Quantum-safe migration: build the inventory before choosing algorithms.

The June 2026 policy surface is unusually clear. The White House executive order on advanced cryptographic attacks and OMB M-26-15 move U.S. federal agencies toward execution planning for post-quantum cryptography. CISA's product-category guidance gives buyers a practical lens for finding software and hardware that depend on public-key cryptography.

Europe is also moving from principle to roadmap. The NIS Cooperation Group roadmap, supported by the European Commission, creates a synchronized transition path for Member States, while EuroQCI keeps a separate quantum-communication track for infrastructure where physics-based secure links may be justified.

That combination matters. The enterprise question is no longer whether quantum-safe cryptography should be considered. It is how to inventory dependencies, prioritize high-risk systems, test standards, coordinate vendors, and preserve evidence that the transition was controlled.

Quantum cryptography is often used loosely. For enterprise planning, the useful split is sharper. Post-quantum cryptography is a standards and software migration path: algorithms designed to resist quantum attacks while running on classical computers. Quantum key distribution is a physical-layer communication approach that can be valuable in narrow high-assurance links but is not a general replacement for internet-scale cryptographic migration.

A serious migration plan should avoid treating QKD as a universal answer. The practical control model is crypto-agility: inventory, algorithm selection, hybrid rollout, vendor validation, policy gates, and evidence retention. QKD belongs only where link budget, geography, physical control, and compliance value justify specialized infrastructure.

Most organizations will not fail because they cannot name ML-KEM. They will fail because cryptographic usage is scattered across applications, devices, SaaS contracts, embedded firmware, certificate authorities, VPN gateways, message queues, databases, and partner APIs.

A practical operating layer has to discover cryptographic assets, classify risk, assign owners, request vendor attestations, run staged tests, approve cutovers, and export a review pack for security, legal, and procurement teams.

The first phase should not be a generic PQC workshop. It should be a scoped assessment of long-lived data, identity systems, signing infrastructure, device fleets, regulated customer data, and third-party dependencies that are expensive to update later.

A credible first phase can produce an asset map, a PQC impact score, vendor outreach pack, algorithm decision register, pilot migration candidates, and a board-level timeline aligned to U.S. and EU policy pressure.

The first failure mode is inventory illusion. A spreadsheet of TLS endpoints is not a cryptographic inventory, because usage hides in embedded firmware, SaaS contracts, certificate authorities, message queues, and partner APIs. Programs that declare coverage early tend to discover the expensive dependencies during cutover, when options are worst. Discovery has to run as a continuous control, not a one-time scan.

The second failure mode is promise collapse. Teams that buy or sell QKD as a general quantum-safe answer inherit an architecture that cannot cover software, signing, PKI, or device identity. PQC and QKD answer different questions, and a migration plan that blurs them will not survive review. The boundary between them is itself a decision that needs an owner and a written justification.

The third failure mode is cutover without rollback. Changing production trust anchors before hybrid pilots complete removes the fallback path that the readiness formula depends on. Rollback confidence is a multiplier, not an afterthought; if it is zero, readiness is zero. Every cutover approval should reference a tested fallback state.

The readiness formula in this article multiplies inventory coverage, algorithm fit, vendor status, test evidence, and rollback confidence. Each factor should be an observable quantity, not an opinion. That means instrumenting the migration itself: which assets are discovered, which have owners, which have vendor attestations, which have passed staged tests, and which have approved fallbacks.

Instrument inventory coverage first, because it gates everything downstream. An algorithm decision register means little if a third of the estate is undiscovered. Track coverage by system class, separating key establishment, digital signatures, code signing, device identity, and archival verification, since each class migrates on a different path and timeline.

Then instrument the decision trail. Every migration decision should carry owner, source, risk, test state, fallback, and approval status, and the whole record should export as a review pack for security, legal, and procurement. In the Neura Parse stack, NowFlow coordinates that work as approval-gated workflows, and QFlow Studio preserves the standards context, assumptions, and test evidence behind each decision.

Quantum-safe is becoming a label before it becomes a property. The credible evaluation lens is the one buyers already have: CISA's product categories identify the hardware and software classes that use or transition to PQC standards, and the finalized NIST standards define what an implementation should conform to. A vendor claim that names neither is a marketing statement, not evidence.

Attestations should be specific and dated. Ask which algorithms ship today, which sit on the roadmap, whether hybrid modes are supported for staged rollout, and how certificates, libraries, and policies can be rotated without a rewrite. That last question is the crypto-agility test, and it separates vendors that can survive an algorithm change from vendors that cannot.

Score vendor status as a readiness input, not a procurement checkbox. The readiness formula treats vendor status as a multiplier, so an estate full of unresponsive vendors caps the whole program. Feed attestations, gaps, and outreach state into the same decision register that holds internal migration decisions, and revisit the scores as products move against the CISA categories.