Explore our new Cybersecurity Terms Glossary, explained by our experts.
Cybersecurity Terms Glossary
Hybrid key exchange runs a classical and a post-quantum algorithm together and combines both results, so the session stays secure as long as either one holds.

What Is Hybrid Key Exchange?
Post-quantum algorithms are new. ML-KEM was standardised in August 2024, and while it has been analysed heavily, no lattice scheme has the thirty-year attack history that elliptic curves have. Classical algorithms, meanwhile, are well understood and known to fall to a sufficiently large quantum computer. Hybrid key exchange refuses to choose: run both, concatenate the two shared secrets, and feed the result into the key derivation function. An attacker must break both to recover the session key.
It is a hedge against two different failure modes at once — a quantum breakthrough on one side, and an unexpected cryptanalytic result against a young algorithm on the other. That second risk is not hypothetical: SIKE, a NIST alternate candidate, was broken on a single laptop in 2022, after years of scrutiny.
In TLS 1.3 the client offers a hybrid named group, for example X25519MLKEM768. The ClientHello carries both an X25519 public key and an ML-KEM encapsulation key. The server performs the classical Diffie-Hellman operation and the ML-KEM encapsulation, then both sides concatenate the two shared secrets and run the standard key schedule over the combination.
Because the derivation treats the concatenation as a single input, breaking one component is not enough: without the other half, the derived key remains unpredictable. The IETF hybrid design draft specifies this construction, and the equivalent approach appears in SSH, IKEv2, and messaging protocols.
sntrup761x25519-sha512 as the default key exchange in version 9.0, released in April 2022, making it the first widely used protocol to deploy hybrid post-quantum key agreement by default.X25519MLKEM768 later that year; Firefox and the major CDNs followed. By 2025 a substantial share of all human-initiated HTTPS traffic was already post-quantum protected.An X25519 key share is 32 bytes. An ML-KEM-768 encapsulation key is 1,184 bytes, and the ciphertext coming back is 1,088 bytes. The ClientHello therefore grows past a single network packet, and that has consequences.
Some middleboxes, load balancers, and embedded TLS stacks assume the first flight fits in one segment, or cap the ClientHello size, and they fail in confusing ways: connections hang rather than returning an error, or the handshake fails only for certain clients. Cloudflare and Chrome both documented this during rollout. If you enable hybrid groups and see intermittent connection failures from a subset of users, this is the first thing to test.
Latency impact is otherwise modest. The lattice operations are fast, often faster than the elliptic curve half, so the cost is bandwidth rather than computation.
X25519MLKEM768: OpenSSL 3.5 or later, recent nginx and Apache builds, or a CDN that offers it as a setting.Key exchange is only half the handshake. Authentication still relies on classical signatures, because post-quantum certificate chains are much larger and the public PKI has not yet moved. That is an acceptable position for now: an attacker cannot retrospectively forge a signature on a handshake that already completed, so the confidentiality risk is addressed first and the authentication risk follows the certificate ecosystem’s own timetable.
National guidance differs slightly on how long hybrids should last. The NCSC and Germany’s BSI both recommend hybrid constructions through the transition, while NIST has indicated that pure post-quantum deployment is acceptable once confidence is established. Either way, hybrid is the right configuration today, and verifying it end to end across an estate is work we include in our network security services.
Because it is relatively new. Combining it with X25519 means a future break of either algorithm leaves the session secure, which costs only a little bandwidth. The SIKE candidate was broken in 2022 after years of analysis, and that episode is the standing argument for hybrids during the transition period.
Computationally, barely: ML-KEM operations are fast, sometimes faster than the elliptic curve half. The real cost is handshake size, since the key share grows by more than a kilobyte in each direction. On normal networks the impact is a fraction of a millisecond; on lossy or constrained links it is more noticeable.
In Chrome, the security panel in developer tools reports the negotiated key exchange group, and X25519MLKEM768 will be shown where it is active. From the command line, recent openssl s_client builds print the negotiated group. Server-side, your TLS logs or CDN analytics will usually break down connections by group.
Not yet. Signatures cannot be forged retrospectively, so protecting confidentiality through key exchange is the urgent step. Post-quantum certificates will follow once the CA ecosystem, trust stores, and protocol size limits can accommodate much larger signatures and public keys, which is expected to take several more years.
You May Be Interested In These
We use a few cookies that are needed to make this website work, and they are always on. We would also like to set analytics cookies, to see which pages are useful; functional cookies, so that our live chat works; and advertising cookies, to measure our campaigns. We will not set any of those until you say yes, and you can change your mind at any time from the Cookie Settings link in the footer.
Choose the cookies you are happy for us to use. Strictly necessary cookies are always on; everything else is off until you turn it on.
These cookies keep the website working: they hold your security checks, your form submissions and the choice you make here. The site cannot run without them, so they are always on.
These let our live chat open and remember your conversation, and let the newsletter sign-up work. Switch them off and the chat window will not appear.
These tell us which pages are read and how people find us, so that we can improve the site. The reports are about groups of visitors, never about you personally.
These let us measure our advertising and show you our adverts on other websites, such as LinkedIn, Google and Meta. Switch them off and you will still see adverts, but they will be less relevant.