Vitalik Buterin has called for a hard ceiling on Ethereum proving costs per unit of block gas, adding a resource-accounting requirement to the push for faster zero-knowledge proofs. For independent prover operators, the challenge centers on how much computational work the most demanding permitted block could require, even as typical blocks process quickly.
In an October 15 post, Buterin outlined three formal-verification priorities: matching the mathematical representation of execution to Ethereum's specification, bounding prover cost relative to block gas, and securing the proof system itself. The cost requirement extends beyond speed benchmarks to address how Ethereum accounts for work and limits resource consumption per block.
Different Workloads, Different Demands
The distinction between execution and proof generation matters because a unit of execution gas can correspond to different amounts of proving work. Research from Nethermind found that cryptographic precompiles and certain modular arithmetic operations were relatively expensive to prove per gas unit, with costs varying across different prover implementations.
ZKsync recently demonstrated progress on the proving side. In an October 13 announcement, the company reported that a single NVIDIA RTX PRO 6000 Blackwell GPU proved 99 percent of 7,200 consecutive Ethereum mainnet blocks in under 10 seconds each, including witness preparation. However, a maximum-cost guarantee must cover every permitted workload, including blocks deliberately assembled to be expensive to prove.
For operators budgeting hardware resources, this distinction is critical. A typical block may fit within a small setup while a difficult but legal workload determines the reserve capacity needed to maintain consistent proving performance.
Defining Resource Limits
The current compute-cost analysis in EIP-7904 recommends no changes to execution gas costs. The proposal targets 100 million gas per second on optimized native execution clients and attributes its conclusion to optimizations enabled by Block-Level Access Lists. However, EIP-7904 expressly does not address ZK-SNARK proof generation, which imposes different computational demands than CPU execution.
The Ethereum Foundation's January worst-case benchmarking framework relates required proving throughput to block gas limits and proving deadlines. Under this approach, difficult blocks must still fit within intended hardware and time budgets. Proof generation involves fixed overhead, aggregation can be nonlinear, and hardware configuration and security parameters affect performance.
EIP-7904 has a November 1 review deadline for its CPU analysis. For independent provers, the next consequential step is establishing a bound tied to explicit workloads, implementations, security settings, and hardware assumptions. This would convert a gas allowance into a dependable limit on the proving resources operators must be prepared to supply.
