Bitcoin’s Quantum Backup Gets 79% Cheaper With AI

A one-week AI-assisted coding challenge cut the modelled GPU cost of preparing StarkWare’s quantum-safe Bitcoin transaction from approximately $320 to $67. Bitcoin’s consensus rules stayed unchanged.
Key Takeaways
- QSB’s compute estimate fell about 79%.
- AI-assisted developers accelerated two computational searches.
- No practical quantum attack threatens Bitcoin today.
- QSB requires eligible outputs and miner coordination.
- The $67 estimate remains unproven on mainnet.
What StarkWare means by a Bitcoin “quantum backup”
“Quantum backup” is shorthand for Quantum-Safe Bitcoin, or QSB, an experimental transaction method developed by StarkWare. It uses hash-based cryptography to move eligible bitcoin under Bitcoin’s existing rules, without waiting for the network to adopt another signature system.
The method addresses a future risk in Bitcoin’s elliptic-curve signatures. A sufficiently powerful fault-tolerant quantum computer could theoretically derive a private key from a visible public key, allowing an attacker to spend the associated funds.
No publicly known quantum computer can perform that attack against Bitcoin today. QSB is being developed as a contingency rather than a response to an active breach.
StarkWare demonstrated the approach on August 26, when the first QSB transaction was included in a Bitcoin block. Preparing it required approximately 3,100 GPU-hours, producing an estimated computing bill of $320.
Mainnet inclusion established that the transaction could be accepted under Bitcoin’s current consensus rules. It did not provide a network-wide security audit or add quantum protection to existing wallets.
Coindoo previously compared the first mined QSB transaction with a separate quantum-resistant signature proposal. The latest work concentrates on reducing the computation behind the QSB method already demonstrated.
AI-assisted developers made two searches faster
StarkWare, Yukon Research and Eigen Labs opened an optimization competition on September 16. Participants used AI coding systems to generate and test potential improvements, while human developers submitted the implementations for benchmarking.
The challenge targeted two computational stages.
Pinning searches for a transaction commitment that satisfies QSB’s required conditions. The published benchmark increased from approximately 146 million verified candidates per second to more than 820 million on an RTX 4090 GPU.
Subset selection searches for the combination of transaction components required by the construction. Its measured rate increased from roughly 62 million to more than 620 million candidate subsets per second.
Those gains lowered the time and GPU capacity needed to prepare a transaction. In its published competition results, StarkWare estimated that the same process would now cost approximately $67 under its hardware assumptions, a reduction of about 79%.
The live Yukon dashboard subsequently displayed an estimate of $66 after additional leaderboard improvements. The result is therefore better understood as a changing benchmark than a fixed retail price.
Using QSB would still require four specialized steps
- Identify a compatible unspent output. QSB is intended for eligible outputs whose public keys remain concealed. Some Bitcoin addresses reveal their public keys when funds are spent, so the method cannot be applied equally to every coin or transaction history.
- Run the off-chain computation. Specialized software and GPUs perform the pinning and subset-selection searches. This is the part of the process that became faster during the competition.
- Construct the QSB transaction. The resulting transaction follows Bitcoin’s consensus rules but falls outside the network’s standard relay policy. Ordinary nodes are therefore not expected to propagate it in the usual way.
- Send it directly to a miner. A participating miner must receive the transaction and choose to include it in a block.
The $67 estimate applies to the modelled GPU workload. It does not represent an all-in user price covering software preparation, engineering support, Bitcoin network fees or an arrangement with a miner.
QSB is also unavailable as a normal wallet feature. A holder cannot currently select a “quantum-safe” option, pay $67 and complete the process through a conventional Bitcoin application.
What the competition has actually demonstrated
A CoinDesk review calculated approximately $83 from the benchmark improvements visible during its examination. StarkWare said subsequent leaderboard records supported the lower figure displayed by Yukon.
The different estimates depend on which benchmark results and hardware assumptions are used. Both indicate a large reduction from the original workload, while neither establishes what a complete commercial service would charge.
StarkWare continues to view a future soft fork as the stronger long-term route. Standardized quantum-resistant spending rules could give wallets, exchanges and custodians a common migration path instead of requiring individually prepared QSB transactions.
The remaining bottleneck is coordination
The competition reduced a measurable technical cost in seven days. Raw computing power is consequently less restrictive than it was when the first QSB transaction reached Bitcoin in August.
Operational access has not advanced at the same speed. A practical service would need to identify compatible outputs, prepare the transaction correctly and maintain a reliable route to miners that accept nonstandard submissions.
That changes where the experiment’s main uncertainty sits. Further software improvements may continue lowering the GPU requirement, but widespread protection depends on wallet integration, miner participation and agreement over Bitcoin’s eventual quantum-resistant standard.
This article is provided for informational purposes only. Quantum-computing capabilities, cryptographic research and benchmark estimates can change as new evidence becomes available.








