proof generation speed. More importantly, the redesign would make Ethereum's base layer more "prover-friendly," allowing ZK applications to integrate directly with Ethereum's state instead of building parallel systems.
Even more ambitious is Buterin's long-term vision for Ethereum's execution engine. He floated the idea of eventually moving beyond the Ethereum Virtual Machine (EVM) toward a RISC-V–based architecture, which is a widely used open instruction set that could offer greater efficiency and simplicity. Buterin argued that Ethereum's increasing reliance on special-case precompiles reflects a deeper discomfort with the EVM itself.
A RISC-V-based VM could reduce complexity, improve raw execution efficiency, and better align with modern zero-knowledge proving systems, many of which already use RISC-V environments internally. In the near term, Buterin proposed a "vectorized math precompile," described as a "GPU for the EVM," which could significantly accelerate cryptographic operations.
However, not everyone is convinced Ethereum needs more deep-layer changes. Analyst DBCrypto criticized what he described as growing abstraction across the Ethereum roadmap, including new frameworks aimed at addressing rollup fragmentation. "Each additional layer," he argued, "increases complexity, introduces trust assumptions, and creates additional potential attack surfaces."
The tension reflects a broader debate over whether Ethereum should continue layering solutions on top of its existing design or rework its foundation. According to Buterin, Ethereum's architecture must evolve as zero-knowledge proofs move from a niche to a necessity, suggesting that the next phase of scaling may occur deep within Ethereum's core rather than on Layer 2.
