Buterin calls for ceiling on Ethereum proving costs

Vitalik Buterin proposed a hard ceiling on Ethereum’s proving cost per unit of block gas. The limit would apply to the hardest valid blocks allowed by the network, rather than to typical blocks alone.

The Ethereum co-founder outlined the proposal in an October 15 post congratulating ZKsync’s Airbender team. He identified three priorities for formal verification: representing execution according to Ethereum’s specification, limiting prover costs relative to block gas and securing the proof system.

The proposal would connect Ethereum’s gas accounting with the computing resources needed to generate zero-knowledge proofs. Independent prover operators would need to support the maximum proving workload permitted in a block if they are expected to produce proofs by a fixed deadline.

Ethereum’s gas system measures execution work and limits the amount of work included in each block. However, one unit of execution gas can require different amounts of proving work. In January 27 measurements, Nethermind Research’s Lin Oshitani found that cryptographic precompiles and some modular arithmetic and division operations were relatively expensive to prove per gas unit.

The study measured marginal proving time by changing the number of selected operations while keeping the surrounding execution context fixed. It tested SP1 version 5.2.3 and RISC0 version 3.0.4 with four NVIDIA GeForce RTX 4090 graphics cards. The tests used representative or best-effort worst-case inputs, excluded block-header validation and accidentally omitted the BLOCKHASH operation. The study also found that proving costs varied by system and that zero-knowledge cycle counts did not produce the same elapsed time for every operation.

ZKsync’s October 13 Airbender announcement reported that one NVIDIA RTX PRO 6000 Blackwell GPU proved 99% of 7,200 consecutive Ethereum mainnet blocks in less than 10 seconds each, including witness preparation. Every sampled block finished in less than 12 seconds. The announcement also described live ETHproofs configurations using two and four GPUs.

Those measurements covered the sampled blocks. A resource ceiling would need to define the workloads, proof systems, hardware, security settings and deadline included in the limit. The Ethereum Foundation’s January worst-case benchmarking framework links a block’s gas limit to the proving throughput required from a selected hardware setup.

The framework also notes that operation-level estimates cannot fully determine proof-generation time. Fixed overhead, aggregation costs, hardware configuration and security parameters affect the result. The execution clients and proof systems expected to participate would also affect the calculation.

One approach outlined by Nethermind would assign weighted zero-knowledge gas to operations that require more proving work and impose a separate per-block limit. That system could restrict a block’s proof workload before the block reached its ordinary execution-gas limit. Its accuracy would depend on how well the weights reflected difficult inputs.

A proving-resource limit would be separate from transaction fees and prover rewards. Repricing opcodes could charge more gas for transactions that use expensive operations, reducing the number of those operations that fit within a fixed block limit. Under EIP-1559, a higher gas charge at the same positive effective gas price would increase the transaction fee.

EIP-7904 recommends no changes to Ethereum’s gas schedule and sets a target of 100 million gas per second for optimized native execution clients. As of October 18, it was an Informational Ethereum Improvement Proposal in Last Call, with a November 1 review deadline. Its analysis covers CPU execution and does not specifically measure zero-knowledge proof generation.

EIP-7999, which proposes a multidimensional fee market, remains in Draft status. The optional execution-proof proposal EIP-8025 also remains in Draft status, provides no prover incentives and retains transaction re-execution when proofs are available. The Ethereum Foundation’s September priorities placed mandatory proofs later in the development sequence, with the order of related network upgrades under review.