Choosing a Quantum-Safe Encryptor: 10 Evaluation Criteria
“Is it quantum-safe?” is close to a useless question by itself. Nearly every network encryption vendor will now answer yes. In a narrow sense, most of them won’t be wrong.
The real differences between products show up in the questions procurement teams don’t always think to ask: what happens when the next algorithm changes, whether the platform still hits line rate with post-quantum key exchange switched on, whether it can protect the legacy and OT links that can never themselves be upgraded, and whether a compliance team can actually produce an audit trail from it.
This page is a checklist for those questions. It includes ten criteria, assembled from what NIST, CNSA 2.0, DORA, and the EU’s post-quantum roadmap already require, plus what tends to break in practice once a “quantum-safe” box has been checked and deployment starts.
The Cryptography Has to Be Able to Change
1. Crypto-agility: a software-defined engine instead of fixed-function hardware. Every algorithm on the table today is a snapshot of where standards happen to be right now but not necessarily a permanent decision. This applies to ML-KEM, the classical algorithms it runs alongside in hybrid mode, and whichever signature scheme eventually succeeds FN-DSA The single most useful question in any evaluation is whether a future algorithm ships as a software update or requires a hardware refresh. Everything else on this list is easier to fix later than that one.
2. Hybrid key exchange as the working default. Hybrid is where a classical algorithm run alongside a post-quantum one, so connection stays secure even if one method is later broken. NIST’s own guidance, CNSA 2.0, and the EU’s transition timeline all treat hybrid mode as the multi-year operating mode for this transition. This is not a temporary stopgap on the way to “pure” post-quantum.
3. No single point of cryptographic failure. NIST added HQC as a second, structurally independent post-quantum key-encapsulation mechanism specifically so the standard wouldn’t rest entirely on lattice-based mathematics being sound forever. Ask whether the platform can run more than one post-quantum algorithm family today, and whether adding the next one, like FN-DSA once finalized, or whatever comes after it, is a configuration change or a new product line.
It Has to Work on Your Actual Network
4. Wire-speed performance at the throughput you run. Post-quantum key exchange adds computational overhead. The question that matters isn’t whether a platform can technically do quantum-safe encryption but whether its symmetric bulk encryption still holds line rate at your actual interconnect speed. It shouldn’t matter whether that’s a 1 Gbps branch link or a 100 Gbps data-center interconnect, but the beef is performance with hybrid or full post-quantum modes switched on. Ask for performance figures measured that way, not classical-only benchmarks with a quantum-safe asterisk.
5. Layer 2 and Layer 3 coverage that matches your topology. Encryption applied at the wrong OSI layer for a given link doesn’t just underperform but it can stop working outright once a topology changes, or fail to support a use case (multicast, jumbo frames, protocol-agnostic transport) the wrong layer was never built for. A platform that supports both Layer 2 (Ethernet) and Layer 3 (IP) keeps that topology decision from getting locked into a purchase you may later regret. Ideally, the solution supports both layers on the same hardware, without a separate product line for each.
It Has to Survive an Audit
6. Automated, auditable key management. Manual key rotation doesn’t scale past a handful of links, and increasingly it’s also the specific thing an auditor asks about directly. Look for automated key generation, rotation, and renewal; forward secrecy by default; and a record of what happened that a compliance team can produce on request.
7. Independent certification and standards conformance appropriate to your sector. Certification requirements vary sharply by buyer. A US defense or federal procurement typically needs FIPS 140-3 module validation; EU and international government buyers often look for Common Criteria, EU-Restricted or NATO-Restricted approval; US national-security-adjacent buyers increasingly ask about CNSA 2.0 algorithm alignment (AES-256, ML-KEM-1024, ML-DSA-87); financial entities need to map a product against DORA’s RTS on ICT risk management. Certification alone doesn’t satisfy the other nine criteria, but its absence is a specific, checkable gap — ask which certifications a vendor holds today versus which are “in progress,” and get the distinction in writing.
8. Transparent deployment across legacy, OT, and systems you don’t control. The highest-value links often run through equipment that will never itself speak post-quantum cryptography. These include SCADA controllers, decades-old SSH servers and VPN appliances, vendor-locked embedded systems, partner and third-party connections you don’t own. A network-layer encryptor that protects that traffic without requiring the endpoints to change is the difference between a migration measured in months and one measured in whenever the vendor in the supply chain gets around to it. For some equipment, that may be never.
9. Central management and compliance-grade visibility. DORA’s technical standards and the EU PQC roadmap’s first milestone both start with the same ask: an accurate, current inventory of what cryptography is running where, such as which algorithm, on which link, since when. A centrally managed solution will give you an answer from a single pane of glass.
It Has to Keep Running
10. Operational resilience and deployment control. Look for redundancy and failover so a single appliance failure doesn’t take a link dark. For government, defense, financial, and critical-infrastructure players in particular, an important requirement is clarity on where keys are generated and held, and whether the deployment model matches the organization’s sovereignty and data-residency needs.
How to Run the Evaluation
- Put all ten criteria in the RFP itself, not just “quantum-safe: yes/no” as a line item — ask each vendor to answer them individually, in writing.
- Ask for performance numbers with hybrid or post-quantum modes enabled, specifically, not classical-only figures with a quantum-safe footnote.
- Ask how a future algorithm change is delivered, and get the answer in writing before you standardize on a platform.
- Pilot on a subset of links that includes at least one legacy or OT connection, not only a green-field one — that’s where deployment friction shows up.
- Score certifications as “held” versus “planned,” and weight the decision toward what’s been achieved, not what’s on a roadmap slide.
Get a head start with network encryption
SSH's NQX applies quantum-safe, wire-speed encryption (up to 100 Gbps) at Layer 2 and Layer 3 to exactly the links this page describes. It is deployed transparently, without touching the systems generating the traffic.
Start by exploring NQX quantum-safe network encryption.
FAQ
What certifications should a quantum-safe network encryptor have? That depends on your sector and country. Government and defense buyers usually need a national approval for classified data. US federal buyers need FIPS 140-3 validation. Many others look for Common Criteria, EU Restricted or NATO Restricted requirements. Ask which certifications are held today and which are only in progress, and get the answer in writing.
How do I verify a vendor's quantum-safe performance claims? Ask for throughput and latency measured with hybrid or post-quantum key exchange switched on, at both small (64-byte) and large frame sizes, on the model you would buy. Then confirm the figures in a pilot on your own links, including at least one legacy or OT connection.
Is "quantum-ready" the same as "quantum-safe"? No. "Quantum-ready" usually means the product can be upgraded later. "Quantum-safe" means post-quantum key exchange is running now. Because of harvest-now-decrypt-later collection, only quantum-safe protects the traffic you send today. Ask which one a vendor means, and on which models.
Can our existing network encryptors be upgraded instead of replaced? Sometimes. If the vendor can add ML-KEM through a software or firmware update on the hardware you own, upgrading may be enough. If post-quantum support needs new hardware, or runs only on newer models, budget for replacement and use the change to require crypto-agility.
Should non-US buyers require CNSA 2.0 alignment? It's a useful reference point but not a must-have. CNSA 2.0 (AES-256, ML-KEM-1024, ML-DSA-87) is mandatory only for US national security systems. European buyers should follow their national authority, which in many countries means hybrid key exchange and acceptance of algorithms such as FrodoKEM.
How does SSH NQX map to these criteria? NQX provides a software-defined crypto engine, hybrid ML-KEM and FrodoKEM key exchange, simultaneous Layer 2 and Layer 3 encryption up to 100 Gbps, automated key management and central management. It is approved by Finland's NCSA-FI for information up to TL III (EU Confidential). EU-Restricted and NATO-Restricted certifications are in progress.
