Key Highlights
- Ethereum researcher Justin Drake says the Ethereum Foundation is abandoning Poseidon for Ethereum L1 and pivoting toward established hash functions such as SHA and BLAKE.
- The shift follows advances in binary-field SNARKs, including Binius and the newer Flock system, that make proving conventional hash computations substantially more efficient.
- Drake said the Ethereum Foundation’s roadmap now targets a production-grade leanVM in 2027, followed by consensus, data and execution-layer deployments in 2028.
The Ethereum Foundation is moving away from Poseidon, a specialized hash function that has played a major role in zero-knowledge proof systems, as advances in SNARK technology make established hashes such as SHA and BLAKE increasingly practical for Ethereum.
Ethereum Foundation researcher Justin Drake announced the change on Thursday, calling it the conclusion of an eight-year research effort into “SNARK-friendly” hash functions.
“Goodbye, Poseidon!” Drake wrote, saying the Foundation is “abandoning Poseidon for L1, pivoting to SHA or BLAKE.”
The change does not mean Poseidon is being removed from existing zk-rollups or applications that use it. Drake’s announcement specifically concerns the cryptographic direction being pursued for Ethereum L1 and its longer-term lean Ethereum roadmap.
Why is Ethereum moving away from Poseidon?
Poseidon was designed to solve a particular problem with zero-knowledge proofs: conventional cryptographic hashes such as SHA-256 are built around bitwise operations that historically have been expensive to represent inside many SNARK circuits.
That made specially designed “SNARK-friendly” hashes attractive because they could be evaluated much more efficiently inside proof systems.
According to Drake, the Ethereum Foundation began investing in this area in 2018, with Poseidon emerging in 2019 and subsequently becoming one of the dominant specialized hashes used across zero-knowledge systems.
But the underlying trade-off is changing. Instead of designing hashes around the arithmetic preferred by SNARKs, researchers have developed proof systems that are themselves better suited to the binary operations used by conventional cryptography.
A central development has been the use of binary fields, mathematical fields built around characteristic two. They map much more naturally to bits and operations such as XOR than the large prime fields commonly used in earlier SNARK systems.
Binius is one prominent example. Its documentation explains that binary fields, combined with protocols such as sumcheck, allow bitwise operations to be proven considerably more efficiently. Binius64 is specifically optimized for conventional hashes including SHA2 and SHA3.
Ethereum co-founder Vitalik Buterin had already highlighted the potential of Binius in 2024, describing binary-field proof systems as a path toward more efficient proofs while retaining the quantum-resistance benefits associated with hash-based constructions.
Flock makes SHA and BLAKE much cheaper to prove
The case for conventional hashes became stronger with the introduction of Flock, a new hash-based SNARK developed by Benedikt Bünz, Ron D. Rothblum and William Wang.
The researchers introduced Flock in a July 2026 research paper, describing it as a system designed to efficiently prove large batches of Boolean computations such as SHA-256, Keccak and BLAKE3.
The performance figures help explain why Ethereum researchers are reconsidering specialized hashes.
On a single core of an Apple M4 Max, Flock demonstrated approximately:
- 82,000 BLAKE3 compressions per second
- 42,000 SHA-256 compressions per second
- 30,000 Keccak permutations per second
The researchers calculated that this represented less than a 250-times overhead compared with native execution.
With ten CPU cores, Flock exceeded 660,000 BLAKE3 compressions per second.
For SHA-256 specifically, the researchers reported that Flock was more than 9 times faster than Binius64, which they identified as the previous state of the art, and more than 500 times faster than the fastest elliptic-curve-based SNARK included in their comparison.
The Flock paper argues that standard hashes are attractive because they are already widely deployed and heavily scrutinized, while specialized SNARK-friendly constructions introduce additional cryptographic design choices.
What happens to Poseidon?
Poseidon is not suddenly obsolete across the crypto industry.
It remains widely used in existing zero-knowledge infrastructure, and research has demonstrated strong performance for Poseidon and Poseidon2 under several existing proving architectures.
The Ethereum Foundation’s change is therefore better understood as a roadmap decision for Ethereum L1, rather than a declaration that projects using Poseidon must replace it.
The attraction of SHA and BLAKE is different. They are conventional, extensively studied cryptographic primitives that do not have to be designed specifically around one proof system.
That distinction becomes particularly important for Ethereum’s long-term security plans.
Drake said relying on conventional hash-based cryptography could reduce the need to wait years for specialized primitives such as Poseidon to undergo sufficient cryptanalysis before being deployed in critical parts of Ethereum.
Ethereum targets Post-Quantum L1 architecture
The broader objective is Ethereum’s transition toward a simpler, more post-quantum-resistant cryptographic architecture.
Drake argued that sufficiently efficient SNARKs could allow hash-based cryptography to support blockchain post-quantum signatures while preserving compact verification.
Instead of placing every individual signature directly into a block, a SNARK can prove that a large collection of signatures is valid and compress that computation into a much smaller proof.
This could also support more complicated authorization structures, including threshold signatures and multisignature schemes, without requiring every underlying cryptographic operation to be directly verified by Ethereum nodes.
Hash-based proof systems such as Binius are particularly relevant here because they can be constructed without trusted setups and around assumptions intended to remain secure against quantum adversaries.
Drake contrasted that direction with more complex post-quantum approaches based on lattices and isogenies, although his claim that hash-based cryptography has effectively “won” for blockchain post-quantum signatures remains his assessment rather than an established industry consensus.
LeanVM roadmap points to 2027 and 2028
The cryptographic shift also feeds directly into the Ethereum Foundation’s leanVM work. According to Drake, the current roadmap points toward a production-grade leanVM in 2027, with deployments across Ethereum’s consensus layer, data layer and execution layer targeted for 2028.
The Foundation’s post-quantum researchers are now working with binary fields as part of that effort. The subject will receive further attention during Ethproofs Call #10 on August 14 at 14:00 UTC, which is dedicated to binary-field proof systems, including Flock, optimization work and leanVM benchmarks.
For Ethereum, the development represents a reversal of the earlier ZK design philosophy. Rather than inventing increasingly specialized cryptographic hashes because SHA and BLAKE were too costly to prove, advances in binary-field SNARKs are making it possible to reconsider the standard hashes themselves.
If those performance improvements hold at production scale, Ethereum could simplify some of its future cryptographic assumptions while simultaneously advancing its longer-term post-quantum roadmap.
Also Read: Ethereum Leads $26.6 Billion RWA Market With 56.6% Share
