PQC Zero Trust

Post-quantum cryptography within a Zero Trust architecture

Zero Trust replaces network location with cryptographic proof of identity for every request. That proof is delivered by certificates and signatures, which is exactly what quantum computing defeats. A Zero Trust estate has more cryptographic dependencies than a perimeter one, and also migrates them faster.

Why Zero Trust concentrates the exposure

The premise of Zero Trust, set out in NIST SP 800-207, is that no request is trusted on the basis of where it originates. Every access decision is made against verified identity, device posture and policy, evaluated per request.

Verification is cryptographic in every case. Workloads authenticate to each other with mutual TLS and short-lived certificates. Devices present attestation backed by hardware keys. Users present tokens signed by an identity provider. Policy decisions are conveyed as signed assertions. Service meshes issue and rotate identity certificates continuously.

If signatures can be forged, each of those controls fails in a way that grants access rather than merely disclosing information. A forged workload identity certificate produces a service that other services trust. A forged identity provider signature produces valid authentication into every system that federates with it.

The architecture therefore concentrates cryptographic dependency at exactly the points where compromise is most consequential.

Zero Trust also removes the perimeter that hid internal traffic

A second consequence is easily missed. Perimeter architectures leave most east-west traffic unencrypted, on the assumption that the internal network is trusted. Zero Trust encrypts it.

That is a substantial security improvement and it changes the harvest now, decrypt later position. Internal traffic that was previously unencrypted was exposed to anyone inside the network; internal traffic that is now encrypted under classical key exchange is exposed to anyone who captures it and waits. For estates carrying long-lived confidential data between services, the encryption should be hybrid rather than classical, or the exposure has been deferred rather than removed.

Where a Zero Trust estate is better placed

The architecture that increases dependency also improves the ability to change it, and this is the useful part of the analysis.

Credentials are short-lived by design. Workload certificates in a service mesh are typically valid for hours, and user tokens for minutes. Since an algorithm change propagates at the speed of credential renewal, a Zero Trust estate converges on a new algorithm within hours rather than years, as set out in how shorter certificate lifetimes enable post-quantum migration.

Issuance is automated and centralised. There is no manual renewal to coordinate, and the issuing authority is a single control point.

Identity is inventoried. Workload identity frameworks maintain a registry of what exists, which addresses the discovery problem that dominates migration duration elsewhere.

Policy is externalised. Cryptographic requirements are expressed as policy rather than compiled into applications, which is a large part of what crypto agility means in practice.

An organisation that has invested in Zero Trust has, without intending to, built most of the capability post-quantum migration requires.

Where it is harder

The mesh case deserves particular attention. A perimeter estate establishes relatively few TLS connections; a mesh establishes them continuously between every pair of communicating services. The per-handshake cost of post-quantum key exchange is modest, but multiplied across mesh traffic it becomes a capacity planning question rather than a rounding error, as covered in what post-quantum algorithms cost in size and latency.

A practical sequence for Zero Trust estates

Start with the identity provider. Its signing keys are trusted by everything downstream, they are long-lived, and forgery of them produces the widest possible compromise. Establish the vendor's post-quantum roadmap and the key rotation position.

Enable hybrid key exchange in the mesh and at ingress. This is generally a library and configuration change, and it addresses the east-west harvest now, decrypt later exposure that Zero Trust encryption created.

Size the issuing infrastructure for post-quantum. Certificate authority and HSM capacity should be modelled against mesh issuance volume with post-quantum key sizes rather than current load.

Test policy enforcement points with larger tokens and certificates. Signed assertions grow, and enforcement points frequently have message size assumptions that were never documented.

Treat device attestation as a refresh-cycle item. Current TPMs generally cannot provide post-quantum attestation, so this is specified into the next hardware procurement rather than remediated in place.

How Unsung helps

Unsung is a UK-based, vendor-neutral consultancy specialising exclusively in public key infrastructure and cryptographic systems, working across central government, defence, healthcare, financial services, nuclear and transport.

We assess the cryptographic dependencies underlying Zero Trust deployments, including workload identity, federation and attestation, through our PKI health check, and design issuing infrastructure sized for the combination of very short credential lifetimes and post-quantum key sizes through our PKI design and build practice. Where issuance volume is the constraint, our certificate lifecycle management work covers the automation and the hardware sizing together.

For the underlying role of PKI in these architectures, see the role of PKI in Zero Trust security strategies.