Zero-knowledge rollups came on to the scene using specialized provers. For rollups aiming to mirror Ethereum functionality, these provers required a zero-knowledge proof system to replicate the behavior of the Ethereum Virtual Machine: a “zkEVM”. The zkEVM was responsible for proving that the rollup blocks correctly built rollup blocks. However, they suffer from high development overhead: every time the EVM changes, the zkEVM must change too. To simplify this process, one can provide the EVM client, and the rollup blocks, as input to a more general purpose zero-knowledge Virtual Machine: a “zkVM”. This ensures that the rollup correctly computes blocks using exactly the same client that could be used for Ethereum itself. In this talk, we’ll review the challenges of changing from a zkEVM to a zkVM after reviewing these designs at a high level. We’ll also comment on the cost of these systems and the performance differences. We’ll end the talk by outlining future directions for zero-knowledge rollup proof systems.