AI Agents Outpace Google’s Quantum Benchmark
A new wave of research has sharply reduced the estimated resources required to mount a quantum attack on Bitcoin and Ethereum, with more than 100 researchers leveraging artificial intelligence coding
The ECDSA.Fail contest, organized by Eigen Labs, became the battleground for these advances. Here, circuit designs for attacking secp256k1—the elliptic curve securing both Bitcoin and Ethereum—were scored by multiplying the number of logical qubits by the count of Toffoli gates (a type of quantum logic gate crucial for complex computations). The leading design used just 1,151 logical qubits and initially about 1.3 million Toffoli gates, with later refinements pushing that number below one million. This result cuts Google Quantum AI’s March benchmark roughly in half, though the two efforts used different scoring methodologies.
The resource gap between human-AI teams and previous industry leaders is closing fast.
Between late May and July 26, the ECDSA.Fail competition saw its resource benchmark for a quantum attack plummet from 10.75 billion to 1.496 billion.
ECDSA.Fail Contest Shakes Up Crypto Timelines
The contest’s rapid progress has forced both industry and academia to reconsider their timelines for quantum preparedness. For context, the tests did not actually crack any real Bitcoin private keys but did prove that the theoretical cost of such an attack could be much lower than previously thought. The leading circuit design was authored by a coalition spanning Eigen Labs, Trail of Bits, StarkWare, Theta Labs, MultiVM Labs, and the Ethereum Foundation—a rare cross-section of crypto security expertise.
This acceleration adds urgency to ongoing efforts: the Ethereum Foundation has set December 2029 as a hard deadline to make all transactions, validators, and storage quantum-resistant. Meanwhile, StarkWare has already pushed what it claims is the first quantum-safe Bitcoin transaction live on mainnet—a symbolic milestone even as practical attacks remain out of reach for now.
Competing Methods Cloud “Best” Quantum Estimate
Despite headline-grabbing reductions in resource estimates, there remains significant ambiguity about which measurement best represents real-world risk. The recent paper shared with CoinDesk highlights that humans and AI agents have both outperformed Google’s March results—but with different counting methods muddying direct comparisons. In practice, some designs optimize for fewer logical qubits while others minimize Toffoli gates; both are critical resources in building scalable quantum computers but may not scale equally in hardware implementations.
This nuance matters because NIST’s draft guidance proposes deprecating classical public-key algorithms at the 112-bit security level after 2030 and outright disallowing them after 2035. As post-quantum cryptography standards are finalized and adopted, choosing which benchmarks to prioritize will shape how urgently protocols must migrate away from current systems like ECDSA.
See Also
Ethereum Sets 2029 Quantum-Resistant Deadline
Ethereum’s leadership is already responding to these shifting benchmarks with concrete proposals. Vitalik Buterin recently co-authored EIP-8288 in June—a plan designed to slash the cost of verifying quantum-safe private transactions by over 99%. Today’s post-quantum signatures can be two or three kilobytes in size and require up to 200,000 gas for verification; STARK proofs can balloon past 128 kilobytes and cost millions of gas units to check on-chain. Under EIP-8288, both private and quantum-safe private transactions could drop into the low tens of thousands of gas by moving heavy cryptographic operations out of Ethereum’s execution path.
Buterin’s proposal depends on another upgrade called Frames—an overhaul that remains unscheduled—underscoring how technical dependencies may delay widespread adoption even amid growing urgency. Nonetheless, Ethereum’s public commitment to a December 2029 deadline signals a willingness to move faster than regulatory timelines from bodies like NIST.
Market Stakes: Price Stability vs Security Race
While these technical contests play out behind the scenes, Bitcoin continues to hold its position as the world’s largest cryptocurrency by market cap. As of September 11, BTC trades at $76,978—down just -0.01% over the last day and -0.05% over seven days, yet still up +0.20% over thirty days. This price stability contrasts sharply with the underlying uncertainty around future-proofing its security model against quantum threats.
Investors might take comfort in BTC’s resilience near record highs (all-time high: $126,080 hit in October 2025), but this calm belies an intensifying race among researchers—and even AI agents—to make or break crypto’s long-term defenses.
There is also a tension between proactive defense funding and practical implementation: In July alone, Galaxy Digital pledged up to $5 million toward quantum defense research; nine other firms including BlackRock and Coinbase committed $15 million across three years for broader Bitcoin security initiatives encompassing quantum risk mitigation. These sums are substantial but must translate into real-world upgrades before looming deadlines like Ethereum’s December 2029 commitment or NIST's post-2030 deprecation guidance come into force.
Signals yet to emerge
If the Ethereum Foundation’s hard December 2029 deadline for quantum resistance is not met, it would immediately leave Ethereum transactions, validators, and storage exposed to quantum attacks; whether the necessary upgrades will be completed by that date remains unclear.