Ethereum researchers outline "Lean Ethereum," a three-to-four-year overhaul of core protocol
Tech & Launches ·
Vitalik Buterin describes a multi-year collection of upgrades meant to replace nearly every major piece of Ethereum's protocol with scalable, quantum-safe alternatives.
Following researcher meetings in Berlin, which continued discussions held with client teams in Svalbard in April, Vitalik Buterin laid out what he calls the third major iteration of Ethereum, comparable in scale to the Merge. Rather than a single upgrade, the plan spans three or four years and touches verification, consensus, gas accounting, state structure, and client architecture.
Central to the plan is a shift toward verification through recursive STARKs rather than direct re-execution, with STARKs becoming an enshrined core component of the protocol. Consensus would move to a decoupled available chain and finality mechanism, aiming for one or two-round finality with what Buterin describes as theoretically optimal security properties that are simpler and faster than the current design. Quantum resistance is framed as a priority that has risen sharply, with a quantum-safe blobs design now described as urgent work already underway for months.
The most disruptive change concerns state. Present-day "dynamic state" would remain largely intact but scale only moderately, while a new, more restrictive but far more scalable state type would be added. Buterin sketches a possible 2030 Ethereum holding roughly 2 TB of dynamic state alongside 100 TB of this new-style state. The new format would suit ERC20 tokens, NFTs, and many DeFi use cases, but not more complex, "central" objects like Uniswap contracts or onchain order books, which occupy a small share of total state. Rewriting apps would not be mandatory, but Buterin says migrating something like an ERC20 token to a new UTXO-based storage design currently being explored could cut transaction fees by more than 10x.
The roadmap also elevates privacy and formal verification to first-class design goals rather than afterthoughts, with questions about quantum-safe, intermediary-free privacy transactions factored into mempool and state-tree design from the outset. Formal verification is described as extending to canonicalization efforts, including an evm-asm effort intended to serve as a canonical proof system for the EVM. Buterin also flags H-star, or Hegota, as likely Ethereum's last "pre-Lean" fork, with the following fork, I-star, marking the start of the Lean-era changes. The update, reported by wublockchain.xyz, points to an updated strawmap published at strawmap.org.
Left unresolved are the specific technical designs for the new scalable state types, which Buterin lists as still being explored, including keyed nonces, ring buffers, UTXOs, and statically accessible structures. No firm timeline beyond the three-to-four-year window has been specified, and the sequencing of forks after H-star remains to be detailed.