Live
USD · 24h
Ethereum & Altcoins

Ethereum’s transition to Verkle trees

Verkle trees reduce Ethereum proof sizes from 150 KB to just 1 or 2 KB using polynomial commitments. This technology enables stateless clients and is targeted for delivery during the Hegota upgrade in Q4 2026.

Ethereum's transition to Verkle trees

Merkle limitations

Ethereum’s Merkle Patricia Trie requires 150 KB witnesses per proof. This size forces nodes to carry heavy data loads as the state expands. Verkle trees reduce these witness sizes to 1 to 2 KB per proof. Verkle trees use polynomial commitments like KZG to compress data. This technology enables stateless clients. Stateless clients verify transactions using witnesses instead of a full state database. This transition reduces storage requirements from hundreds of gigabytes toward levels manageable on consumer hardware. The Merkle Patricia Trie relies on hexary branching and Keccak-256 hashing. This design makes Ethereum flexible but not optimal for scalability. Verkle trees use a high branching factor, typically 256 children per node, which allows for very shallow trees. A blockchain stores a cryptographic commitment to a state. This commitment uses a root that changes if any element changes. For an element, the proof contains sibling hashes along the path. Merkle trees scale logarithmically, but they do not scale efficiently enough in practice. The Merkle Patricia Trie uses separate tries for accounts and storage, creating deeper and more complex paths. The current Merkle implementation requires recomputing multiple trie nodes for every update, creating pressure on disk I/O and CPU hashing. One proposed Verkle scheme uses extension nodes to handle address and storage key components. This approach prevents attackers from filling storage in locations that result in excessively long branches.

Property Merkle Verkle
Proof size $\log(n)$ $\sim$constant
Model hash polynomial
Verification hashing pairing

The Hegota roadmap

The Hegota upgrade targets Q4 2026 to deliver Verkle Trees, FOCIL, and account abstraction. Core developers moved these features from the H1 2026 Glamsterdam fork to Hegota to manage scope. Glamsterdam instead prioritizes ePBS, parallel execution, and a gas limit increase to 200 million. Verkle Trees produce proofs roughly 90 percent smaller than Merkle proofs. This reduction enables stateless clients. A node verifies any transaction without storing the full state. If you follow the roadmap closely, you see that this compression targets the network’s growing state. Validators capture the direct benefit because running a node becomes cheaper, which may compress staking yield from the current 2.86 percent APR. However, dApp developers face migration costs as they refactor code for archive-proof integration. I favor the move toward Verkle Trees for its scalability benefits, but current implementations remain 40 percent slower than Merkle Patricia Trees. The Hegota upgrade bundles Verkle Trees with state expiry and stateless clients. State expiry moves inactive state into a separate archival layer. dApps relying on old contracts must integrate with archive providers. Wallets must source proofs for inactive accounts. This transition enables more independent operators to enter the staking set, which improves censorship resistance and finality guarantees. Developers test Verkle Trees on the Beverly Hills and Kaustinen testnets. The Beverly Hills testnet focuses on EL clients and the Verkle Tree spec, while Kaustinen implements changes to the Ethereum Consensus Layer alongside Verkle Trees. Testing includes chain reorganizations, but developers have not yet assessed JSON RPC performance or shadow fork behavior.

Post-quantum security concerns

A rival proposal, EIP-7864, suggests a binary state tree instead of Verkle Trees. This binary tree uses hash functions to provide post-quantum security. Verkle Trees rely on KZG commitments, which use elliptic curves. A sufficiently powerful quantum computer could break these elliptic curve mathematics. This risk drives interest in the binary tree alternative. The Hegota upgrade remains the target for Verkle Trees, but developers continue testing. The current transition requires nodes to maintain both Merkle Patricia Trees and Verkle Trees in parallel until they synchronize, at which point the consensus rules switch to the new tree. The conversion from Merkle Patricia Trees to Verkle Trees takes around 18 hours on the Erigon implementation and much longer on other clients. One concern involves whether all clients can use Verkle Trees without adopting a database architecture similar to Erigon. If they do not adopt Erigon’s architecture, they must rely on nodes to gossip the Verkle Trees, which poses a risk. Nethermind currently works on a similar database architecture to mitigate this. Will the network revert to a binary tree if proving-time engineering fails to meet block production deadlines under load?

Join the discussion

Leave a Reply

Your email address will not be published. Required fields are marked *