Agile, sovereign, edge-ready: rewiring defence IT for a contested decade

Brought to you in partnership with HPE

Europe’s land forces are operating under the most demanding security conditions seen since the end of the Cold War. The conflict in Ukraine has compressed into a few years a doctrinal shift that defence planners had been preparing for over decades: the assumption that command posts will stay connected, that data will stay in well-defined sanctuaries, and that “edge” is a buzzword from the commercial cloud has been comprehensively broken. 

Forces must now plan for full disconnection as a baseline, treat sovereignty as a condition of engagement rather than a procurement preference, and push useful compute as close to the soldier as physics allows.

For HPE, that is not a marketing inflection. It is an engineering one, and it is reshaping four conversations the company is having with government, primes, and the wider defence industrial base across Europe.

From the rear echelon to the line of contact

The first conversation is geographic. Sensor volumes from drones, signals intelligence and ISR feeds have long passed the point where rear-echelon processing can keep pace, and the bandwidth required to backhaul that data simply does not exist in contested electromagnetic environments. The architectural answer is not more pipe, but more compute moved forward, into command vehicles, forward operating bases, even individual platforms, with the thermal, vibration and security envelope that field deployment demands.

What that requires of an IT supplier is a portfolio that does not assume an air-conditioned room or a permissive radio environment. Open, standards-based ruggedised servers and hardened tactical infrastructure are now baseline expectations, not optional extras. They are also what makes the difference between an intelligence cycle that closes inside the tactical window and one that arrives too late to matter.

Capability to know: HPE ProLiant Compute EL, the ruggedised edge server line designed for AI-enabled processing in deployable and hardened environments.

Sovereignty as an operating model, not a label

The second conversation is jurisdictional. “Sovereign cloud” risks becoming the most overused phrase in European defence procurement; it is also, when defined precisely, the single most consequential. For a government ministry, it means knowing, and being able to prove, that classified workloads run on infrastructure assembled, operated and supported by cleared nationals, under domestic law, with the option to disconnect entirely. 

For a prime contractor delivering a service to its armed-forces customer, it means being able to operate in fully air-gapped mode without losing the agility benefits the commercial cloud world has spent fifteen years industrialising.

What ties those two patterns together is a single hybrid control plane that can govern a classified national tenant, an allied coalition tenant, and a deployable tactical tenant from the same operating model, and that can run that model whether the site has a fibre link to the rear or none at all.

Capability to know: HPE GreenLake for Private Cloud, including disconnected and air-gapped configurations governed by a single hybrid control plane.

Networking: the contested baseline

The third conversation is about the network itself, and it is arguably the most underestimated of the four. The classical defence network, perimeter-defended, statically segmented, designed for predictable traffic, is being replaced by a fabric that has to assume contested links, hostile reconnaissance and adversaries with state-scale compute. 

This changes the cryptographic baseline: post-quantum protection is no longer a research topic but a near-term requirement, and it has to be embedded where it actually matters, at the silicon root of trust on every server, and inside high-performance routing on the backbone, without forcing a throughput compromise that operational users will not accept.

How seriously the threat is taken can be read in operational behaviour at the events where defence buyers gather. During Eurosatory 2026, HPE will deploy its own private 5G bubble on-site, so that confidential conversations with defence organisations can take place inside an infrastructure the company controls end-to-end. The same architecture, self-deployable, cryptographically hardened private 5G, is what allows a tactical command post to stand up its own trusted radio environment within minutes of arrival.

Capability to know: HPE Networking, combining Juniper and Aruba portfolios with self-deployable private 5G and post-quantum cryptography embedded in high-performance routing.

Trusted AI: sovereign by construction

The fourth conversation is the one reshaping the most procurement roadmaps. AI is now an operational supply chain in its own right, with dependencies on energy, infrastructure, components and skills that are themselves becoming targets. The question for defence buyers is no longer whether to deploy AI, but under whose jurisdiction, with what provenance, and protected at which states of the data life cycle.

For sensitive workloads, that means extending the sovereignty guarantees applied to the cloud layer all the way down to the workload itself. Data has long been encrypted in motion and at rest; the missing piece, now moving into production, is keeping it encrypted while it is actively being processed, so that even a privileged operator on the host cannot read what is being computed. 

Combined with a tactical edge able to run inference locally and a network designed to resist quantum-era attack, that closes the loop: a chain of custody for defence data that holds from the sensor in the field to the model in the sovereign cloud, and back to the operator.

Capability to know: HPE confidential computing, protecting data in use through hardware-rooted encryption in compute.