Skip to content
All guides
Migration engineering

A post-quantum migration roadmap: sequencing the move to PQC

8 min
TL;DR

Read this first

A post-quantum migration is a sequencing problem, not a single cutover. Five phases: inventory what you have, prioritize by risk and deadline, protect exposed traffic with hybrid TLS, migrate signatures and PKI, then bake in crypto-agility so the next change is cheap. Sequence by data confidentiality lifetime, not by how easy a system is to touch — the systems holding your longest-lived secrets go first even when they are the hardest.

Every organization that takes the quantum threat seriously arrives at the same question: not whether to migrate to post-quantum cryptography, but in what order. The algorithms are standardized and the deadlines are published; the hard part is turning a sprawling estate of TLS endpoints, PKI, code-signing pipelines, and embedded firmware into a sequenced plan that closes the highest risks first without breaking anything in flight. A roadmap is what converts an inventory into that plan.

Phase 1 — Build the inventory

You cannot sequence what you cannot see. The roadmap begins with a cryptographic inventory: a machine-readable map of where every algorithm, key, certificate, and protocol version actually lives — in application code, libraries, TLS configurations, hardware modules, and third-party dependencies. Without it, prioritization is guesswork and you will discover the forgotten key-exchange in a legacy service the hard way. The inventory is the source of truth every later phase reads from.

Phase 2 — Prioritize by risk and deadline

With the estate mapped, rank each data class two ways. First by exposure, using Mosca's inequality: if data confidentiality lifetime (X) plus migration time (Y) exceeds time-to-quantum (Z), the data is already at risk from harvest-now-decrypt-later. Second by mandate, using the deadlines that apply to you:

  • **CNSA 2.0** — NSA requires post-quantum algorithms exclusively across national security systems, with software and firmware signing first and full adoption by 2033.
  • **NIST IR 8547** — the federal timeline deprecates RSA-2048 and ECC after 2030 and disallows them after 2035, setting the pace most enterprises should plan against.
  • **OMB M-23-02** — obliges US federal agencies to inventory quantum-vulnerable systems and report them, making Phase 1 a compliance artifact, not just an engineering one.
Decision

Sequence by data lifetime, not by ease

The tempting order is to migrate whatever is easiest first — a stateless service you fully control. Resist it. The correct first movers are the systems holding secrets whose confidentiality must outlive the time-to-quantum, even when those systems are the oldest and most fragile. A quick win on short-lived data does nothing for the archive an adversary is recording today; a hard migration of long-lived data closes the exposure that actually matters.

Phase 3 — Protect exposed traffic first

The one class of risk that is accruing right now is recorded key exchange. So the first migration to ship, ahead of the rest of the estate, is hybrid TLS key exchange: wrapping a post-quantum KEM around the classical exchange (see hybrid key exchange) so recorded handshakes cannot be decrypted retroactively. This closes the harvest-now-decrypt-later window on externally facing traffic while remaining interoperable with classical clients — you get protection before the full programme is anywhere near done.

Phase 4 — Migrate signatures and PKI

Signatures are not exposed to harvesting — a forged signature is only useful after a quantum computer exists — so they follow key exchange rather than lead it. But they carry their own engineering weight: ML-DSA (FIPS 204) and SLH-DSA (FIPS 205) produce signatures and keys far larger than ECDSA, which stresses certificate sizes, handshake budgets, protocol limits, and storage. For firmware and long-lived boot chains, the stateful hash-based schemes LMS and XMSS (SP 800-208) are the standardized choice, with careful state management to avoid one-time-key reuse. Plan PKI, code signing, and document signing as a distinct workstream.

Phase 5 — Bake in crypto-agility

The last phase makes the next migration cheap. Crypto-agility means algorithms are named in configuration, negotiated at runtime, and swappable without redeploying application logic — so the transition after this one is a config change, not another multi-year programme. Treat the current migration as the forcing function that finally decouples your systems from any single algorithm.

Avoiding breakage on the way

Hybrids exist precisely so migration is not a flag day: they keep backward compatibility with peers that have not moved. Roll out in stages behind feature flags, run interoperability testing against real client populations before enforcing, and keep a rollback path for every change until the post-quantum side is proven in production. The goal is a continuous ratchet — never a single irreversible cutover.

How quantakrypto helps

We run this roadmap end to end: an audit produces the inventory and the risk-and-deadline ranking, a migration engagement sequences and ships the phased rollout — hybrid TLS first, then signatures and PKI, then agility — and training leaves your teams able to reason about X + Y > Z and design for algorithm change on their own. Start with the inventory: everything downstream depends on knowing what you have.

Frequently asked questions

What should we migrate first?

Externally facing key exchange, via hybrid TLS. It is the only class of risk actively accruing today through harvest-now-decrypt-later, and hybrids let you deploy protection without waiting for the rest of the estate. Signatures and PKI follow, because forged signatures are only useful after a quantum computer exists and cannot be produced retroactively.

Should we sequence by how easy a system is to migrate?

No. Sequence by data confidentiality lifetime. A system holding secrets that must stay confidential for a decade belongs early even if it is the hardest to touch, because that is exactly the data an adversary gains by recording traffic now and decrypting later. Easy wins on short-lived data feel productive but close no real exposure.

Why are post-quantum signatures called out as a distinct phase?

Because their size changes system behavior. ML-DSA (FIPS 204) and SLH-DSA (FIPS 205) keys and signatures are much larger than ECDSA, stressing certificate chains, handshake sizes, protocol limits, and storage. Firmware often needs the stateful hash-based schemes LMS/XMSS (SP 800-208), which require disciplined key-state management. These constraints make PKI and code signing their own workstream rather than a step within TLS.

How do we migrate without breaking production?

Use hybrids for backward compatibility, roll out in stages behind feature flags, run interoperability testing against real clients before enforcing, and keep a rollback path for every change. Combined with crypto-agility — algorithms in configuration, negotiated at runtime — the migration becomes a reversible ratchet rather than a single flag-day cutover.

Take the next step

Related reading

References

Get started

Turn quantum risk into a credential.

Book a discovery call and get an indicative scope and pricing for your organisation.