After twenty-four years running live, constructive and virtual training; Military Synthetic Training in the Norwegian Army — the technology works. What consistently fails is the systems thinking that connects it.
Military Synthetic training encompasses the full spectrum of technology-supported training: Live — real forces and equipment operating in the field, with the outcomes of battle simulated through instrumentation systems; Virtual — crews and units operating in computer-generated environments using simulation software; and Constructive — commanders and staffs working with fully simulated forces in command-and-control training systems. When integrated coherently, these three domains form a synthetic training ecosystem that should allow an army to train any unit, at any echelon, at any time. The challenge — and the subject of this article — is that coherent integration remains the exception rather than the rule.
There is a moment that everyone involved in military synthetic training recognises. At the crew level, that ecosystem largely works. The trainer works. The soldiers are engaged, the procedural fidelity is good enough, and the instructors can see what is happening and intervene when it matters. Army leadership is satisfied, and a new training capability is declared.
Then someone asks: Can we do this at the company level?
And that is where it gets complicated.
I spent twenty-four years at Camp Rena — first as Chief Instructor Mechanised Infantry Battalions at the School of Armour and Cavalry, then across fourteen years at the Norwegian Combat Training Centre (NORCTC) as Executive Officer and an Exercise Manager, and later Commanding Officer, and finally eight years as Head of the Norwegian Army Training and Simulation Centre (NATSC). That arc took me from teaching mechanised infantry tactics, through live instrumented exercises with thousands of troops in the field, to designing the Norwegian Army’s LVC training architecture and driving the procurement programmes that underpin it. Throughout this period, I observed a persistent gap between what military synthetic training promises and what it actually delivers when scaled up. That gap is real, it is structural, and the industry does not talk about it honestly enough.
Building Military Synthetic Training Capability
Before making the argument about what is hard, it is worth being honest about what works.
The Norwegian Army has developed a meaningful military synthetic training ecosystem over the past decade. It spans the full LVC spectrum, and at each level, there are capable, well-used systems.
At the live end, NORCTC remains the foundation — instrumented force-on-force training using the SAAB GAMER system at Rødsmoen or elsewhere in Norway, the single most important tool the Norwegian Army has for developing soldiers, leaders, and commanders in realistic combat conditions. That capability is not standing still. P5066 — the Future Army Combat Training System (FACTS) — is a publicly announced programme currently in its pre-procurement phase and heading to the Parliament (Stortinget) in the autumn of 2026. With a government-disclosed cost range of between one and five billion NOK, it represents the most significant investment in the Norwegian Army training capability in a generation. Crucially, LVC integration and interoperability is a core requirement of the future system — and rightly so. Realistic training of artillery location radars, air defence batteries, loitering ammunition, drone operators, and forward observers adjusting fire against virtual impacts cannot be achieved purely through live instrumentation. A virtual integration layer is not a nice-to-have for FACTS. For much of what the future force needs to train, it is the only technically viable solution.
At the virtual level, the military synthetic training ecosystem is broader than most outside observers realise. The KDA Leopard 2A4 gunnery trainer — a desktop solution with a real gunner’s handle — has been in service long enough to fall outside vendor support and is now run and maintained independently by NATSC. That is both a testament to its continued utility and a quiet warning about the lifecycle costs of platform-specific solutions. Remote Weapon Station gunnery trainers, each with ten student stations and an instructor station, are deployed at four locations. Javelin indoor trainers are available at most Army camps.
The JFATS — Joint Terminal Attack Controller and Forward Observer Advanced Training System, delivered by Fidelity Technologies Corporation and operated at Rena by OBSIMA on behalf of Fidelity — is the centrepiece of the Norwegian Air Ground Operations School’s JTAC and fire support officer training. The main installation at Rena features a seven-metre immersive dome, a pilot-in-the-loop station, and a four-person TACP team environment. A smaller dome operates at Setermoen, and two mobile satellite systems support training at other locations. JFATS was purposefully designed to be HLA-compliant — meaning it was built, in principle, to federate with other military synthetic training systems. The significance of that design decision will become clear shortly.
The backbone of collective virtual training is a network of Virtual Battlespace (VBS) classroom sets distributed across six sites: three at Rena, two at Setermoen, and single sets at Skjold, Porsangmoen, Høybuktmoen, and — supporting Norwegian forces in the enhanced Forward Presence — in Lithuania. Each set is a NATSC asset, technically managed centrally from Rena, with locally based instructors trained by NATSC leading training in their respective units. The platform configurations reflect the force structure: CV90, Leopard 2A4, fire support, observation posts, and multi-role variants.
That network is currently undergoing an upgrade to VBS4. This is not a simple platform refresh but a strategic restructuring of the Norwegian Army’s relationship with training technology. The procurement — which I initiated on behalf of the Land Warfare Centre, drove through successive levels of decision-making, and coordinated as both user representative and as the interim Army G-10 Total Project Coordinator, with the contract signed in December 2025 — transitioned from 250 standalone licences with no support backbone to a 500-licence enterprise agreement including Builder, Blue IG, TerraTools Premium, a four-year support plan, and a hardware refresh across the classroom network. The upgrade has been publicly covered in Teknisk Ukeblad (Weekly Technical Digest) and on NRK (Norwegian Broadcasting Corporation).
At the constructive end, the Command Staff Trainer uses JCATS — Joint Conflict and Tactical Simulation — connected to the Norwegian Army’s C2 information systems, NORCCIS and NORBMS, which automatically populate blue-force tracks. It is used to train commanders and their staff in battlefield leadership and military decision-making. And it is already federated with the virtual layer in ways that point toward what a full military synthetic training architecture could look like.
This is the right direction of travel, which makes the structural problems that remain all the more frustrating — because they are not technology problems.
Military Synthetic Training That Actually Works
Before describing where the system falls short, it is worth describing two examples of where military synthetic training integration works — because they illustrate both the potential and the design thinking required.
The first is the JCATS/VBS federation in support of the Norwegian Army’s intelligence officer course. This grew from an initiative by the engineering team at NATSC during my time as Head — they saw the opportunity, tested it on a course, found it worked, and it has since been developed further.
ISTAR units man observation posts in VBS, positioned in a virtual terrain that mirrors the tactical scenario being played out simultaneously in JCATS. What those virtual observers see depends entirely on where they are positioned, which sensor they are using, and whether they are looking in the right place at the right time — exactly as in real operations. When they observe something, they report it through their command post, which aggregates the incoming reports and builds a recognised intelligence picture, which it then reports to the S2 students who are the primary training audience for the course.
Those students — the intelligence officers — must build their own recognised intelligence picture from what their command post provides and present it to the course director with recommendations for actions and reactions. The course director plays the role of the commander, pushing back, challenging, and demanding clarity.
The result is a military synthetic training system that simultaneously develops ISTAR operators in observation and reporting discipline, command post staff in aggregating and processing intelligence, and intelligence officers in analysis and presentation — all within a single, coherent, unclassified scenario. The federation runs on unclassified instances of both systems, with fictitious scenarios and order of battle, making it fully accessible without classification constraints.
Figure 1 — The JCATS/VBS Intelligence Federation: a single coherent unclassified scenario training operators, staff, and officers simultaneously across virtual and constructive domains.
The second example comes from earlier in my time at NORCTC. During the early years of UAV introduction into the Norwegian Army, getting actual unmanned platforms airborne during instrumented exercises was genuinely difficult — due to airspace restrictions, platform availability, and the operational immaturity of the capability itself. We integrated UAV virtual training stations based on VBS2 into our live instrumented exercises at Rødsmoen to fill that gap.
The terrain was modelled in VBS2. All troops, vehicles, and weapons instrumented through the SAAB GAMER system were visible to the virtual UAV operators — conditionally. Right area, right time, right sensor mode. Exactly as in real operations. What this gave us was a military synthetic training solution for something that was genuinely hard to train at the time: getting UAV operators to share what they observed with their commanders, and getting commanders to build situational awareness from UAV feeds and act on it. The information asymmetry, the communication discipline, the cognitive load of integrating an aerial picture into a ground tactical decision — all of that was trainable because the integration was realistic enough to create genuine friction.
Both examples share the same characteristic. They did not emerge from procurement requirements. They emerged from practitioners who identified a training problem and solved it with the tools available. That is both encouraging and instructive: some of the best military synthetic training integration in the Norwegian Army exists because people saw the opportunity, not because a programme specified it.
The challenge is that this kind of integration remains the exception. And the structural reasons for that are worth examining.
The HLA Promise and What Vendor Restrictions Actually Deliver
JFATS was designed to be HLA-compliant. That design decision was supposed to enable federation with other simulation systems — VBS, JCATS, live instrumented forces. The military synthetic training value of such a federation would be considerable. Imagine a JTAC embedded with live-instrumented troops at Rødsmoen, coordinating in a real tactical scenario, talking a pilot in a simulator onto a live instrumented target, and receiving simulated weapon effects back into the exercise via the GAMER EXCON solution. The full close air support kill chain — trained end-to-end, in a single coherent synthetic and live environment, without the cost or availability constraints of a real F-35 sortie. Technically, with the systems available in Norway today, that is achievable.
It has not been done in any sustained way because contractual and procurement restrictions on JFATS have prevented the integration that HLA compliance was supposed to enable. NOBLE — the Norwegian Battle Lab & Experimentation at the Joint Headquarters — conducted an experiment integrating JFATS with an unclassified F-35 desktop simulator. It worked. The technical path was proven. What it demonstrated is that the barrier to the kind of joint, multi-system JTAC training environment described above is not technology. It is contractual/programmme requirements that did not mandate integration, and the institutional habit of treating interoperability as someone else’s problem.
This is the same pattern, expressed at the joint fires level, that runs through the rest of the Norwegian Army’s military synthetic training ecosystem.
The Connectivity Problem Nobody Solved
The VBS classroom network not yet operationalised as a distributed network.
The nine classrooms are isolated from each other. Cross-garrison online exercises are not possible. Remote maintenance and updates from the central hub at Rena are not possible. Each classroom is an island. The network exists in name but not in practice.
This is not a technology failure. The solution should run on an unclassified but dedicated military network operated and maintained by CYFOR. The technical path to connecting these classrooms has been clear for years. We have been working toward it for years. It has not happened.
It is a procurement and priority failure. A solvable problem that should have been solved years ago.
The consequences are direct and concrete. A distributed military synthetic training network that cannot connect is not a distributed capability — it is a collection of local capabilities that happen to run the same software. The training ceiling in each garrison is determined by the number of stations in that garrison’s single classroom. A platoon exercise requires a platoon’s worth of stations in the same room. A company exercise across garrisons is not possible in a virtual environment without either transporting people or connecting the classrooms — the former defeats much of the purpose of military synthetic training. The latter is not currently an option.
The ambition of scaling from crew to collective training — the exact transition that synthetic environments are supposed to enable — runs directly into this wall. And it has run into it for years, not because the wall could not be removed, but because removing it was not prioritised.
The vendor delivered the capability. The system works as specified. The gap is in the infrastructure and prioritisation that surrounds it — and the institution, not the vendor, owns that gap. Which means the institution has to be honest about it.
Figure 2 — The Disconnected Network: nine classroom sets across Norway and Lithuania, each an island. The network exists in name but not in practice.
The Right Answer at the Wrong Time — and the Pattern It Reveals
The Norwegian Army recently started test usage of the Combined Combat Simulator at Camp Rena — four CV90 cabins and four Leopard 2A4 cabins, each a full mock-up of the real vehicle, presenting an environment as close to the actual platform as current technology allows. Vehicle commander, gunner, driver — every crew position represented, every interface replicated. In a single configuration, it can train up to half a mechanised battalion, with simulated subunits filling the roles that are not crew-manned. Official handover from vendor KNDS is expected in autumn 2026. It is an impressive piece of engineering.
It is also an illustration of a recurring procurement pattern in defence training: the right answer to a question that has since moved.
What the Norwegian Army needs now — expanding toward three brigades, with two standing brigades and a mobilisation brigade that can only be trained at scale through military synthetic training — is not a single high-fidelity simulator at a single location. What it needs is enough stations to train the force that actually exists, distributed across the locations where that force is based, at the tempo required by operational demand.
Figure 3 — The question is not which system. It is which philosophy.
What I argued for while still in post, and continue to believe, is a larger number of lower-cost, VBS4-based crew trainers that replicate the essential crew interfaces without the full physical mock-up. Not the perfect simulator at one site, but a sufficient number of simulators at every site. A larger number of distributed stations will train more soldiers, more often, and with more operational relevance than a small number of perfect cabins at a single garrison. The Combined Combat Simulator will deliver genuine value — I am not arguing otherwise. But it represents a procurement philosophy and the answer for a problem we had almost 20 years ago, a problem that the Norwegian Army’s current force development trajectory has outgrown.
Now consider what is about to happen at Camp Rena.
This autumn, alongside the Combined Combat Simulator handover, the first Leopard 2A8 gunnery trainers, driver trainers, and VBS4-based crew trainers will begin arriving as part of the ongoing Leopard 2A8 procurement. In 2028 or 2029, a Leopard 2A8 Combat Simulator — the AGPT — running the same backbone as the Combined Combat Simulator, will follow. Two Combat Simulators from the same vendor, running the same system architecture, at the same garrison.
They can technically be connected. But tjey will probably enter service as standalone non-connected systems, because connectivity was not specified as a requirement in the Leopard 2A8 procurement. Which means the Norwegian Army will likely have to initiate a separate procurement action to connect two near-identical systems sitting at the same camp. The architecture for interoperability exists by accident of vendor selection, not by design. And the cost of that oversight — paid in additional procurement effort, additional funding, and additional time — is entirely predictable from where we stand today.
This is the institutional habit worth naming clearly: procuring live, constructive, and virtual training capabilities in isolation and paying for integration later, repeatedly, as if it were a surprise every time. The connectivity gap in the VBS classroom network, the contractual/programme constraints limiting JFATS federation, and the forthcoming gap between the two Combat Simulators are not separate problems. They are the same problem, expressed at different levels of the training system, in different procurement cycles, over and over again.
The answer is not more technology. It is better requirements. Connectivity, interoperability, and the ability to operate across the training architecture as a coherent system — these need to be mandatory requirements in every procurement, not afterthoughts negotiated at additional cost after delivery.
FACTS has the opportunity to break that pattern. LVC integration and interoperability are already stated as core requirements. The question is whether the procurement discipline to enforce those requirements across every subsystem, every vendor, and every delivery phase will hold. That question will be answered over the next several years. The answer matters enormously.
The Military Synthetic Training Question Nobody Is Asking
The defence training industry is good at answering the question that is asked. The question is usually some version of: can your system replicate this platform, at this fidelity, in this environment?
But that is the wrong question.
The right question is: what architecture allows us to scale military synthetic training from the crew to the brigade level without rebuilding from scratch at each echelon? What decisions do we need to make now — about enterprise licensing, about connectivity infrastructure, about the relationship between live, virtual and constructive layers, about where fidelity matters and where distribution matters more — that will determine whether our training investment compounds over time or depreciates?
I have sat on both sides of that question. As an instructor at the School of Armour and Cavalry, I knew what a CV90 crew needed to be able to do before they touched a real vehicle. As Commanding Officer of NORCTC, I ran instrumented exercises at brigade scale and watched where the military synthetic training design broke down. As Head of NATSC, I helped design the organisation, drove the procurements, and lived with the gap between what the network could do and what it should have been able to do years earlier.
By the time the Leopard 2A8 Combat Simulator arrives at Rena in 2028 or 2029, the Combined Combat Simulator will have been in service for two to three years. Whether those two systems can operate together in the same synthetic environment will depend on decisions being made — or not — right now. That is not a hypothetical future problem. It is a present procurement decision with a predictable future consequence.
The Norwegian Army is at an inflexion point in its military synthetic training capability. The investment is real, the direction is largely right, and the technology is genuinely good. What has consistently lagged is the systems thinking that connects the pieces into a coherent, scalable training architecture. That is not a technology problem. It is a requirements problem. And requirements are written by the buyers — not the vendors.
Twenty-four years at Camp Rena gave me that view. It is worth sharing while decisions shaping the next generation of Norwegian Army training are still being made.
