Omnia turns Army training into a digital factory

A Raytheon UK-led consortium has secured a £2 billion contract to rebuild the British Army’s collective training system around live exercises, virtual environments, artificial intelligence, and advanced analytics.

Running for 15 years and intended to train around 60,000 soldiers annually, the programme brings together Raytheon UK, Capita, Cervus, Rheinmetall UK, and Skyral, supported by 44 companies across the wider supply chain.

The contract is expected to support around 400 jobs, including 270 newly created skilled positions, alongside 100 apprenticeships. Those figures reflect the programme’s technical breadth, since Omnia combines training delivery with long-term software development, hardware integration, data engineering, communications, modelling, and support.

Collective training covers formations working together rather than individuals learning isolated tasks. Exercises must represent headquarters, communications networks, vehicles, artillery, air defence, uncrewed systems, logistics, intelligence, electronic warfare, and opposing forces, while producing enough reliable data to assess performance.

By connecting activity conducted on physical ranges with virtual and constructive elements, the programme will allow soldiers operating real equipment to train alongside simulated units. Headquarters will respond to a combined operational picture generated from live positions, computer-controlled forces, and virtual platforms.

Training as a maintained technical product

A synthetic representation of the Army can never remain static. Vehicle characteristics change, software-defined radios receive updates, new drones enter service, threat behaviours evolve, and weapons acquire revised ranges, sensors, or operating procedures.

Maintaining that representation creates a continuous production pipeline for terrain, platform models, artificial behaviours, scenarios, interfaces, and data services. Each element requires design, verification, version control, cybersecurity assessment, and controlled distribution to training locations.

Physical equipment will remain central to delivery. Instrumentation packages, communications gateways, displays, servers, wearables, tracking devices, controllers, and simulation hardware must be manufactured, installed, repaired, and replaced, often in outdoor environments where ruggedisation standards exceed those of conventional computing centres.

An AI battle laboratory will support experimentation, although useful output will depend on the quality, consistency, and accessibility of exercise data. Models trained on incomplete or poorly structured information can produce confident but unreliable assessments, while disconnected datasets may leave instructors with attractive dashboards that offer little operational insight.

Advanced analytics can identify delays in command decisions, communications congestion, repeated logistics failures, or gaps between sensing and engagement. To perform those functions across a mixed supplier base, Omnia will need common data standards and clearly governed interfaces.

Proprietary formats may accelerate individual work packages, but they make later integration and replacement more expensive. Over a 15-year contract, several generations of servers, processors, networking equipment, and software frameworks will pass through the training estate, placing a premium on modularity and open architecture.

Connecting training to equipment design

Exercise data can also inform the design and support of operational equipment. Repeated problems with interfaces, maintenance procedures, power demand, operator workload, or communications can be fed back to programme teams before faults become embedded across a large fleet.

The Army’s rapid expansion of its drone fleet illustrates the value of a shorter feedback loop between users, manufacturers, software developers, trainers, and procurement teams. Systems arriving in units at speed need realistic training, while factories need structured evidence about reliability and usability.

Keeping development activity separate from major exercises will require disciplined release management. Soldiers need stable systems during training events, while engineers need access to representative environments where updates can be tested without disrupting delivery.

Cybersecurity will extend across the programme because training systems contain realistic information about force structures, tactics, communications, and equipment performance. Networks must support distributed users and suppliers without creating unnecessary routes into sensitive military infrastructure.

Commercial technology cycles will add continuing pressure. Servers, graphics processors, headsets, networking products, and software libraries may become obsolete within a fraction of the contract period, requiring the Army and consortium to replace components without rebuilding the complete architecture.

The breadth of Omnia’s industrial team offers access to specialist capability, although it also increases interface risk. Technical authority, configuration ownership, acceptance criteria, and security responsibilities must remain clear if the programme is to operate as one integrated environment rather than a collection of connected products.

Smaller suppliers could find opportunities in modelling, sensing, analytics, specialist hardware, and scenario generation, provided technical standards and procurement routes allow them to integrate without excessive dependence on bespoke prime-contractor interfaces.

Unlike a conventional factory, Omnia’s production line will generate models, software releases, data services, exercise scenarios, and readiness rather than physical platforms. It will nevertheless face familiar industrial pressures around standards, quality, skills, obsolescence, supply-chain continuity, and throughput.