Post-quantum migration starts with an inventory, not with an algorithm

What the publication says

On 13 August 2024, NIST published FIPS 203, standardising ML-KEM, a lattice-based key-encapsulation mechanism, in three parameter sets (512, 768, 1024). The selection phase is over: what to deploy is settled, and libraries are catching up.

Why that does not solve your problem

The algorithm was never the hard part. The hard part is that the threat is already running: an attacker capturing encrypted traffic today does not need a quantum computer today — storing and waiting is enough. That is the harvest now, decrypt later scenario, and it makes the question urgent for any data whose value outlasts ten years: health records, industrial secrets, root keys, legal instruments.

What we actually find on assignment

Cryptography is never where the organisation believes it is. TLS terminated at the load balancer is the visible part and, in practice, the easiest to handle. Underneath, we consistently find:

• firmware signatures and OTA update roots of trust;
• site-to-site IPsec tunnels configured once and never revisited;
• HSMs, key ceremonies and written procedures built around frozen algorithms;
• smart cards and secure elements whose hardware lifecycle runs into years;
• internal certificates issued for ten years by an authority nobody administers any more.

Industrial equipment commissioned in 2026 with an RSA-2048 root will still be running in 2040. That lifetime, not algorithm availability, sets the schedule.

The sequence that works

1. Cryptographic inventory. Where, which algorithm, which key size, who administers it, until when. A CBOM, in the same spirit as an SBOM.
2. Rank by dual lifetime: that of the protected data and that of the equipment protecting it. The intersection gives the true priority order — not the declared criticality level.
3. Crypto-agility before migration. Being able to change algorithm without changing the product beats migrating once, under pressure.
4. Hybridisation on exposed traffic. France's ANSSI considers it essential: combine a classical and a post-quantum algorithm so that a weakness in either does not bring down the whole.

What this means for you

If the answer to "which algorithms run where, on which equipment, and until when" fits in one sentence, the migration has not started. The inventory is not a preliminary formality: it is 80 % of the work, and it is the one deliverable that stays useful whatever comes next.