01
STARK proving
Built on hashes
STARK proving relies on hash functions rather than the elliptic-curve assumptions threatened by Shor’s algorithm. Starknet does not need to replace its proving system as part of its post-quantum migration.

Architecture
StarkWare created STARKs and develops S-two, the production prover securing Starknet. Starknet's native account abstraction lets accounts adopt new signature schemes.
The throughline
Cryptographic agility lets a system change cryptography without redesigning the whole stack. Starknet’s two architectural advantages for post-quantum migration are STARK proving and native account abstraction.
01
Built on hashes
STARK proving relies on hash functions rather than the elliptic-curve assumptions threatened by Shor’s algorithm. Starknet does not need to replace its proving system as part of its post-quantum migration.
02
Signatures can change
Programmable Starknet accounts can use post-quantum signature verifiers today without a protocol-wide hard fork.
STARK proving
At the proving layer, yes. STARKs use hash-based commitments and do not rely on elliptic-curve hardness. Shor's algorithm targets the elliptic-curve assumptions used by common signatures. STARK proving uses a different construction.
StarkWare created STARKs, and co-founder Eli Ben-Sasson co-authored the 2018 STARK paper. S-two now powers Starknet's production proving stack.
In November 2026, Anthropic published FlashProve, a research prototype that uses a hash-based, post-quantum proof system to prove an AI model’s answers, and named Starknet as an example of where this kind of proof is deployed. Read Anthropic’s post
Built on hashes
STARK proving relies on collision-resistant hash functions rather than elliptic-curve hardness. This keeps the proving layer outside the assumptions threatened by Shor's algorithm.
Proves heavy computation
Post-quantum signature checks cost more than common schemes. STARK proving gives Starknet a way to absorb that extra computation.
No trusted setup
STARKs do not require secret setup material. No party needs to create or safeguard hidden parameters.
Account validation
Starknet launched with native account abstraction. Each account is a programmable contract that can define its own signature checks. A new account implementation can adopt a post-quantum signature verifier without a protocol-wide change.


Crypto agility
NIST published FIPS 203, FIPS 204 and FIPS 205 in August 2024, standardizing ML-KEM, ML-DSA and SLH-DSA. NIST IR 8547 is separate draft transition guidance, useful for planning but not a deadline set by those publications and not a determinant of Starknet's schedule.
No date for a cryptographically relevant quantum computer is known. Migration work is therefore sequenced as engineering rather than timed against a forecast, and its status is tracked publicly per component rather than asserted as a whole.
Early migration is an engineering project you schedule; late migration is a coordination problem you inherit.Read the threat and standards context
The record
Dated public events, ordered by date rather than by significance. Component-level status lives in the migration register on the Roadmap page.
2018
StarkWare's founders publish the STARK proof system: no trusted setup, no elliptic-curve hardness assumption in the proving layer.
June 2020
13 Aug 2024
NIST publishes FIPS 203, 204 and 205, standardizing ML-KEM, ML-DSA and SLH-DSA.
12 Nov 2024
NIST IR 8547 published as Initial Public Draft transition guidance.
November 2025
S-two, StarkWare's hash-based next-generation STARK prover, goes live on Starknet Mainnet.
February 2026
BIP-360 (Pay-to-Merkle-Root) merged into the Bitcoin Improvement Proposal repository.
July 2026
Starknet v0.14.3 replaces Pedersen with a BLAKE2s-256-based hash construction for the live OS program and configuration hashes.
October 2026
93.52% of voting power backed research into whether Starknet should move to its own post-quantum L1.
The 2018 paper
Scalable, transparent, and post-quantum secure computational integrity. Eli Ben-Sasson, Iddo Bentov, Yinon Horesh, Michael Riabzev · Cryptology ePrint Archive, Paper 2018/046 · 2018

The paper introduced STARKs, a proof system that assumes only that cryptographic hash functions are collision-resistant. Most zero-knowledge systems in production, including Groth16 and PLONK with KZG commitments, rely instead on elliptic curves, which Shor's algorithm breaks on a large enough quantum computer. Grover's algorithm, the strongest generic quantum attack on hashes, is handled by sizing security parameters.
The paper built this choice into the foundation of the proof system; it is not a later patch or a parameter choice. It is a property of the proving layer rather than a claim about every cryptographic component in Starknet.
Read the threat and standards contextThe construction requires no trusted setup. There are no secret initialization parameters whose compromise would let an adversary forge proofs; verifier inputs are public. This is a property of the proof construction.
The construction was designed for succinct verification of large computations. Its practical production record is separate evidence from the cryptographic assumptions that support the proof system.
Merkle commitments and Fiat-Shamir place cryptographic assumptions in hash functions rather than elliptic-curve hardness in the proving layer. The resulting security boundary still depends on selected hashes, parameters, protocol details, and implementation.
The paper describes a proof system. Starknet’s native account abstraction gives a new account implementation programmable validation, but existing accounts, wallets, custody, keys, and adoption require separate migration work. Migration tracks the other surfaces.
Last updated: 3 November 2026.