Organisations preparing for quantum computers are being offered two very different answers, quantum key distribution (QKD) and post-quantum cryptography (PQC), and choosing the wrong one wastes budget while leaving the real exposure untouched. QKD uses the physics of single photons over dedicated links to agree keys; PQC replaces vulnerable public key algorithms with new mathematical ones that run in ordinary software. This guide compares the two on what they protect, what they cost and how mature they are, and sets out what the UK NCSC, the US NSA and the joint French, German, Dutch and Swedish agencies have said about QKD. It covers the NIST standards FIPS 203, 204 and 205, how to decide, and how to run a migration programme. It is for security architects, cryptography owners and security leaders who need a defensible position. Status as checked on 26 October 2026.
Key takeaways
- The national agencies agree: PQC first. The NCSC, NSA, and a joint paper from ANSSI, BSI, NLNCSA and Swedish authorities all recommend post-quantum cryptography as the primary mitigation and limit QKD to niche or research use.
- QKD does not authenticate anyone. It still needs pre-shared keys or classical signatures (in practice post-quantum signatures) to know who is at the other end, so it cannot replace PQC.
- PQC standards are final and deployable. NIST published FIPS 203 (ML-KEM), FIPS 204 (ML-DSA) and FIPS 205 (SLH-DSA) on 13 August 2024, and hybrid ML-KEM key exchange for TLS is in the IETF RFC Editor queue.
- QKD is hardware and distance bound. Commercial fibre systems typically reach about 100 km, longer paths need trusted nodes, and it cannot be delivered in software or as a network service.
- The real work is inventory and agility. Whichever approach is chosen, the deadline pressure comes from discovery: the NCSC asks organisations to finish discovery and planning by 2028 and complete migration by 2035.
The quantum threat in one paragraph
A large, fault tolerant quantum computer running Shor's algorithm would break the public key schemes used for key exchange and signatures today: RSA, Diffie-Hellman and elliptic curve cryptography. Symmetric ciphers such as AES and hash functions are far less affected; Grover's algorithm at most calls for larger keys. Two risks follow. "Harvest now, decrypt later" means encrypted traffic recorded today can be decrypted once such a machine exists, which matters for data that must stay confidential for many years. Forged signatures matter later, but signing keys embedded in devices, firmware and PKI roots have long lifetimes, so migration must start early. For the fundamentals, see the Symmetric vs Asymmetric Encryption Guide and the Post-Quantum Readiness Guide.
How QKD and PQC work
- QKD: two endpoints exchange quantum states (typically single photons) over a dedicated optical fibre or a free-space link. Any measurement by an eavesdropper disturbs the states, so the endpoints can detect interception and distil a shared secret key. That key then feeds a conventional symmetric cipher. QKD only does key agreement: it does not encrypt data, sign anything or identify the parties.
- PQC: new public key algorithms based on mathematical problems believed hard for both classical and quantum computers, such as module lattices and hash functions. They run on existing CPUs, HSMs and smart cards, slot into existing protocols such as TLS, SSH, IPsec and X.509 PKI, and cover both key establishment and digital signatures.
The joint European position paper puts the authentication gap plainly: the classical channel used by QKD must be authenticated, either with keys already shared at both ends or with post-quantum signatures, in which case the security relies on the post-quantum scheme anyway. That single fact shapes every comparison below.
Side-by-side comparison
| Criterion | Quantum key distribution (QKD) | Post-quantum cryptography (PQC) |
|---|---|---|
| What it provides | Key agreement only | Key establishment (ML-KEM) and digital signatures (ML-DSA, SLH-DSA) |
| Authentication | None; needs pre-shared keys or classical/PQ signatures | Built in through signatures and certificates |
| Security basis | Physics in theory; in practice the security of specific hardware | Hardness of mathematical problems, public analysis since the NIST process began |
| Infrastructure | Dedicated fibre or free-space optics, single photon sources and detectors | Software and firmware updates on existing hardware |
| Distance | Commercial systems typically about 100 km; demonstrations at most a few hundred km on fibre | No distance limit; works over the internet |
| End to end security | Not over long fibre paths: trusted nodes hold keys at each hop | Yes, end to end between endpoints |
| Availability risk | Detection of eavesdropping means an attacker can cause denial of service | Same availability profile as current cryptography |
| Standards status | ETSI ISG QKD specifications for interfaces and implementation security; no FIPS standard | FIPS 203, 204, 205 final (August 2024); FN-DSA (FIPS 206) and HQC in progress |
| Government approval | NSA: not recommended for National Security Systems; NCSC: not supported for government or military use | Recommended by NSA, NCSC, ANSSI, BSI and NLNCSA |
| Cost | Very large acquisition and lifecycle costs (joint European paper) | Mostly engineering and testing effort, plus hardware refresh where devices cannot be updated |
What national security agencies say about QKD
| Agency | Publication | Position (paraphrased from the source) |
|---|---|---|
| UK NCSC | Quantum networking technologies (white paper) | Will not support QKD for government or military applications; other sectors should not rely on QKD alone; QKD provides no authentication; PQC is the best mitigation. Anyone still considering QKD should pair it with robust quantum-resistant authentication. |
| US NSA | Quantum Key Distribution (QKD) and Quantum Cryptography (QC) | Does not recommend QKD for National Security Systems unless its limitations are overcome and does not expect to certify QKD products until then. Lists five limitations: partial solution without source authentication, special-purpose equipment that cannot run in software, trusted relays that add cost and insider risk, difficulty securing and validating real hardware, and denial-of-service risk. |
| ANSSI (France), BSI (Germany), NLNCSA (Netherlands), Swedish NCSA and Swedish Armed Forces | Position Paper on Quantum Key Distribution, 26 January 2024 | QKD has functional limitations, needs specialised infrastructure and is not yet sufficiently mature from a security perspective; it can only be applied in niche use cases. Migration to PQC, in hybrid with symmetric keying or classical public key cryptography, is "the clear priority". Research on QKD should continue. |
Not every government is negative on QKD as a research and industrial programme. The EU's European Quantum Communication Infrastructure (EuroQCI) is building terrestrial fibre segments in member states and a space segment with ESA, with the Eagle-1 prototype satellite due to launch in late 2027 and a testing and evaluation infrastructure (NOSTRADAMUS) for certifying QKD technologies. That is an infrastructure and evaluation programme, not advice to enterprises to replace PQC, and the joint European agencies' paper sits alongside it.
The PQC standards: FIPS 203, 204 and 205
| Standard | Algorithm (original name) | Purpose | Status |
|---|---|---|---|
| FIPS 203 | ML-KEM (CRYSTALS-Kyber) | Key encapsulation for general encryption and key establishment | Final, 13 August 2024 |
| FIPS 204 | ML-DSA (CRYSTALS-Dilithium) | Primary digital signature standard | Final, 13 August 2024 |
| FIPS 205 | SLH-DSA (SPHINCS+) | Stateless hash-based signatures; a backup built on different mathematics from ML-DSA | Final, 13 August 2024 |
| FIPS 206 (planned) | FN-DSA (FALCON) | Compact lattice signatures | Draft in preparation; check NIST CSRC for current status |
| HQC (to be standardised) | HQC | Code-based KEM, backup to ML-KEM | Selected 11 March 2025; NIST expected a draft about a year later and a final standard in 2027 |
In protocols, the practical first step is hybrid key exchange. The IETF draft draft-ietf-tls-ecdhe-mlkem defines TLS groups that combine classical ECDHE with ML-KEM: X25519MLKEM768 (code point 0x11EC) as the general purpose option, and SecP256r1MLKEM768 (0x11EB) and SecP384r1MLKEM1024 (0x11ED) for environments that need FIPS-approved curves. As of October 2026 it is in the RFC Editor queue, not yet a published RFC. Hybrid groups protect against harvest-now-decrypt-later while keeping classical security if a flaw is ever found in ML-KEM. NIST IR 8547 (initial public draft, November 2024) sets out NIST's proposed transition away from quantum-vulnerable algorithms; read the current version before fixing internal dates.
Decision guidance: which to use
For almost every organisation the answer is PQC, deployed through crypto-agile systems. QKD is a possible addition in a narrow set of cases, never a substitute.
| Situation | Recommended approach | Why |
|---|---|---|
| Internet-facing services, VPNs, SaaS, APIs | PQC (hybrid ML-KEM in TLS, IPsec, SSH) | Works end to end over any network; QKD cannot reach these paths |
| PKI, code signing, firmware signing | PQC signatures (ML-DSA; SLH-DSA for long-lived roots) | QKD offers no signatures at all |
| Machine identities, workload certificates, API tokens | PQC through PKI and library updates | Scale and automation need software; see the Post-Quantum Migration for NHIs Guide |
| Point-to-point fibre between two owned data centres under 100 km, very high confidentiality, existing budget | PQC as the baseline; QKD only as an additional layer if risk owners accept the cost and the authentication still uses PQC | The only setting where QKD's distance and hardware constraints fit |
| Government or defence systems in the UK or US | PQC per national guidance (CNSA 2.0 in the US) | NCSC and NSA do not support QKD for these systems |
| Research, telecom operators exploring quantum networks | QKD pilots in labs or testbeds alongside PQC deployment | The joint European paper supports continued research; for PQC plans in future mobile networks, see our 6G Security Guide |
- Small organisations: PQC will arrive mostly through vendors: browsers, operating systems, cloud providers and VPN products. Your job is to keep software current, ask suppliers for PQC roadmaps and avoid buying hardware that cannot be updated.
- Large organisations: run a programme with inventory, prioritisation by data lifetime, hybrid pilots and supplier management. Treat any QKD proposal as a business case to be tested against the agency positions above.
- Layering QKD on PQC: where QKD is used, the defensible design combines its keys with a PQC or pre-shared symmetric key, so that a break in either mechanism alone does not expose traffic. This mirrors the hybrid approach the joint European agencies recommend for PQC itself, and it means the QKD link adds defence in depth rather than becoming a single point of trust.
- Questions to ask a QKD vendor: how the classical channel is authenticated; whether trusted nodes are involved and who controls them; what independent evaluation the hardware has passed; behaviour under denial of service; total cost over the equipment lifetime.
How to implement a quantum-safe programme
- Discover (complete by 2028 in the NCSC timeline): build a cryptographic inventory of protocols, libraries, certificates, keys and HSMs, ideally as a CBOM. See the Cryptographic Inventory and CBOM Guide.
- Prioritise by data lifetime and exposure: long-lived confidential data crossing untrusted networks first (harvest-now risk), then long-lived signing roots, then everything else.
- Build agility: centralise cryptographic configuration, remove hard-coded algorithms and make sure key management systems can hold larger PQC keys. See the Crypto-Agility Guide and the Cryptographic Key Management Guide.
- Pilot hybrid key exchange: enable X25519MLKEM768 on a test TLS endpoint, measure handshake size, latency and middlebox compatibility (larger ClientHello messages can break poorly written network devices), and keep classical-only groups as a rollback.
- Roll out in rings: internal services, then partner connections, then customer facing endpoints, with monitoring for handshake failures at each step and a documented switch back to classical groups.
- Migrate signatures and PKI (2031 priority work in the NCSC timeline): plan new PQC root and issuing CAs, test certificate size impacts on constrained devices, and coordinate with relying parties.
- Complete and verify (by 2035): retire quantum-vulnerable algorithms, keep the inventory current and re-test after every major platform change.
The step-by-step migration programme is covered in more depth in the Post-Quantum Cryptography Migration Guide; this page focuses on the QKD versus PQC decision.
Detection and measurement
Monitor what is actually negotiated, not what policies say. TLS termination points, load balancers and VPN concentrators can log the negotiated key exchange group; certificate inventories show signature algorithms; HSM logs show key types in use.
- Percentage of external TLS connections negotiating a hybrid PQC group.
- Percentage of cryptographic assets in the inventory with a recorded owner, algorithm and migration date.
- Number of systems with hard-coded or non-updatable cryptography, trending down.
- Percentage of strategic suppliers with a documented PQC roadmap.
- Handshake failure rate after enabling hybrid groups, per ring.
- For any QKD link: key generation rate, link downtime and time spent falling back to classical keying.
Relevant adversary behaviour today is interception: Adversary-in-the-Middle (T1557) and Network Sniffing (T1040) are how traffic gets harvested for later decryption.
Common failure patterns
- Buying QKD as the quantum strategy. QKD covers one link and no signatures; the PKI, VPNs, code signing and APIs stay exposed.
- Forgetting QKD's authentication dependency. A QKD link authenticated with classical RSA or ECDSA is only as quantum-safe as those signatures.
- Ignoring trusted nodes. Over long fibre paths keys exist in the clear at relay sites, creating insider and physical security risk the NSA explicitly flags.
- Starting with algorithms instead of inventory. Teams switch on PQC in one product while not knowing where RSA keys live across the estate.
- Waiting for "final" everything. FIPS 203, 204 and 205 are final; waiting for FN-DSA or HQC delays protection against harvest-now attacks that are happening now.
- Neglecting key protection. Quantum-safe algorithms do nothing if keys are stolen. The GoTo breach showed the risk: attackers took encrypted backups and also exfiltrated an encryption key for a portion of them, so for that portion the encryption gave no protection. Key management failures beat any cipher, classical or post-quantum.
Framework mapping
| Activity | NIST CSF 2.0 | NIST SP 800-53 Rev 5 | ISO/IEC 27001:2022 Annex A | CIS Controls v8.1 |
|---|---|---|---|---|
| Cryptographic inventory | ID.AM-02 | CM-8 | A.5.9 | Control 1, Control 2 |
| Quantum-safe protection in transit | PR.DS-02 | SC-8, SC-13 | A.8.24 | 3.10 |
| Quantum-safe protection at rest | PR.DS-01 | SC-28, SC-13 | A.8.24 | 3.11 |
| Key establishment and management | PR.DS-02 | SC-12 | A.8.24 | 3.10 |
| Supplier PQC roadmaps | GV.SC-05 | SR-3 | A.5.21 | Control 15 |
Practitioner checklist
- Write a one-page position: PQC is the primary quantum-safe approach; QKD considered only as an additional layer for named point-to-point links.
- Cite the NCSC, NSA and joint ANSSI, BSI and NLNCSA positions in any QKD business case review.
- Complete a cryptographic inventory with owners and data lifetimes, aligned to the NCSC 2028 discovery milestone.
- Pilot hybrid ML-KEM key exchange (X25519MLKEM768) on internal TLS, with rollback to classical groups.
- Require FIPS 203 and FIPS 204 support, or a dated roadmap, from strategic suppliers and HSM vendors.
- Plan PQC PKI: new roots and issuing CAs, certificate size tests on constrained devices.
- Track FIPS 206 (FN-DSA), HQC and the final NIST IR 8547 for date changes.
- If QKD is deployed, authenticate it with PQC or pre-shared keys, document trusted nodes and monitor link availability.
Standards and references
- NIST: First three finalised post-quantum encryption standards (FIPS 203, 204, 205)
- NCSC: Quantum networking technologies
- NSA: Quantum Key Distribution (QKD) and Quantum Cryptography (QC)
- ANSSI: The uses and limits of quantum key distribution (joint position paper)
- NCSC: Timelines for migration to post-quantum cryptography
- NIST: NIST selects HQC as fifth algorithm for post-quantum encryption
- IETF: Post-quantum hybrid ECDHE-MLKEM key agreement for TLS 1.3
- European Commission: European Quantum Communication Infrastructure (EuroQCI)
Related NHI Mgmt Group resources: Post-Quantum Cryptography Migration Guide · Post-Quantum Migration for NHIs Guide · Crypto-Agility Guide · Cryptographic Inventory and CBOM Guide · Encryption Guide