TL;DR:

  • Defence is one of the most sustained XR investment sectors — live training is expensive, dangerous, and logistically constrained; simulation and synthetic training environments address all three problems
  • The US Army’s IVAS programme (Microsoft HoloLens-based) represents the highest-profile military AR deployment; UK MoD, NATO allies, and dozens of national militaries have parallel programmes
  • The use cases that have proven out are training simulation, mission rehearsal, maintenance guidance, and synthetic environment generation — battlefield AR as direct combat overlay is still in development

Defence was among the earliest enterprise adopters of VR simulation and is now one of the most active markets for AR and spatial computing. The economics are clear: live military training is expensive, sometimes dangerous, and constrained by geography, weather, and the availability of actual equipment. A synthetic training environment that can replicate any scenario — from urban combat to carrier deck operations — at a fraction of the cost, with no risk of equipment damage or injury, is a compelling proposition.

What makes defence XR interesting in 2026 is that the technology has matured enough that some of these programmes have moved from proof-of-concept to large-scale deployment. The gaps between the marketing and the operational reality are also becoming visible.

Training Simulation: The Proven Use Case

Military simulation has existed since the 1970s in the form of flight simulators and tank gunnery trainers. Modern XR extends the concept into infantry training, joint force exercises, and combined-arms scenarios that were previously only possible at live training ranges.

Synthetic training environments (STEs) are the current major investment area. These are high-fidelity virtual representations of real or fictional operating environments where soldiers can train individual skills, small unit tactics, and large-scale exercises. The US Army’s STE initiative, anchored by PEO STRI, aims to connect live, virtual, and constructive (LVC) training modes so that physically present forces and simulated entities can operate in the same exercise.

The training benefits are well-documented. Studies from multiple military training commands show that VR-trained personnel achieve similar or better performance on specific tasks compared to live training — at significantly lower cost. Marksmanship, vehicle operator training, medical evacuation procedures, and close-air support coordination have all been successfully trained in synthetic environments.

Dismounted soldier training is where headset-based XR has the clearest near-term application. Companies like Bohemia Interactive Simulations (BISim), CAE, and L3Harris supply military-grade training systems that use VR headsets to place soldiers in realistic patrol environments, building clearance scenarios, and checkpoint operations. The headsets handle tracking and rendering; the software handles scenario fidelity, after-action review, and instructor control.

The UK’s Collective Training Transformation Programme and the Australian Army’s Project Overlander are both pursuing similar approaches: replacing some portion of live range time with synthetic training that can be conducted in garrison, reducing the cost and logistics overhead of moving units to live training areas.

IVAS and Battlefield AR: The Complicated Story

The most high-profile military XR programme is the US Army’s Integrated Visual Augmentation System (IVAS), built on Microsoft HoloLens hardware. The programme has been in development for several years and has been the subject of significant scrutiny — both for its technical challenges and its contract economics.

IVAS’s concept is ambitious: a head-worn display that gives dismounted soldiers a persistent heads-up display showing position, compass bearing, teammates’ locations, map overlays, and sensor feeds — reducing the need to look down at a phone or tablet in a combat environment. The system also includes a thermal weapon sight that can feed its image to the headset, allowing soldiers to see around corners or over cover without exposing themselves.

The operational testing reports have been candid about the challenges. Early versions produced nausea and visual discomfort after extended use. The form factor added weight and complexity to an already heavily laden dismounted soldier. The software required in the field was often different from what had been tested in controlled environments. The programme was restructured in 2024 to address these issues with hardware revisions.

By 2026, the picture is more mixed than the early marketing suggested. IVAS has demonstrated genuine capability in controlled environments. The challenge is the gap between “works well enough in a demonstration” and “works well enough that soldiers trust it under stress.” Military equipment must be rugged, reliable, and intuitive enough to use under adrenaline. Getting XR hardware to that standard has proven harder than the technology’s enterprise track record suggested.

Other NATO militaries are watching IVAS closely. The UK MoD, Bundeswehr, and French DGA each have parallel programmes for dismounted AR, but most are deliberately later in the cycle — learning from IVAS’s experience before committing to large-scale procurement.

Mission Rehearsal and Planning

One of the most operationally mature XR applications in defence is mission rehearsal — using 3D spatial environments to walk through planned operations before executing them.

Traditional mission planning uses 2D maps, briefing slides, and sand tables (physical 3D terrain models). XR replaces the sand table with a photorealistic 3D environment generated from satellite imagery, lidar terrain data, and building intelligence. Commanders and squad leaders can literally walk through the planned route, identify chokepoints, preview extraction routes, and rehearse decision points — all before setting foot in the actual environment.

Companies like Vrex (formerly vrgineers), Bohemia Interactive, and Presagis supply mission rehearsal systems to NATO militaries. The US Special Operations Command (SOCOM) has been a particularly active adopter — high-value target raids and hostage rescue operations benefit significantly from the ability to rehearse building layouts in photorealistic 3D.

The intelligence fusion aspect is increasingly compelling: when photogrammetry-generated models of real buildings can be overlaid with SIGINT, imagery intelligence, and ground sensor data, the mission rehearsal environment becomes an intelligence product as much as a training tool.

Maintenance, Repair, and Technical Training

Away from the operational front end, AR-guided maintenance is one of the clearest success stories across both military and defence industrial applications.

Military equipment is highly complex, maintenance-intensive, and the subject of extensive technical manual requirements. An AR system that overlays step-by-step maintenance instructions on actual equipment — showing exactly which fastener to remove, in what order, with torque specifications — reduces training time, error rates, and the need to consult paper technical manuals while working on an aircraft or armoured vehicle.

Boeing, BAE Systems, Raytheon, and Lockheed Martin all have AR-assisted maintenance programmes. The US Navy has deployed AR guidance for shipboard maintenance. The British Army has piloted AR maintenance support for the Challenger 2 fleet. The ROI calculations are consistent: AR-assisted maintenance reduces task completion time by 30–40% and error rates significantly for complex multi-step procedures.

The advantage in defence is particularly pronounced because military equipment has long lifecycles (decades), which means the documentation complexity compounds over time. AR systems that can surface the right version of the right technical procedure for a specific configuration of a specific vehicle are genuinely difficult to replicate with paper or static digital documentation.

Synthetic Environment Generation: Where AI and XR Converge

One of the faster-moving areas is AI-assisted synthetic environment generation — using generative AI and automated photogrammetry pipelines to create high-fidelity 3D environments for training and planning at scale.

Previously, creating a realistic 3D representation of a specific operating environment was a time-consuming manual process. Terrain artists and technical content teams had to build environments from imagery and reference data. The bottleneck limited how many environments could be created and how frequently they could be updated.

Newer pipelines combine satellite and aerial imagery, lidar elevation data, and neural rendering techniques (including Gaussian splatting) to generate 3D environments automatically, with human review and refinement applied where needed. This changes the equation: instead of a fixed library of training environments, a command can generate a photorealistic synthetic environment from any location in the world within hours.

NVIDIA’s Omniverse platform and Epic Games’ Unreal Engine (through its defence-focused partnerships) are both competing in this space. The UK’s Defence and Security Accelerator (DASA) has funded several programmes exploring AI-assisted environment generation for defence training systems.

The Security and Logistics Reality

Defence XR programmes face constraints that commercial XR deployments don’t. Security classifications mean that mission rehearsal systems handling sensitive intelligence data must operate in air-gapped environments. Headsets used in classified environments can’t be standard commercial products — they require hardware modifications to prevent data leakage and meet TEMPEST and emissions control requirements.

The supply chain implications are significant. A commercial AR headset manufactured by a company with supply chain exposure to adversary nations may be unsuitable for classified military applications regardless of its technical capability. This creates a bifurcation between commercial XR hardware used for training in unclassified environments and hardened, modified, or custom-built systems for operational use.

The UK’s CADET (Cyber and Digital Environment Technology) programme and US DoD Zero Trust Architecture initiatives are both grappling with how to integrate XR hardware into defence IT environments in ways that don’t introduce new security vulnerabilities.

Where This Heads

The near-term trajectory is clearer for training and maintenance than for operational AR. Synthetic training environments will continue displacing live range time for scenarios where simulation fidelity is sufficient. Maintenance AR will expand as the ROI case becomes more established and the hardware matures.

Operational AR — the IVAS concept of a heads-up display on the battlefield — will continue development, but the timeline to widespread deployment is longer than early programme projections suggested. The human factors challenge is non-trivial: military personnel operating under stress need hardware that works reliably at the extremes of their operational envelope, which is a significantly higher bar than enterprise users navigating an office.

Defence is, however, one of the few sectors with the procurement budgets to push XR hardware requirements beyond what the commercial market has demanded. The requirements that emerge from military programmes — ruggedness, battery life, display performance in bright sunlight, radio frequency compatibility — tend to pull the whole ecosystem forward. Many advances in consumer XR hardware over the past decade trace their origin to defence and aerospace research programmes.

The relationship runs in both directions: commercial XR advances (higher-resolution displays, more efficient rendering, better inside-out tracking) reduce the cost of building capable military systems. In 2026, both directions are active.