The Ethereum Foundation is pivoting away from the Poseidon hash function for its future layer-1 (L1) architectural plans, opting instead for established hashes such as SHA or BLAKE. Ethereum researcher Justin Drake announced the strategic shift on August 13, marking a departure from an eight-year investment in specialized, proof-friendly cryptography.

Drake emphasized that the decision does not stem from any newly discovered weakness in Poseidon, nor does it necessitate an immediate migration. Instead, he explained that recent advancements in proof systems have fundamentally altered the performance trade-offs that once made Poseidon the more favorable option.

The Changing Landscape of Cryptographic Proofs

The original rationale for using Poseidon was tied to the high cost of proving standard hash functions within SNARKs (Succinct Non-Interactive Arguments of Knowledge). SNARKs generate compact proofs verifying that a computation was executed correctly. Many of these systems rely on arithmetic over large prime fields, a mathematical environment where the bitwise operations inherent to conventional hashes like SHA-256 and Keccak have historically been computationally expensive to prove. Poseidon was specifically engineered to be "SNARK-friendly" by aligning with the preferred mathematical structure of these proof systems, thereby reducing the proving overhead.

However, recent research has effectively flipped this dynamic. The Binius research introduced a binary-field construction that works natively with the 0s and 1s of digital computation, closely mirroring the bitwise nature of Keccak. Furthermore, a paper titled "The Flock," posted on July 29, demonstrates how newer techniques can be applied to efficiently process large batches of standard hash computations.

Performance Breakthroughs and a Design Reversal

The authors of The Flock paper reported significant performance gains. On a single M4 Max core, their system can prove approximately:

• 82,000 BLAKE3 compression evaluations per second
• 42,000 SHA-256 compressions per second
• 30,000 Keccak permutations per second

Scaling to ten cores, the system surpassed 660,000 BLAKE3 compressions per second. In a direct comparison within the paper, SHA-256 proving was shown to be more than nine times faster than Binius64. It is important to note that these benchmarks measure internal hash operations, which is a different metric from Ethereum's transaction capacity or production throughput.

Drake described the advance as a "reversal in cryptographic design." The focus has shifted from creating SNARK-friendly hashes to building hash-friendly SNARKs. This breakthrough means that conventional, time-tested hashes can now be integrated into Ethereum's design process without incurring their former performance penalty during proof generation.

Implications for Ethereum's Security and Roadmap

This pivot allows Ethereum to leverage cryptographic primitives that have benefited from years of public analysis, offering a greater degree of confidence in their security. For instance, SHA-256 is part of the NIST's Secure Hash Standard, and BLAKE2 has had a public informational specification since 2015, distinct from the newer BLAKE3 implementation benchmarked by Flock. The ability to use these mature standards without a speed trade-off aligns with Ethereum's security roadmap, which identifies protocol simplification as a key method for shrinking the network's potential attack surface.

One concrete architectural change this enables is part of Ethereum's plan to move from validator BLS signatures to hash-based leanXMSS signatures. The subsequent aggregation of their significantly larger data into compact proofs using a system called leanVM would be more feasible with cheap proofs of conventional hashes.

According to Drake's personal strawmap, a production-grade leanVM could be ready around 2027, with deployments across the consensus, data, and execution layers following in 2028. For comparison, Ethereum's official post-quantum roadmap sets a non-binding target of approximately 2029 for core infrastructure changes, with a full execution-layer migration extending beyond that date.

For now, this strategic change is forward-looking and has no immediate impact on the network. Drake’s announcement does not describe a deployed fork and calls for no changes to existing rollups or zkVMs. Likewise, Ethereum's roadmap advises users to leave their wallets unchanged. The decision is focused on which hash functions will form the foundation of a future, proof-driven iteration of the layer-1 protocol.