Ethereum Foundation abandons the Poseidon hash algorithm, shifting to SHA-2 or BLAKE2 to advance its post-quantum cryptography roadmap.

Ethereum Foundation core researcher Justin Drake announced in a post that Ethereum Layer 1 (L1) will abandon Poseidon, the SNARK-friendly hash function that has been dominant since 2019, in favor of traditional hash functions such as SHA2 or BLAKE2s. This reversal stems from a breakthrough in SNARK design, centered on "hash-friendly SNARKs" rather than the prior approach of "SNARK-friendly hashes". By natively aligning Boolean operations in traditional hashes with binary fields, proof performance for traditional hash calls in SNARKs has reached 1 million per second, with overhead of just around 100x—a stark contrast to the previously extremely expensive operations in large prime fields. Drake dubbed this "science-fiction-level cryptography" and named contributors to key research breakthroughs including Binius and Flock. The shift will push Ethereum’s hash-based cryptography to the peak of minimal assumptions, while drastically accelerating deployment: there is no need to wait for years of cryptanalysis maturity for Poseidon. The Ethereum Foundation’s (EF) post-quantum team is advancing rapidly, with a roadmap targeting a production-grade leanVM in 2027, and deployment of the consensus, execution, and data layers in 2028. Drake also noted that AI’s enhanced cryptanalysis capabilities have recently dealt successive setbacks to lattice-based schemes (HAWK) and isogeny-based schemes (SQIsign), while hash-based schemes are emerging as the leading candidates for post-quantum signatures in blockchains. The trend of open-source automated research is also accelerating; SNARK.fast has achieved 1.8 million BLAKE3 proofs per second.