Ethereum’s cryptography roadmap is shifting as new proof systems make traditional hash functions more practical for the network’s Layer 1. Vitalik Buterin acknowledged Poseidon’s past contributions to privacy proofs, while Ethereum Foundation researcher Justin Drake said the Foundation is now moving toward SHA or BLAKE. Drake linked the change to faster binary-field SNARKs and ongoing post-quantum research.

Binary-field SNARKs can make SHA2 and BLAKE2s practical inside zero-knowledge proofs, reducing the advantage Poseidon once held. The shift matters beyond performance: SHA2 and BLAKE are long-standardized, widely audited hash functions with decades of cryptanalytic scrutiny, whereas SNARK-friendly designs like Poseidon are far newer and have faced ongoing security analysis, including third-party cryptanalysis of their algebraic structure. Ethereum targets a production-grade leanVM in 2027, with broader consensus and execution deployments potentially following in 2028.

Poseidon’s Role in Ethereum

Buterin said years of work went into Poseidon hashes, which enabled fast client-side SNARK generation for privacy protocols. However, he highlighted progress in general-purpose STARKs, noting that current systems can prove broadly batched computations with mid-three-digit overhead — a figure he said could soon reach double-digit levels or lower.

Drake described Poseidon’s development as an eight-year effort backed by eight-figure funding. The Ethereum Foundation began researching SNARK-friendly hashes in 2018; Poseidon emerged in 2019 and later supported zk-rollups and zkVMs. If traditional hashes can now be proven efficiently inside SNARKs, the specialized designs built specifically to avoid that cost become less necessary on Ethereum’s roadmap.

Binary Fields Change the Equation

According to Drake, newer SNARK designs can use traditional hashes without SNARK-friendly functions. He named SHA2 and BLAKE2s as examples that can now match Poseidon inside a SNARK.

Drake explained that binary fields use the prime number 2 and fit traditional hash operations naturally, in contrast to large prime fields, where bit manipulation requires more work. Researchers can prove 1 million hash calls per second on a laptop, he said, with roughly 100 times the overhead of native CPU Boolean computation.

Ethereum’s Post-Quantum Roadmap

Drake credited Jim and Ben with Binius in 2023, and Ron, Benedikt and William with Flock in June. Emile and Thomas are developing binary-field systems, he added.

The roadmap targets a production-grade leanVM in 2027, with consensus, data availability and execution-layer deployments potentially following in 2028. Drake also cited setbacks involving the lattice-based HAWK and isogeny-based SQIsign schemes, both of which appeared in NIST’s Round 3 signature competition. Quantum computers, once sufficiently powerful, would threaten the elliptic-curve signatures Ethereum relies on today, which is why post-quantum signature research is part of the roadmap; hash-based approaches are considered resistant to such attacks.

He noted that SNARK.fast has exceeded 1.8 million BLAKE3 proofs per second on an M3 Max, and referenced the ECDSA.fail, zk.golf and autoresearch efforts. The tenth Ethproofs call will focus on binary fields at 2 p.m. UTC.

Drake said hash-based cryptography has won for blockchain post-quantum signatures, and that SNARKs can compress signatures — including k-of-n thresholds and complex multisigs — into a single block proof.