POST-QUANTUM CRYPTOGRAPHY / ADVANCED

Hybrid Post-Quantum Cryptography In Practice: X25519MLKEM768, Composites And Agency Views

Most post-quantum deployments today pair a new algorithm with a classical one. Here is how hybrid TLS works, who has shipped it, and why the NSA and European agencies disagree on whether it is needed.

Checked against primary sources and independently reviewed on . Sources are listed at the end.

A hybrid scheme combines a post-quantum algorithm with a classical one so that an attacker has to break both. The idea is a hedge. If the new algorithm turns out to have a flaw, the classical one still protects today’s traffic. If a quantum computer arrives, the post-quantum one still protects it. NIST’s draft transition guidance describes hybrids in exactly these terms, while warning that they add complexity and are usually expected to be temporary.1

Hybrid key establishment is already the default in much of the software that secures the internet. This article explains how it works in TLS, where else it has shipped, how hybrid signatures differ, and where national security agencies part ways.

How Hybrid TLS Works

The TLS 1.3 handshake lets a client offer a “key share” for a named group, and the server answers with its own. In August 2026 the IETF published RFC 10024, a Proposed Standard that defines three hybrid groups combining ML-KEM with elliptic curve Diffie-Hellman.2 The first of them, X25519MLKEM768, pairs ML-KEM-768 with X25519. The RFC describes it as widely deployed, and it is the only one of the three marked as recommended in the IANA registry.

  1. Client Sends Its Hybrid Key Share

    The ClientHello carries an ML-KEM-768 encapsulation key and an X25519 public key together: 1,216 bytes in all.

  2. Server Responds

    The server encapsulates a secret to the ML-KEM key and adds its own X25519 public key: a 1,120-byte share in the ServerHello.

  3. Both Sides Compute Two Secrets

    Each side now holds a 32-byte ML-KEM shared secret and a 32-byte X25519 shared secret.

  4. Secrets Are Joined

    The two are concatenated into a 64-byte secret, ML-KEM first, and fed into the normal TLS 1.3 key schedule.

  5. Traffic Is Encrypted As Usual

    From here the connection uses AES or ChaCha20 as before. An attacker would need to break both ML-KEM and X25519 to recover the keys.

A TLS 1.3 handshake using X25519MLKEM768, with sizes from RFC 10024.

RFC 10024 also defines SecP256r1MLKEM768 and SecP384r1MLKEM1024, which use NIST curves. In those two groups the elliptic curve secret comes first; in X25519MLKEM768 the ML-KEM secret comes first. The ordering lets each variant fit FIPS-approved key derivation, so that a FIPS-validated component sits in the expected position.2 The RFC also marks the earlier pre-standard Kyber groups, such as X25519Kyber768Draft00, as obsolete.

Who Has Shipped It

Deployment ran well ahead of the final RFC, based on earlier IETF drafts.

  • Chrome added support for hybrid X25519Kyber768 in Chrome 116 in 20233 and announced in 2024 a switch to the standard ML-KEM group from Chrome 131.4
  • OpenSSL 3.5.0, released on 8 April 2025, added ML-KEM, ML-DSA and SLH-DSA and made X25519MLKEM768 one of its default TLS key shares.5
  • OpenSSH made the hybrid sntrup761x25519-sha512 method its default in version 9.0 in April 2022, citing the risk of traffic being captured now and decrypted later.6 Version 10.0, released on 9 April 2025, switched the default to mlkem768x25519-sha256.7
  • Apple announced iMessage PQ3 on 21 February 2024, starting with iOS 17.4. It combines Kyber with elliptic curve keys at setup and adds periodic post-quantum rekeying, roughly every 50 messages and at least every seven days.8
  • Signal added Kyber to its initial key agreement with PQXDH in September 2023.9 On 2 October 2025 it announced SPQR, an ML-KEM-768 ratchet that runs beside its existing Double Ratchet, with keys from both mixed so that an attacker must break both.10

Hybrid Signatures And Composite Certificates

Hybrid key establishment protects confidentiality. Hybrid signatures protect authenticity, and they are harder. A certificate that carries two public keys and two signatures is larger, and every relying party has to understand the format.

The IETF’s main effort here is composite ML-DSA for X.509 certificates. It defines 18 combinations that pair ML-DSA-44, 65 or 87 with RSA, ECDSA, Ed25519 or Ed448, so that one composite key and signature behave like a single algorithm. As of 7 October 2026 the latest version is draft 19, dated 21 April 2026; it has been submitted for publication and is with the RFC Editor, but it is not yet an RFC.11

Where Agencies Disagree

National agencies agree on the destination, which is post-quantum cryptography, but not on whether hybrids are needed along the way.

AgencyPosition On HybridsNotable Detail
NSA (US, national security systems)Not required; not to be used on mission systems except where NSA itself recommends itTrusts CNSA 2.0 algorithms alone and warns hybrids add complexity and a second transition. IKEv2 is the example exception it gives.
NCSC (UK)Acceptable as an interim measureHybrids should sit within a flexible framework that allows a straightforward move to PQC-only later.
ANSSI (France)Mandatory in its regulated scope; strongly recommended elsewhereNot needed systematically for hash-based signatures such as SLH-DSA, XMSS or LMS.
BSI (Germany)Recommends post-quantum schemes only in hybrid formApplies to key agreement and signatures; hash-based signatures may in principle be used alone.
Agency positions on PQ/T hybrids, from the documents cited. Positions may be updated; check the source before relying on them.

The NSA’s reasoning is set out in its CNSA 2.0 FAQ. It is confident in ML-KEM and ML-DSA, will not require hybrids for security, and notes that more products fail through implementation or configuration errors than through broken algorithms, so extra complexity can reduce security. It also tells national security system owners not to use hybrid or other non-standard post-quantum solutions on mission systems, apart from exceptions that NSA itself recommends for standards or interoperability reasons. IKEv2 is one: its message size limits mean NSA expects to keep classical key establishment there, reinforced with ML-KEM-1024.12 The UK NCSC takes a middle position: if an organisation chooses a hybrid, it should treat it as interim.13

France and Germany are more cautious about the new algorithms themselves. ANSSI’s December 2023 paper says end products that include post-quantum protection shall implement hybridation, with an exception for hash-based signatures.14 Its current FAQ stresses hybridation in the short and medium term, makes it mandatory within its regulated scope (including certified products) and recommends it, without imposing it, everywhere else.15 BSI’s technical guideline, version 2026-01, recommends post-quantum key agreement and signatures only in combination with classical schemes, noting that these methods are comparatively new and less studied, especially in terms of implementation security.16

Outside US national security systems, hybrids fit the European guidance and are already the default in much mainstream software, so they are the usual starting point. Whatever the choice, it helps to keep the configuration flexible enough to drop the classical half later, and anyone supplying national security systems should check NSA’s specific profiles and approved exceptions rather than assume a hybrid is acceptable.

Footnotes

  1. NIST, IR 8547 (initial public draft), “Transition to Post-Quantum Cryptography Standards”, November 2024, section 3.2. nvlpubs.nist.gov ↩

  2. K. Kwiatkowski, P. Kampanakis, B. E. Westerbaan and D. Stebila, RFC 10024, “Post-Quantum Traditional (PQ/T) Hybrid Key Agreement Mechanisms for TLS 1.3”, IETF, August 2026. rfc-editor.org ↩ ↩2

  3. Chromium Blog, “Protecting Chrome Traffic with Hybrid Kyber KEM”, August 2023. blog.chromium.org ↩

  4. Google, “A new path for Kyber on the web”, Google Online Security Blog, 13 September 2024. security.googleblog.com ↩

  5. OpenSSL Project, “NEWS for OpenSSL 3.5”, OpenSSL 3.5.0 entry, 8 April 2025. github.com ↩

  6. OpenSSH, “OpenSSH 9.0 release notes”, 8 April 2022. openssh.org ↩

  7. OpenSSH, “OpenSSH 10.0 release notes”, 9 April 2025. openssh.org ↩

  8. Apple Security Engineering and Architecture, “iMessage with PQ3: The new state of the art in quantum-secure messaging at scale”, 21 February 2024. security.apple.com ↩

  9. Signal, “Quantum Resistance and the Signal Protocol”, 19 September 2023. signal.org ↩

  10. Signal, “Signal Protocol and Post-Quantum Ratchets”, 2 October 2025. signal.org ↩

  11. IETF, “Composite Module-Lattice-Based Digital Signature Algorithm (ML-DSA) for use in X.509 Public Key Infrastructure”, draft-ietf-lamps-pq-composite-sigs-19, 21 April 2026, Datatracker status checked 7 October 2026. datatracker.ietf.org ↩

  12. NSA, “The Commercial National Security Algorithm Suite 2.0 and Quantum Computing FAQ”, version 2.1, December 2024. media.defense.gov ↩

  13. UK National Cyber Security Centre, “Next steps in preparing for post-quantum cryptography”, 14 August 2024. ncsc.gov.uk ↩

  14. ANSSI, “ANSSI views on the Post-Quantum Cryptography transition (2023 follow up)”, 21 December 2023. messervices.cyber.gouv.fr ↩

  15. ANSSI, “FaQ sur la Cryptographie post-quantique (PQC)” (in French), checked 7 October 2026. cyber.gouv.fr ↩

  16. BSI, TR-02102-1, “Cryptographic Mechanisms: Recommendations and Key Lengths”, version 2026-01, 23 January 2026. bsi.bund.de ↩

Knowledge Hub content is general information. It is not legal advice, a compliance certification, a guarantee of security or a substitute for an assessment of your own systems. Standards and rules change; check the sources for the latest position.