The State of Post-Quantum Cryptography
If you've been following recent discussions in bitcoin, you've probably noticed how often post-quantum cryptography comes up.
A cryptographically relevant quantum computer (CRQC) running Shor's algorithm can solve the elliptic curve discrete logarithm problem (ECDLP), which both ECDSA and Schnorr rely on today. That's why preparing for a migration to post-quantum cryptography is becoming increasingly relevant.
We don't know when Q-day will happen, but advancements in quantum computing are pushing us to prepare for the transition [3]. Migration takes time. A recent report from Chaincode Labs estimated around two years for short-term contingency measures and seven years for a comprehensive bitcoin migration [1]. The US government has set 2035 as a broad goal for mitigating quantum risk [4], while Google recently announced its own PQC migration target of 2029 [3].
We should also make a distinction between two types of threats:
Data confidentiality: This is concerned with an attack called "harvest now, decrypt later". Imagine Alice and Bob sending encrypted sensitive text messages to each other. Eve, an attacker, can see the encrypted messages but doesn't have the key to decrypt them. However, Eve can store these messages until a sufficiently powerful quantum computer becomes available. If the messages used quantum-vulnerable key exchange, Eve could then decrypt and read them. That's what we mean by "harvest now, decrypt later", and that's why transitioning to PQC is already urgent for this case [2].
Data authentication: This is where bitcoin comes in. An attacker with a powerful quantum computer running Shor's algorithm could derive a private key from an exposed public key and produce a valid signature to spend the coins [1].
So even though we don't know exactly when Q-day will arrive, preparing for the transition is important, especially for systems like bitcoin, where introducing new cryptography and migrating users could take years.
What Can Quantum Computers Actually Break?
When talking about quantum threats, we need to distinguish between two main algorithms: Shor and Grover.
Shor's Algorithm
Shor's algorithm is able to solve the discrete logarithm problem that ECDSA and Schnorr rely on, and the integer factorization problem that RSA relies on today [2].
Grover's Algorithm
Grover's algorithm, on the other hand, provides a quadratic speedup for unstructured search [2].
Let's use SHA-256 as an example to explain this.
We have to distinguish between two types of attacks on SHA-256:
- Preimage search: Given H(x) = y, finding an input x for a known hash value y is computationally infeasible. For SHA-256, this requires approximately 2^256 classical operations. However, a quantum computer running Grover's algorithm can reduce this to approximately 2^(256/2) = 2^128 quantum queries [2].
- Hash collision: Finding two different inputs that hash to the same value, such that H(x1) = H(x2), is also computationally infeasible. On a classical computer, this requires approximately 2^128 operations (see @bitcoin_devs post about the birthday paradox).
2^128 is still enormous. And also keep in mind that Grover's algorithm cannot be parallelized as efficiently as classical brute-force search. Someone could allocate more quantum computers to parallelize the work, but the speedup would not be linear.
You might ask: What if future research develops a new algorithm that could outperform Grover?
Christof Zalka showed that Grover's algorithm achieves the best possible query complexity for generic unstructured search [6]. So a new algorithm cannot outperform Grover in this model.
The TL;DR is that SHA-256 is still considered secure against known quantum attacks and can still be used in a post-quantum world. And this is what sets the foundation for post-quantum hash-based signature schemes.
We won't be able to rely on ECDSA and Schnorr for digital signatures in bitcoin once powerful quantum computers become available. So we need a new kind of digital signature that is resistant to quantum attacks.
Hash-based signatures are one family of these schemes. What's interesting about them is that they rely on hash functions rather than assumptions about discrete logarithms or lattice problems, and we've been using and studying hash functions for decades.
So What Are the Alternatives?
In 2024, NIST published its first three post-quantum cryptography standards [7]:
- ML-KEM: Based on lattice cryptography and used for establishing shared secret keys.
- ML-DSA: A lattice-based digital signature algorithm, also relevant to bitcoin since it concerns digital signatures.
- SLH-DSA: A hash-based digital signature algorithm.
NIST selected ML-DSA as its primary post-quantum digital signature standard, and SLH-DSA as a backup [7].
You might ask, why does NIST even need to standardize multiple signature algorithms? Shouldn't we just rely on lattice-based signatures since they're generally faster and have smaller signature sizes compared to hash-based signatures [2]?
That's a valid question, but the main reason is cryptographic diversity.
Rather than relying on lattice-based signatures alone and discovering a few years later that researchers have found a practical attack against them, we have another family of signatures that relies on different cryptographic assumptions.
Hash-based signatures are interesting because their security relies on well-studied hash functions, rather than the mathematical hardness assumptions used by lattice-based cryptography. However, this comes with trade-offs, particularly larger signature sizes [2].
The transition to post-quantum cryptography has already started. Companies like Cloudflare and Google have been working on deploying post-quantum cryptography on the web.
According to Cloudflare, around 70% of browser-generated traffic reaching its network was already protected using hybrid post-quantum key agreement with ML-KEM as of September 2026 [8].
However, this mainly concerns key agreement, not digital signatures. The transition to post-quantum authentication, which involves certificates and digital signatures, is still ongoing [8].
But Can We Really Trust Them?
The fact that NIST standardized the two digital signature schemes we talked about earlier, SLH-DSA and ML-DSA, doesn't mean they can never be broken.
To allow for more cryptographic diversity, NIST started another standardization process for additional digital signature schemes, so we're not relying on just two schemes [10].
One of the candidates was HAWK, which uses lattice-based cryptography. HAWK had already passed two rounds of expert review and reached the third round of NIST's evaluation process.
However, in July 2026, Anthropic's Claude Mythos helped discover an attack in just 60 hours of work that substantially reduced HAWK's estimated security strength [9]. Following this discovery, the HAWK team withdrew the scheme from NIST's standardization process [10].
This shows another interesting angle: cryptographic schemes are not only facing potential future quantum attacks. Advances in AI are also helping researchers discover weaknesses in cryptographic constructions. The HAWK attack did not break ML-DSA or lattice-based cryptography in general [9].
Why Is bitcoin Different?
You might be wondering, if NIST has already standardized two digital signature schemes, SLH-DSA and ML-DSA, why can't we just choose one of them for bitcoin? The web is already adopting NIST's post-quantum key agreement standard, so why can't bitcoin do the same?
Bitcoin is different because we have to take into account signature size, verification time, block space, and wallet compatibility.
According to Blockstream's analysis, bitcoin could process around 6.5 transactions per second if all transactions used Schnorr signatures, assuming an average transaction with 2.27 inputs and 2.64 outputs (the 90-day average as of March 30, 2026, per https://mainnet.observer) [11].
Let's analyze the trade-offs of each standardized NIST signature scheme under those assumptions:
- Adopting ML-DSA: ML-DSA is lattice-based, as we discussed before, and has good characteristics like small signatures and fast verification compared to hash-based signatures. However, throughput would drop to about 0.5 transactions per second (at NIST security level 3) [11]. ML-DSA also introduces additional security assumptions based on lattice hardness.
- Adopting SLH-DSA: SLH-DSA is based on hash functions, which we already trust and use everywhere in bitcoin. It doesn't introduce new lattice-hardness assumptions. The problem is its large signature size, which would reduce modeled throughput further to about 0.36 transactions per second (at NIST security level 1) [11].
So what's the plan?
SHRINCS is one of the experimental proposals being developed for bitcoin. It combines two signing paths: stateful and stateless.
The stateless path uses SLH-DSA algorithms with different parameters, reducing the signature budget from 2^64 to 2^40 signatures per key. This brings the signature size down to 5,777 bytes [12].
The stateful path is the main innovation that SHRINCS brings. In the current draft, its signatures range from 548 to 4,619 bytes [12], but this comes with the trade-off of managing state, which can be challenging for bitcoin wallets, especially when restoring backups.
We'll dive into how SHRINCS works and these trade-offs in a follow-up article.
Where Do We Go From Here?
As we've seen, post-quantum signature schemes already exist, but choosing one for bitcoin comes with trade-offs in signature size, security assumptions, and performance.
In the follow-up articles, we'll explore the SHRINCS proposal in detail by going through its building blocks one by one, starting with WOTS, XMSS, FORS, and hypertrees, before putting everything together to understand how SHRINCS works.
References
[1] Milton, A., & Shikhelman, C. (2025). bitcoin and Quantum Computing: Current Status and Future Directions. Chaincode Labs.
[2] Banerjee, A., et al. (2026). RFC 9958: Post-Quantum Cryptography for Engineers. IETF.
[3] Adkins, H., & Schmieg, S. (2026). Quantum Frontiers May Be Closer Than They Appear. Google, March 25, 2026.
[4] NIST (2022). NIST Role and Activities Relative to the Post Quantum Cryptography White House Memo.
[5] Brassard, Høyer & Tapp. Quantum Algorithm for the Collision Problem.
[6] Christof Zalka. Grover's Quantum Searching Algorithm Is Optimal.
[7] NIST (2024). First Three Finalized Post-Quantum Standards.
[8] Cloudflare (2026). Is Your Domain Using Post-Quantum Encryption?.
[9] Anthropic (2026). Discovering Cryptographic Weaknesses with Claude.
[10] NIST (2026). Nine Candidates Advance to the Third Round.
[11] Blockstream. OP_CHECKSHRINCS: A Hash-Based Signature Opcode for Post-Quantum Bitcoin.