Why Post-Quantum Cryptography Matters for TLS 1.3
TLS protects an enormous amount of Internet traffic, but the security of a TLS connection depends on several different cryptographic operations. The key-establishment stage is especially important to the post-quantum transition: if an attacker records encrypted traffic today, a sufficiently capable future quantum computer could threaten confidentiality when the session's secret depended only on quantum-vulnerable public-key mathematics. Standards bodies are addressing that exposure by defining hybrid key exchange for TLS 1.3.
What changes in a TLS handshake?
In a conventional TLS 1.3 handshake, client and server negotiate cryptographic parameters and derive shared secrets used to protect the connection. Elliptic-curve Diffie–Hellman exchange is widely used for this purpose. Its security relies on mathematical problems that a large, fault-tolerant quantum computer could attack with quantum algorithms. The post-quantum question is therefore not whether TLS uses encryption at all, but how the two parties agree on the secrets that feed that encryption.
NIST's FIPS 203 defines ML-KEM, a key encapsulation mechanism based on the module learning-with-errors problem. A KEM allows parties to establish a shared secret over a public channel. The resulting secret can then support symmetric encryption and authentication. ML-KEM is not itself a replacement for every part of TLS: it does not replace the record-layer encryption algorithm, certificate validation or the entire protocol.
The IETF's RFC 9954, published in July 2026, describes how multiple key-exchange algorithms can be combined within TLS 1.3. The objective is that the resulting exchange can remain secure if all but one of its components resist attack. In the current transition, the practical combination is a traditional elliptic-curve exchange plus a post-quantum KEM.
The standardized hybrid groups
RFC 10024, published in August 2026 as an Internet Standards Track document, specifies three post-quantum/traditional hybrid key-agreement mechanisms for TLS 1.3:
| Named group | Traditional component | Post-quantum component |
|---|---|---|
X25519MLKEM768 | X25519 | ML-KEM-768 |
SecP256r1MLKEM768 | secp256r1 | ML-KEM-768 |
SecP384r1MLKEM1024 | secp384r1 | ML-KEM-1024 |
These are named groups, not three interchangeable spellings for one algorithm. A client and server need compatible support to negotiate a particular group. The ML-KEM suffix indicates the NIST parameter set; the traditional component indicates the elliptic-curve method paired with it. RFC 10024 supplies the protocol definition rather than leaving each vendor to invent its own combination.
The design is called hybrid because both components contribute to key establishment. It does not mean that a TLS connection automatically runs two complete independent TLS handshakes. Nor does the term mean that the post-quantum component has been proved unbreakable. The value of combining different assumptions is that an unforeseen weakness in one component need not immediately expose the shared secret if the other component remains secure under the construction's assumptions.
What hybrid TLS does not solve
A TLS deployment still needs server authentication. Certificates and their signatures are a distinct migration problem from the ephemeral key exchange. Replacing the latter with an ML-KEM hybrid does not by itself turn an RSA- or ECDSA-signed certificate chain into a post-quantum signature chain. NIST standardized ML-DSA under FIPS 204 and SLH-DSA under FIPS 205 for digital signatures, while FIPS 203 addresses key encapsulation. These roles should not be conflated in architecture diagrams or compliance reports.
Likewise, adding ML-KEM does not repair certificate misissuance, weak endpoint security, compromised application secrets or incorrect trust-store configuration. Hybrid key exchange addresses a specific cryptographic risk: the establishment of session keys against classical and future quantum threats. Existing TLS operational controls still matter.
Deployment questions for operators
Suppose an organization runs TLS at a cloud load balancer, an API gateway and an application server. Updating the application library alone may accomplish nothing for public-facing traffic if the load balancer actually terminates the external TLS session. The team should first identify each termination point, record the negotiated groups in testing and check whether clients, proxies and monitoring systems handle the new handshake material correctly.
A realistic pilot should measure connection establishment under normal and constrained network conditions. ML-KEM uses different public-key and ciphertext sizes from classical elliptic-curve exchanges, so implementers should examine handshake bytes, fragmentation, retransmissions and compatibility with middleboxes. The relevant comparison is an end-to-end connection through the actual network path, not a standalone algorithm benchmark.
Operators should also decide what happens when a peer does not support a hybrid group. A deployment can only claim post-quantum protection for connections that actually negotiate an appropriate mechanism. A server configuration flag, by itself, does not establish that every client session received the new protection. Negotiation telemetry and clearly defined fallback policy are essential to making deployment claims accurate.
Why the timing matters
NIST finalized FIPS 203 in August 2024, while the IETF published the TLS hybrid design and named-group standards in 2026. Those milestones make interoperable implementation more concrete, but do not mean that every TLS stack has deployed the new groups. The urgency stems partly from the possibility that encrypted data captured today could remain valuable to an adversary in the future. Organizations should therefore prioritize long-lived sensitive data, test interoperable implementations and distinguish hybrid key establishment from the separate work of migrating digital signatures.
