The predicted scale of a computer capable of compromising Bitcoin’s cryptography has significantly decreased.
Scientists from the quantum computing company Alice & Bob, based in Paris, have discovered that a machine with approximately 30,000 physical qubits could potentially uncover a Bitcoin private key from its public key in just over one hour. They urge Bitcoin developers to prioritize a transition to quantum-resistant security solutions now, as implementing these upgrades could take years.
To provide some background on the rapid decline of this number, earlier estimates suggested that millions of physical qubits would be necessary to breach Bitcoin’s security.
In March 2026, Google Quantum AI reduced this estimate to below 500,000 physical qubits. Now, Alice & Bob’s models suggest around 30,000 qubits, making their requirement roughly 17 times lower than Google’s upper limit.
The focus is on secp256k1, the elliptic-curve cryptography utilized to secure Bitcoin wallets. The attacking method employed is Shor’s algorithm, which allows for deriving the private key from its public counterpart.
For clarity on qubits: physical qubits are the real hardware components and come with a high degree of error. In contrast, logical qubits are made up of multiple physical qubits that correct errors. Reducing the need for physical qubits thus simplifies the most complex and costly part of the machine.
Alice & Bob’s technique uses superconducting “cat qubits,” featuring their unique hardware design. The efficiency of their model arises from integrating this design with a more streamlined method for conducting the attack.
No machine of this kind has been built yet. Fully operational fault-tolerant quantum computers at this scale are still projected to be several years away; the researchers anticipate such potentially disruptive technology may emerge around the early 2030s.
Alice & Bob is actively working on this challenge. Founded in 2020, the company has raised around €130 million, with a goal of developing a fault-tolerant system featuring 100 logical qubits by 2030.
Google’s whitepaper from 2026 presented another key insight, indicating that an “on-spend” attack could be executed in as little as 9 to 23 minutes using only 1,200 to 1,450 logical qubits.
An on-spend attack focuses on transactions in process. Once a Bitcoin transaction is broadcast, the corresponding public key becomes visible before the transaction is finalized. A sufficiently rapid quantum computer could theoretically determine the private key within this timeframe and reroute the funds.
The other vulnerability involves dormant coins. Approximately 6 to 7.1 million BTC, representing about 30% of the total supply, are held in addresses where public keys are already accessible.
The crux of the researchers’ argument pertains less to technical intricacies and more to governance. Bitcoin is not engineered for swift upgrades.
To make any protocol changes, including a shift to quantum-resistant signatures, extensive testing is necessary before implementation. Alice & Bob warns that this process could span several years. If the required hardware becomes available in the early 2030s and the upgrades take significant time to complete, the timeline for action becomes clear.
Researchers also noted that advancements in hardware efficiency could unexpectedly speed up the predicted timeline. Their study serves as a case in point: the needed qubit count has significantly dropped within a year, largely due to improved algorithms and designs rather than a groundbreaking machine.
Individuals most at risk are those holding coins in addresses with disclosed public keys. This approximately 30% portion of the total supply would be the primary target for a capable quantum adversary, and protecting it through a protocol upgrade would be challenging without users actively relocating their assets.
For developers, the major hurdle lies in coordination. Selecting a quantum-resistant signature method, validating it, and persuading a decentralized network to adopt it is an extensive project that may take years, even if there is unanimous agreement.