Researchers have significantly reduced the estimated computing resources needed for a critical operation in a theoretical quantum attack against the cryptography used by Bitcoin and Ethereum.
The open research project, known as ECDSA.Fail, produced a circuit requiring 1,151 logical qubits and about 1.30 million Toffoli gates at the study’s July 26 data cutoff. Its combined resource score fell 86.1% from the project’s starting point to roughly 1.496 billion. The findings were published as an arXiv preprint on Sept. 9, 2026.
The result does not mean Bitcoin or Ethereum can currently be broken by a quantum computer. Instead, it reduces the estimated resources required for one important calculation that would form part of a much larger attack using Shor’s algorithm.
KEY TAKEAWAYS
- Researchers reduced the ECDSA.Fail resource score by 86.1% in about two months.
- The leading circuit at the study cutoff used 1,151 logical qubits and 1,299,453 average executed Toffoli gates.
- The result was more than 50% below Google’s previously published point-addition benchmark, although the researchers warned that the two measurements use different accounting methods.
- The research does not represent a complete quantum attack and excludes major hardware and error-correction requirements.
- More than 100 contributors, including human researchers and AI agents, participated in the open optimization effort.
OPEN RESEARCH CUTS THE ESTIMATED QUANTUM COST
Bitcoin and Ethereum rely on the secp256k1 elliptic curve for transaction signatures. A sufficiently powerful fault-tolerant quantum computer could theoretically use Shor’s algorithm to derive a private key from an exposed public key, potentially allowing an attacker to authorize transactions.
ECDSA.Fail focused on elliptic-curve point addition, a mathematical operation that would be repeatedly performed during such an attack. Researchers measured each circuit using a score based on peak logical qubit requirements multiplied by the average number of executed Toffoli gates. Logical qubits represent error-corrected quantum information, while Toffoli gates provide a way of estimating computational workload.
The initial circuit required 2,715 logical qubits and an average of 3,960,753 Toffoli gates. By the July cutoff, the leading design had reduced those figures to 1,151 qubits and 1,299,453 gates, producing a score of approximately 1.496 billion. The researchers said the improvement came through more than 400 promoted submissions during the challenge. AI coding agents were used for implementation, testing and incremental optimization, while human researchers generally selected research directions and made larger architectural decisions.
The project was launched by Eigen Labs after Google Quantum AI published its own quantum resource estimates earlier in 2026. Google did not publish its underlying circuit, but it provided verification tools that enabled researchers to test alternative designs. The new result is numerically more than 50% below Google’s reported point-addition benchmark. However, the ECDSA.Fail authors explicitly cautioned against treating this as a direct performance victory because the two approaches use different interfaces, assumptions and accounting conventions.
THE QUANTUM THREAT REMAINS A FUTURE PROBLEM
The findings do not indicate that a Bitcoin or Ethereum attack is imminent. The benchmark covers only one major component of Shor’s algorithm. It does not include the complete algorithm, physical error correction, hardware-specific compilation or other infrastructure that a real fault-tolerant quantum computer would need.
The researchers also produced a version designed to work more closely with the requirements of a windowed Shor implementation. That circuit used 1,162 logical qubits and 1,684,161 average executed Toffoli gates, with an adjusted score of about 1.961 billion. The research continued after the formal cutoff. A later design reduced the combined score to approximately 1.259 billion, while another circuit reduced the logical-qubit requirement to 813 at the cost of substantially more computation.
Theta Labs CTO and lead author Jieyi Long stressed that the issue is about preparation rather than an existing attack:
“None of this is urgent because an attack is imminent. It is urgent because the remedy takes years and cannot be applied retroactively.”
That concern is reflected in Ethereum’s current development plans. The Ethereum Foundation says its protocol team is targeting quantum resistance across Ethereum’s execution, consensus and data layers by December 2029, while acknowledging that the timetable is a planning target and could change.
Ethereum is already working on replacing or adapting cryptographic systems vulnerable to quantum computing, including ECDSA account signatures, BLS validator signatures and KZG commitments. Its roadmap also includes mechanisms intended to give accounts greater flexibility in adopting post-quantum signatures.
CONCLUSION
The ECDSA.Fail research does not bring Bitcoin or Ethereum closer to an immediate quantum breach. What it does show is that algorithmic improvements can substantially reduce the theoretical computing resources required for a future attack.
The 86.1% reduction demonstrates why quantum security estimates cannot be based solely on advances in physical quantum hardware. Better circuit design, software optimization and collaborative research can also narrow the gap.
For blockchain networks, that makes post-quantum migration a long-term engineering issue rather than something that can simply be addressed when a sufficiently powerful quantum computer appears. The research provides a lower benchmark for one part of the problem, while significant technical challenges remain before a complete cryptographic attack becomes practical.
