On Sunday, Ethereum Foundation researcher Dankrad Feist submitted a proposal known as Ethereum Improvement Proposal 9698. The plan involves allowing the blockchain's gas limit to increase automatically over the course of four years. Feist, known for his work on scaling the Ethereum network, including the 'danksharding' component of the Dencun update, has introduced a deterministic exponential schedule. This schedule, which will be integrated into client defaults, incrementally raises the gas limit by a small, predetermined amount every epoch, allowing current validators to keep their systems up to date without needing sudden upgrades.
If the proposal is approved and implemented, the gas limit would rise from 36 million units to approximately 3.6 billion units. This increase could support around 6,000 simple transfers per block and over 2,000 transactions per second, a significant improvement from the current 15-20 transactions per second.
To implement this change, client teams would need to integrate the code and keep the setting enabled by default, although operators would have the option to manually override it. Feist explained the motivation behind the proposal, stating that the current gas limit mechanism, which relies on voting by miners and operators, lacks coordination and predictability. By introducing a predictable exponential growth pattern as a client default, the proposal aims to create a sustainable and transparent gas limit trajectory that aligns with expected advancements in hardware and protocol efficiency. The proposal is set to be activated at Beacon-chain epoch 369017, which is expected to occur around June 1.
Following this, the proposal will be reviewed by client maintainers and core developers. If approved, Ethereum's throughput could see significant changes by 2029 without requiring a hard fork. The proposal comes at a time when Ethereum is facing increased competition from other blockchains offering faster settlement times and lower fees, which has led to a decline in financial activity and user base on the network.