Vitalik Buterin Proposes ‘Partially Stateless Nodes’ to Scale Ethereum Without Sacrificing Decentralization
Ethereum co-founder Vitalik Buterin has introduced a forward-looking proposal aimed at addressing one of Ethereum's longest-standing challenges: scalability. His solution—partially stateless nodes—may offer a way to raise Layer-1 gas limits significantly without compromising the network's decentralization or node accessibility.
In a newly published roadmap, Buterin outlines several short- and medium-term ideas to ease the strain on Ethereum full nodes, which currently store enormous amounts of data—around 1 terabyte for state and roughly 500 gigabytes of historical data. Increasing the gas limit traditionally means more data to process and store, which can make it harder for individuals to run a full node—an essential aspect of Ethereum's decentralized ethos.
To address this, Buterin reiterated the importance of proposals like EIP-4444, which would limit nodes to storing only the most recent 36 days of historical data. This change could drastically reduce disk space requirements and make running a node more accessible.
Beyond this, Buterin recommends prioritizing the development of a distributed history storage solution, allowing the wider Ethereum network to maintain long-term data accessibility without burdening individual nodes.
On the medium-term horizon, Buterin emphasizes a shift toward stateless verification—a method that allows nodes to verify transactions without storing complete Merkle trees. This could reduce node storage needs by nearly half, making Ethereum lighter and more efficient.
However, it’s Buterin’s partially stateless nodes concept that could be the real game-changer. These nodes would not store the full Ethereum state but instead retain a relevant subset of data. Using technologies like zkEVMs and stateless verification, they would still be able to fully validate transactions and blocks. If successfully implemented, this design could pave the way for 10x to 100x increases in the L1 gas limit, significantly boosting throughput while preserving the ability for users to participate in the network in a "trustless, censorship-resistant, and privacy-friendly" manner.