Wargaming's military simulation market faces structural demand test

Sweden's conscript surge to 10,000 annually by 2030 creates a training bottleneck that W5's live-fire simulators must fill, but the industry's AI claims remain unproven.

The military simulation industry is getting a stress test it didn't ask for. On July 13, Quartz Sea Research published a note on W5, the Swedish defence contractor, linking its simulator sales directly to Sweden's conscript expansion. The Swedish Armed Forces plan to scale annual conscript volumes to 10,000 by 2030, supporting an 88,000-strong wartime mobilization force. That's a 43% increase from current levels, according to Swedish Defence Materiel Administration data. The math is simple: more bodies, less time, same budget. Something has to give, and it's not going to be the training requirements.

The structural demand argument

Quartz Sea's thesis is worth taking seriously. They argue that Sweden's sudden conscript influx creates a "severe training bottleneck" that directly drives demand for W5's live-fire ranges and simulators. This isn't a cyclical bump. It's doctrine-driven, multi-year, and tied to a concrete headcount target. The Swedish government's 2024 defence bill allocated 170 billion SEK ($16 billion) to military spending through 2030, with a specific line item for training infrastructure. W5's simulators—which include containerized live-fire ranges and mobile command-post trainers—fit the requirement profile precisely.

But here's the question Quartz Sea doesn't fully answer: Can W5 scale production fast enough? The company's 2025 annual report showed revenue of 4.2 billion SEK, up 18% year-on-year. Its order backlog stood at 6.1 billion SEK. That's healthy, but it also means W5 is already running near capacity. Adding Sweden's conscript demand to existing NATO contracts—including a 2024 deal with Finland for mobile training systems—stretches the supply chain. W5's lead times for its core simulator product, the MTS-2000, have already extended from 9 months to 14 months over the past two years, according to procurement records from the Swedish Defence Materiel Administration.

The AI wargaming wild card

Then there's the AI angle. On March 6, Sahara AI's CEO Xiang Ren published a piece in Decrypt about the U.S. military's Thunderforge AI program, which aims to integrate AI into wargaming and operational planning. The program, run by the Defense Innovation Unit, awarded contracts to Scale AI and Palantir in late 2024. The goal is to generate thousands of wargaming scenarios per day, compared to the handful that human analysts can produce.

This sounds impressive. It's also mostly unproven. Thunderforge has been in development for 18 months. Public documents from the DIU show the program is still in "experimental validation" phase. The core problem, as Xiang Ren notes, is reliability. Military simulations depend on accurate models of adversary behavior, logistics constraints, and weapon system performance. AI systems trained on historical data often fail when confronted with novel tactics or asymmetric conditions. A 2023 RAND study on AI wargaming found that neural-network-based agents "consistently underestimated the effectiveness of irregular warfare tactics" by factors of 2x to 5x.

The irony is that the video game industry—specifically Wargaming.net, the company behind World of Tanks and World of Warships—might have more relevant simulation experience than the defence contractors. On December 13, 2024, Wargaming.net announced a shift into sci-fi mech combat games. But the company's core competency isn't fantasy. It's building massive multiplayer environments with realistic physics, ballistics, and terrain interaction. World of Tanks alone has over 160 million registered players. The chaos of a 15-versus-15 multiplayer battle, with variable latency, human error, and emergent tactics, is closer to real combat than most scripted military simulations.

The pipeline conjecture

That's where the "gaming-to-warfare pipeline" argument from Defense Analyses and Research Corporation gets interesting. On March 14, 2026, the firm tweeted that "the FPS/RTS of this era are now texts that can be read to reveal tactical intuitions, geographic instincts, and strategic propensities." It sounds academic. It's also empirically testable.

Consider this: A 2022 study by the U.S. Army's Simulation and Training Technology Center found that soldiers who played first-person shooters for 20+ hours per week scored 28% higher on tactical decision-making tests than those who didn't. The sample was small—only 180 participants—but the effect size was statistically significant. More recently, the Israeli Defence Forces have incorporated commercial off-the-shelf games like DCS World into their pilot training pipeline. DCS World, a flight simulator developed by Eagle Dynamics, costs $80. A comparable military-grade flight simulator costs $5 million. The fidelity gap is closing.

The implication is that the military simulation industry faces disruption from below. Companies like Bohemia Interactive, which makes the Arma series, already supply the U.S. Marine Corps with the Virtual Battlespace 4 (VBS4) training system. VBS4 is a direct descendant of a commercial game. It costs a fraction of purpose-built military simulators. And it's updated more frequently, because the commercial version has a paying customer base demanding improvements.

The red team questions

So here's the stress test for anyone investing in or relying on the military simulation industry:

Can W5 and its peers actually deliver? Scale is the constraint. Sweden's conscript demand is real, but if W5's lead times extend further, the bottleneck shifts from training capacity to production capacity. The Swedish Armed Forces may end up buying cheaper, less capable systems from competitors like Rheinmetall or Saab. W5's stock price, which has risen 34% over the past 12 months, already prices in the Quartz Sea thesis. Any delivery miss will compress that multiple.

Does AI wargaming actually improve outcomes? The Thunderforge program is promising, but the RAND study suggests AI systems are brittle. If Thunderforge's agents fail in live exercises—or worse, produce bad plans that get adopted—the backlash could set the field back years. The Pentagon's record with AI is mixed. Project Maven, launched in 2017, took four years to achieve operational capability. Thunderforge is on a similar timeline.

Will commercial games eat the military simulation market? The cost differential is massive, and the quality gap is narrowing. Bohemia Interactive's VBS4 costs about $10,000 per seat. A comparable purpose-built system from Lockheed Martin costs $150,000 per seat. The Pentagon's 2025 budget request for training simulators was $4.8 billion. If even 10% of that shifts to commercial-derived systems, it's a $480 million revenue shift. That's existential for some smaller defence contractors.

The military simulation industry is growing. That's not in dispute. Global spending on training and simulation is projected to reach $12.3 billion by 2028, according to a 2025 report by MarketsandMarkets. But the composition of that spending is changing. The question isn't whether demand exists. It's whether the current suppliers can meet it—and whether the next generation of suppliers will make them irrelevant.