International Security Journal hears exclusively from Simon Pamplin, Chief Technology Officer of Certes about evolving risks in 2026.
Something shifted in 2026.
Not a single event, not a headline breach, not a government announcement.
A convergence.
The harvest-now, decrypt-later campaigns that security teams have been warning about for years are no longer theoretical positions in a risk register, they are maturing operations, running quietly in the background of geopolitical and commercial competition, accumulating encrypted data at scale.
Meanwhile, regulatory timelines are tightening, the attack surface has expanded dramatically through supply chains, edge infrastructure and AI pipelines, and most organisations are still, at best, mid-plan.
The threat did not announce itself, it accumulated and 2026 is the year that gap between where most organisations are and where they need to be becomes genuinely dangerous.
The attack that already happened
There is a persistent and dangerous assumption in the way organisations frame quantum risk: that it is a future problem, contingent on the arrival of a cryptographically relevant quantum machine.
Google’s revised Q-Day estimate of 2029 reinforced that framing for some, positioning it as a deadline to plan toward.
However, that framing misses the most urgent dimension of the threat.
Adversaries running harvest-now, decrypt-later campaigns are not waiting for 2029 – they are operating today, systematically exfiltrating encrypted data from networks, third-party systems, cloud environments and edge infrastructure, storing it and banking on the certainty that quantum capability will eventually give them the means to unlock it.
The attack, in other words, has already happened and what remains is the decryption.
The clock started the moment the first packet was intercepted and stored.
Q-Day does not create the risk, it realises it.
This matters most for long-life data such as: financial records, health information, identity data, intellectual property and legal communications.
These are not assets that lose value quickly as Social Security numbers do not expire, genomic data does not become less sensitive over time and commercial IP harvested from a competitor today may still be strategically valuable in five years.
Organisations that think in terms of current breach impact are measuring the wrong thing.
The question is not what an attacker can do with this data now. It is what they will be able to do with it later.
With Gartner predicting that a cryptographically relevant quantum computer could arrive by 2029, the migration timelines that organisations once treated as comfortable have collapsed.
Post-quantum transition is a multi-year programme for most.
The arithmetic is not forgiving.
The surface expanded while organisations were planning
Even organisations that took quantum risk seriously two or three years ago face a compounding problem: the attack surface they were planning to protect has fundamentally changed shape beneath them.
Supply chain exposure is perhaps the most structurally underestimated risk.
A single managed service provider, software vendor or cloud platform can provide indirect access to hundreds of downstream organisations.
From an attacker’s perspective, it is a concentration of value with diffusion of responsibility.
Data flows constantly between organisations, platforms and jurisdictions, but accountability has not shifted.
Regulators are explicit: responsibility remains with the organisation that owns the data, regardless of where it is processed.
That creates a dangerous tension as the data owner retains the liability.
However, the data itself travels through infrastructure they do not control, protected by cryptographic assumptions that are already under pressure.
The edge compounds this further.
The rapid expansion of 5G, IoT and operational technology has pushed sensitive data to the furthest and hardest-to-defend points of any organisation’s infrastructure.
Many edge devices run weak or bespoke cryptography, cannot support heavyweight post-quantum algorithms natively and are genuinely difficult to patch at scale and adversaries have noticed.
The edge has become the easiest place to harvest data precisely because it sits outside the coverage of most traditional security programmes.
AI pipelines represent the newest and least-understood exposure.
Sensitive data now flows into training sets, inference pipelines and third-party analytics platforms that organisations do not fully control, often without consistent cryptographic protection applied at the data level.
Once data is exposed in cleartext to an AI model or external service, the organisation loses meaningful control over how it is stored, retained or leaked.
The attack surface for harvest-now, decrypt-later is no longer confined to the network, it extends into every system that touches sensitive data in motion.
The perimeter that most PQC roadmaps were designed to protect no longer exists in the shape it once did.
The organisations that recognise this now are the ones with room to respond.
Legacy makes this harder. It does not make it optional.
The structural challenge of quantum migration deserves honest acknowledgement.
Many of the systems that matter most to large organisations, core banking platforms, clinical record systems, trading infrastructure and operational technology, were built decades ago.
Cryptography is embedded deep within application code, firmware and proprietary vendor software.
Changing it can involve rewriting applications, recertifying devices or replacing hardware entirely. This is genuinely expensive, slow and operationally risky.
The complexity of legacy environments is a reason to start earlier and think differently, not a reason to defer.
The organisations waiting for a clean architectural moment to begin migration will find that moment does not arrive before the window closes.
The practical answer lies in separating the protection of data in motion from the modernisation of the systems that carry it.
Rather than attempting to update cryptography inside every legacy application or device, organisations can apply quantum-safe protection at the data layer itself, wrapping flows independently of application logic or infrastructure.
Legacy platforms continue operating.
Security is modernised around them. The blast radius of any cryptographic failure is contained by segmenting flows, so that a single weakness does not expose everything at once.
This is not about avoiding the harder work of long-term migration. It is about ensuring that data already in motion is not left exposed while that work proceeds.
What 2026 demands: Data-centric, crypto-agile, sovereign
Three principles define what effective post-quantum protection looks like in 2026, and they are inseparable.
Data-centric protection means security that travels with the data itself, not the infrastructure carrying it.
Perimeter controls, VPNs, network-layer defences: all of these assume a defined boundary that no longer holds.
Data moves across cloud environments, legacy systems, edge devices and AI pipelines constantly.
Protection that stops at the infrastructure boundary provides no guarantee once that boundary is crossed or compromised.
The only durable model is one where the data itself remains cryptographically useless to any unauthorised party, regardless of where it travels or what path it takes.
Crypto agility is not a feature, it is a business requirement.
No algorithm lasts forever, and the post-quantum landscape will continue to evolve as NIST standards develop and some approaches are eventually superseded.
The organisations that will navigate this well are not the ones that pick the right algorithm today, they are the ones that build the ability to swap, rotate and adapt cryptographic controls without redesigning systems or interrupting operations.
Frequent key rotation, decentralised key ownership and algorithmic flexibility all serve this goal, and they serve it across every environment, from server workloads to edge devices.
Sovereign key ownership is the mechanism by which an organisation retains genuine control over its own data.
In multi-cloud environments, across third-party processors and into AI platforms, there are many points at which control can quietly slip away.
If the keys used to protect sensitive data are held or accessible by a provider, a platform or a third party, the sovereignty claim is incomplete.
Only organisations whose encryption keys are entirely their own, never visible to infrastructure providers, can genuinely assert that their data remains under their control wherever it travels.
As regulatory scrutiny of data residency and jurisdictional control intensifies, this distinction is becoming a compliance reality, not just a security preference.
These three principles are not separate workstreams, they are one coherent posture that can be deployed across legacy environments, hybrid cloud and edge without breaking what already exists.
The opportunity in the urgency
There is a version of this conversation that is purely about:
- Threat
- What adversaries are doing
- How fast timelines are closing
- The weight of regulatory expectation
That version is accurate, but it is not the complete picture.
The security and technology leaders who move decisively on post-quantum readiness in 2026 are not reacting to pressure.
They are positioning.
PQC readiness is increasingly a signal to regulators, to boards, to customers and to partners.
It demonstrates that an organisation understands not just current risk but long-term structural exposure.
It shifts the posture from best-endeavours compliance to demonstrable, auditable protection, whilst building the cryptographic resilience that will matter as quantum capability matures, whether Q-Day arrives in 2029 or later.
The data being created, transmitted and harvested today is the data that will define exposure in five or ten years.
The organisations that act now will not be the ones that were forced to, they will be the ones that saw the landscape clearly when it still offered room to move.
Quantum is not coming, for the data already harvested and waiting, it is already on its way.
The question is whether the data it unlocks will be yours.
