How estimates have fallen
Resources for the attack's core step
Score = logical qubits × Toffoli gates, in billions.
Dashed line: Google published upper thresholds · about 2.99B
- 10.75B
- 2.99B
- 3.01B
- 1.496B
- 1.259B
- 0.915B
View resource estimates and accounting
| Date | Circuit | Logical qubits | Toffoli gates | Q × T score |
|---|---|---|---|---|
| 30 May 2026 | ECDSA.fail starting circuit | 2,715 | 3,960,753 | 10.75B |
| Mar 2026 | Google Quantum AI | up to 1,425 | up to 2.1M | 2.99B |
| Jun 2026 | Schrottenloher · open design | 1,462 | 2.06M | 3.01B |
| 26 Jul 2026 | ECDSA.fail best at cutoff | 1,151 | 1,299,453 | 1.496B |
| Reported Sep 2026 | ECDSA.fail after cutoff | 1,321 | about 953,000 | 1.259B |
| 27 Oct 2026 | Current ECDSA.fail record | 1,234 | 741,629 | 0.915B |
What the estimate covers
Point addition inside Shor's algorithm
Public keyvisible on-chain
Shor's algorithm on secp256k1
- Repeated lookup, point addition and unlookupECDSA.fail benchmarks the point-addition part.
- Quantum Fourier transform
- Measurement and classical post-processing
Private keyrecovered
Paper and contributors

ECDSA.Fail: Open Autoresearch for Optimizing Elliptic-Curve Point Addition in Shor's Algorithm
arXiv 2609.09582 · 9 Sep 2026
Eigen Labs launched ECDSA.fail as an open competition, and more than 100 people took part, alongside AI agents. Four StarkWare researchers and three Starknet Foundation researchers co-authored the resulting paper.
Read the paper on arXiv