Author: Bitcoin Security Consortium, a coalition of industry leaders dedicated to the long-term security and stability of the Bitcoin protocol
Compiled by Gu Yu, ChainCatcher
A Bitcoin Security Consortium report, completed in collaboration with Presidio Bitcoin.
Key Points (TL;DR)
Recent advances in quantum computing are real—but only incremental: developments such as Google's Willow and recent optimizations to Shor's algorithm have raised concerns, but a cryptographically relevant quantum computer (CRQC) has not yet been demonstrated.
The risk Bitcoin faces from quantum computing is limited to its signature scheme, manifesting in two ways: static attacks against vulnerable coins and in-transit attacks on mempool transactions.
The solution has been identified and is being actively advanced: (1) moving funds out of reused addresses, (2) developing post-quantum signature schemes, and (3) resolving consensus issues regarding P2PK tokens that have likely been lost.
The Revival of Quantum Computing?
After years of apparent stagnation, an increasing number of academic papers demonstrating progress in quantum computing, along with accompanying headlines, have sparked concerns among investors about the security of bitcoin. These headlines include:
Google's Willow (December 2024). First demonstration that adding physical qubits to protect a single error-corrected "logical" qubit makes that logical qubit more reliable, not less.
Google Quantum AI, Oratomic/Caltech, and ECDSA.fail optimize Shor’s algorithm (2026). Google Quantum AI, in collaboration with the Ethereum Foundation and Stanford, compressed the circuit required to break Bitcoin’s elliptic curve using Shor’s algorithm, reducing it by approximately 20 times compared to prior estimates. Oratomic and Caltech assert that their optimized Shor’s algorithm can crack a single Bitcoin private key within days. Finally, the ECDSA.fail project—an Eigen Labs leaderboard—brought together researchers and AI agents who further shrunk Google’s circuit by nearly 70% over the following months.
IonQ's end-to-end Shor algorithm optimization (September 2026). A complete compilation blueprint for running the Shor algorithm, co-optimized for software and hardware—theoretically solving a single private key in approximately 26 days per attempt, with a 63.3% success rate.
Google's magic state synthesis (October 2026). Quantum computing requires "magic states"—resource states that are consumed with each use and are expensive to produce. Google demonstrated on Willow that its filtering protocol reduces errors in these states by approximately 40 times, achieving 99.99% logical fidelity.
Quantinuum’s “five nines” logical fidelity (October 2026). Quantinuum’s Helios computer has five errors per million operations (99.9995% logical fidelity), approximately 100 times lower than Google’s demonstration. Errors decrease by a factor of 12 with each scaling of the code (λ = 12).
Although no cryptographic-relevant quantum computers (CRQCs) currently exist—and the technology still faces numerous engineering challenges, such as physically scaling error-corrected qubits—why are these headlines capturing the attention of the Bitcoin community? What is the relationship between Bitcoin and quantum computers? Let’s explore.
Brief Overview of Quantum Computing
The core of quantum computing is the quantum bit, or qubit. Unlike classical bits that store values of 0 or 1, qubits exhibit two properties unique to the quantum mechanical world:
Superposition: A qubit can simultaneously hold a mixture of 0 and 1 values, and measurement causes these values to collapse into one of them with weighted probability.
Entanglement: Two or more qubits can share a joint state, such that measuring one determines the value of the other, regardless of the distance between them.
These features enable quantum computers to simultaneously explore many candidate answers while canceling out incorrect ones. However, physical qubits are highly fragile and prone to decoherence—the loss of quantum information—which is why many companies’ current roadmaps aim for the minimum number of logical qubits: error-corrected qubits built from multiple redundant physical qubits, enabling errors to be detected and corrected, ensuring reliable computation and information retention.
To find the lowest physical qubit to logical qubit ratio, quantum computing laboratories research the most efficient hardware architectures, including but not limited to:
Superconducting circuits (Google, IBM): Small loops of wire cooled to near absolute zero, where current behaves like a qubit. This technology enables faster computation times, but its development and scaling are slow.
Neutral atoms (Oratomic): Uncharged atoms trapped by a laser array called "optical tweezers," with their electrons serving as qubits. This technology is scaling rapidly, but theoretically has slower computation times.
Trapped ions (Quantinuum, IonQ): Charged atoms confined in electromagnetic fields and manipulated with lasers—in this case, the ion’s ground or excited energy states serve as qubits. Computation is relatively slow; scalability is moderate.
Photonics (PsiQuantum, Xanadu): Photons—particles of light—travel through chips and optical devices; each photon’s properties serve as a qubit. Enables faster computation times, but development timelines appear slower.
The quantum attack surface of Bitcoin
Theoretically, a cryptographically relevant quantum computer running Shor's algorithm would be able to break Bitcoin’s Elliptic Curve Digital Signature Algorithm (ECDSA)—the signature scheme that maintains ownership and property rights on the network by proving possession of a private key without revealing it. Importantly, most of Bitcoin’s core game-theoretic and economic principles fall outside this attack surface, such as its transaction consensus, issuance schedule, and supply cap.
Attacking a Bitcoin private key presents two attack surfaces to the network:
Long-term exposure or static attacks. Tokens in reused addresses or older address types with lower security (pay-to-public-key, P2PK) have already exposed their public keys, which quantum computers could exploit to derive the corresponding private keys. Today, most active addresses using modern output types—where public keys remain hidden until spent—are not vulnerable to such attacks. However, Project Eleven estimates that approximately 7.1 million BTC, about 35% of the total supply, would be vulnerable to theft if a capable quantum computer existed today. About three-quarters of these are held in reused addresses—mostly at large exchanges—and the risk could be reduced if exchanges moved funds to one-time addresses and adopted better custody practices and standards. However, another 1.8 million BTC are held in P2PK addresses and are largely assumed to be lost, making migration to secure addresses potentially infeasible. Quantum computers, regardless of speed, could exploit these exposed public keys one at a time to obtain the private keys controlling investors’ funds.
Short-term exposure or in-transit attack. Tokens on the Bitcoin network reveal their public keys only when spent, creating a vulnerable window of approximately 10 minutes (or more precisely, the time between transaction submission and inclusion in a block). A sufficiently fast quantum computer could exploit this window to steal tokens during transit. Google estimates that a pre-warmed, fast-clock quantum device could complete the computation to obtain the private key in as little as 9 minutes, putting even tokens originally held in addresses not inherently vulnerable to attack at risk of theft during typical transactions.
Misunderstandings that may exaggerate threats
Quantum hardware has been scaled up and is ready. The system with the best current demonstration results—Quantinuum’s—has produced 48 error-corrected logical qubits. At least about 1,200 are needed to break Bitcoin’s elliptic curve.
Quantum demonstrations have not proven cryptography-relevant quantum computing. Willow demonstrated error-corrected memory, not large-scale error-corrected computation. The 2026 papers from Google, Oratomic, and IonQ are blueprints, not operational machines. Low-error operations between many logical qubits remain a major bottleneck to be overcome.
All bitcoin would be immediately at risk of theft. Most bitcoin are not vulnerable while in a static state, because public keys remain hidden behind hashes until spent. The largest single exposure—address reuse—can now be mitigated by rotating to new unused addresses. Even if an attack were to occur today, it would target individual private keys one at a time, rather than compromising the entire vulnerable supply at once.
Misconception of underestimating the threat
“The optimization of Shor’s algorithm is irrelevant.” In March, Google reduced the theoretical requirement for logical qubits needed to break Bitcoin’s ECDSA by approximately 20 times, noting that elliptic curves have received far less optimization attention than the RSA (Rivest-Shamir-Adleman) cryptosystem, implying there is still significant room for further optimization. For example, the ECDSA.fail project has further optimized Google’s circuit by nearly 70%.
“That’s something that will happen decades from now—why worry?” Recent expert assessments suggest there is approximately a 50% chance that a capable machine will emerge between 2030 and 2035. Governments aim to complete post-quantum migration by the early to mid-2030s, while Google and Cloudflare have set internal deadlines for 2029. It’s important to note that these deadlines are more urgent for governments and corporations defending against “harvest now, decrypt later” attacks—primarily involving private information that may remain relevant in the future. Since Bitcoin can defend itself by having participants migrate to addresses with hidden public keys, these deadlines can be viewed as best practices for the network rather than absolute deadlines.
We will see it coming in advance. Breakthroughs may remain private or be considered classified by nation-states, and early thefts may appear sporadic and difficult to distinguish from routine custodial turnover or payment transactions.
What is being done now
Today’s Bitcoin holders—including exchanges, institutions, and investors—are already actively discussing solutions to mitigate quantum threats, with priorities likely as follows:
Move vulnerable supplies from reused addresses to unused addresses. This is the easiest step to implement, requiring no network consensus or soft fork—it relies entirely on large exchanges and custodians, as well as self-custodial investors, adopting better security practices such as change output rotation and absolutely avoiding address reuse.
Design a post-quantum signature scheme. To date, the leading proposal for achieving quantum-resistant spending on Bitcoin is Blockstream’s SHRINCS (“Shrunken SPHINCS”). Built on SHA-256, it can process up to 3 transactions per second, compared to just 0.36 for NIST’s standard SPHINCS+. Blockstream tested the scheme on the Liquid Network in March 2026 and, together with other contributors, published a Bitcoin Improvement Proposal (BIP) in August. Formal security proofs are still pending, and full network deployment will require a soft fork.
Consensus has been reached on how to handle tokens in Bitcoin’s vulnerable, and likely lost, P2PK addresses. As Bitcoin faces quantum threats in the coming years, this may become the most contentious issue. Approximately 1.7 million BTC are held in around 20,000 P2PK addresses, including an estimated 1.1 million BTC attributed to Satoshi Nakamoto. According to Presidio Bitcoin, the most prominent options under discussion are: maintaining legacy spending validity (allowing quantum attackers to “steal”); slowing legacy spending (restricting the movement of vulnerable tokens, or “speed-limiting”); and freezing tokens with a recovery pathway (disabling all legacy Bitcoin spending but allowing owners to reclaim their tokens via a zero-knowledge proof of their seed phrase, known as “soft freezing”). For broader reading on this critical topic, Presidio Bitcoin elaborates and updates the fundamentals here.
Featured image: Google Sycamore quantum processor (Google Quantum AI, CC BY 3.0, via Wikimedia Commons)



