
Firmware updates rarely make headlines, and the expiry of a certificate buried in a PC's boot firmware sounds like the kind of detail that belongs in a vendor's release notes. It is not. Secure Boot depends on a chain of certificates held in firmware, and some of the certificate authorities anchoring that chain across a large installed base of PCs reach their expiry dates in 2026. Replacing them is not a single patch. It is a rotation of trust anchors across an estate few organisations can describe with confidence, on hardware that may not still receive updates.
For the Global Tech Awards, which recognises engineering across the technology sector, the subject sits where firmware, public-key infrastructure, hardware supply chains and IT operations intersect. The cryptographic work is well understood. The difficulty is inventory, sequencing and the long tail of devices whose manufacturers have moved on.
Secure Boot is a chain of trust that begins before the operating system loads. Firmware holds a set of databases: a platform key, a key exchange key, a signature database of trusted signing certificates and a revocation list. When a boot component such as a bootloader or a signed driver loads, the firmware checks its signature against those stored entries. If the signature validates against a trusted certificate, the component runs; if it does not, the machine refuses to load it.
What expires in 2026 is not the device's own identity but certificate authority certificates used to sign the components that firmware trusts. An expired certificate does not, by itself, stop an existing machine from booting; firmware does not usually re-check the validity period of a certificate it already trusts at every power-on. The effect is narrower and slower. New signed boot components issued after the expiry can no longer be validated against the old chain, so updates to bootloaders, recovery tools and security fixes that depend on that chain stop installing cleanly. The failure is not a cliff edge on a single date but a gradual loss of the ability to update the boot path, which becomes serious when a security fix is needed.
Most organisations do not hold a reliable record of which firmware version and which certificate entries each device carries. Asset registers typically record make, model, operating system and perhaps a warranty date. They rarely record firmware revision, and almost never the contents of a Secure Boot database. The gap widens across the estate: laptops, servers, virtual machine images cloned from old templates, point-of-sale terminals, kiosks and lab equipment running embedded operating systems.
Virtual machines deserve particular attention. A guest image inherits a firmware model from its host or template, and a template built years ago may carry an old trust store replicated every time a new instance is created. The result is a fleet that looks uniform in an inventory system and is anything but uniform underneath.
OEM update availability compounds the problem. Firmware updates are released per model, and often per regional variant, on the manufacturer's own schedule. A model still on sale may receive the new certificates promptly; a model withdrawn two years ago may receive them late, or not at all. Organisations discover the shape of their exposure only by asking each supplier, model by model, what is planned and when.
Rotating a trust anchor is a firmware flash, and a firmware flash can leave a device unusable if it is interrupted. Power loss, a drained battery, a forced reboot or a failed verification step during the write can leave a machine unable to complete its boot sequence. The risk is manageable, but only if it is planned for rather than discovered during a fleet-wide push.
Several operational constraints follow. The ordering of key updates matters, because applying the steps out of sequence can leave a device trusting neither the old chain nor the new one. Some updates require physical presence at the machine, or console access for servers, which turns a remote patch campaign into a scheduling exercise across offices, data centres and home workers. Recovery keys need to be escrowed and retrievable before the work begins, because a firmware change can trigger a recovery prompt that locks out a user without the key.
Rollback planning belongs in the same conversation. A firmware update that fails on a small percentage of a fleet is an inconvenience; the same failure without a documented recovery path is an outage. Pilot groups, staged rollouts and a tested recovery procedure are the ordinary controls, applied to firmware with the discipline usually reserved for operating-system changes.
The devices that will never receive new firmware decide how well this transition goes. Industrial controllers, medical instruments, retail terminals and refurbished hardware sold into secondary markets often run firmware that the manufacturer stopped updating years ago. Some sit on networks that cannot easily be reached; others run software that cannot be revalidated against a new chain without recertification the vendor will not fund.
Where a trust anchor cannot be rotated, the compensating controls are familiar but imperfect. Network segmentation limits what an unpatched device can reach. Allow-listing and monitoring at the network edge reduce the chance that a compromised device goes unnoticed. Some organisations will isolate such equipment behind a gateway and treat it as a fixed-function appliance rather than a general-purpose computer. Others will replace it, and that cost belongs in the planning conversation from the start: it is a capital expense disguised as a security patch.
The durable lesson is that certificate lifetime is a product-design variable. A trust anchor that cannot be replaced in the field turns an ordinary expiry date into a hardware replacement programme. Designers can avoid that by making trust stores updatable through a supported, recoverable process, and by ensuring that a device can hold both an old and a new anchor during a transition.
Procurement has an equal part to play. Contracts should state how long firmware updates will be supplied, and for how long after the last unit ships. They should ask whether trust anchors can be rotated, whether the vendor publishes a schedule for certificate changes and whether recovery procedures are documented. These are ordinary questions, rarely asked at the point of purchase, when they are cheapest to resolve.
Certificate validity periods should also be set with device lifetimes in mind. A trust anchor designed to last a decade will expire inside the service life of equipment expected to run for fifteen years. Shorter validity periods, paired with a credible rotation process, are the more honest arrangement.
The 2026 expiry is being read in many organisations as a vendor note to be filed. It is better understood as a rehearsal. The same pattern will recur with every certificate, key and trust store embedded in hardware that outlives its supplier's attention. The organisations that treat trust anchors as replaceable components, and firmware support as a term of the purchase, will find the next deadline considerably less disruptive than this one.