The lights dim, and the screens flicker to life. Across the room, a dozen general officers lean over virtual terrain maps, eyes scanning for anticipated enemy movements tied to intelligence feeds as they look for candidates for possible pulse strikes. Each general is fighting another general, guided by only an AI staff assistant and a playbook with decision points and tailored options in lieu of a long operational plan. There is no staff circus churning out endless amounts of PowerPoint briefs. No legion of contractors running white cells and resurrecting their inner dungeon masters as they throw dice and debate overly complicated combat adjudication tables. It’s more Thunderdome than Title 10 and global games of old.
One general tries to pull the enemy out of position using deception. Another gambles on a high-risk, high-payoff multidomain deep strike against command-and-control nodes. There are no referees—just results. A publicly visible leaderboard ranks all officers in the room by their ability to synchronize effects and outfight a thinking opponent.
This isn’t science fiction. It’s what should happen when we talk about preparing our senior leaders for war. It describes a modern kriegsakademie built not on endless seminars, lectures, and games too big to fail, but actual competition. In this professional fight club commanders are forced to prove they can actually fight against someone trying to beat them.
While Washington is filled with calls to cut the number of generals and consolidate headquarters, these measures will prove insufficient to increasing lethality barring a deeper change to modern military culture. Cutting billets won’t make the joint force better at its core mission—fighting and winning wars—if the officers who remain don’t regularly compete, rehearse, and prove they can outthink and outfight a capable adversary. The core question we should be asking is simple: Do our senior leaders fight enough to understand advantage in modern war? Being the best marksman or top gun graduate when you’re young is no guarantee you have what it takes to fight large formations in multiple domains.
The sad truth is that senior officers spend more time navigating policy meetings than practicing the art and science of war. Without a culture of competition—of trial by simulated fire—we risk fielding a military led by assistant managers of violence rather than warfighters. Lethality isn’t a budget line or a briefing slide; it’s a habit built through pressure, repetition, and the humility of losing in a safe-to-fail environment.
This article is a call to renew the “fight club” spirit in the profession of arms, but from the top down. Secretary of Defense Pete Hegseth should sign a policy requiring general officers to fight each other in wargames tailored to modern scenarios. These games should capture objective data that helps leaders refine their ability to make decisions under pressure and hone their judgment in relation to tactics, doctrine, enemy order of battle, and emerging capabilities. Getting it right will require more than the proverbial stroke of the policy pen. The Department of Defense will need to invest in AI and data infrastructure, using the new fight club as an experiment for broader, more dynamic, and creative AI applications across the services integrated with emerging ideas about training and education in the US Army and the US Marine Corps in the era of agentic AI.
Why the Revolution Stalled
This isn’t a new idea. It’s a return to one long championed by voices across the profession of arms. Over a decade ago, multiple think tanks and federally funded research corporations created new wargaming centers and outlets like War on the Rocks starting publishing more appeals for renewing a culture of wargaming to transform the military profession. These calls paralleled service initiatives such as the Warfighting Society in the US Marine Corps. In many ways, they were made possible by senior leaders like Bob Work. Underpinning this wargaming movement was a shared premise that the best way to prepare for future wars was to create spaces where officers compete, fail, and learn. The hope was that gaming would unlock a revolution transforming military training and education. The effort later influenced a range of academic efforts including work at the University of California, Berkeley’s Center for Long-Term Cybersecurity and Risk and Security Lab, MIT, Stanford University’s Hoover Institution, and King’s College London, among others. This academic turn viewed games as a method for systematically analyzing decision-making in international relations.
Yet, as one of those authors and an early proponent of integrating AI into wargaming, it’s clear to me that we made a fatal assumption. We assumed that a culture of warfighting could be organically built from the bottom up. As fight clubs proliferated—largely led by officers like Arnel David and others across NATO countries—they never seemed to take hold and produce enduring institutional change in military services. Why?
There is a simple but sad answer: No one required the leaders to fight. And as Stephen Rosen told us back in the 1990s, military innovation is a function of clashing communities inside the military profession. Change requires making champions and champions need to be flag officers. In other words, no change to the promotion pathways and expectations of general officers, no military innovation and corresponding increase in lethality.
In place of honing the craft of modern joint warfighting, today’s generals find their days destroyed by endless meetings and bureaucratic nonsense. Each passing year takes them further from honing their ideas about large-scale combat through constant sets and reps. The bureaucracy and a Gordian knot of policy nonsense have taken their toll. Contrast the state of affairs with how UFC fighters train. They don’t get better by talking about fighting. They get better by sparring and aligning entire training regiments against their strengths and weaknesses alongside those of their opponents.
The dominant culture in the military today sadly still rewards briefing fluency over battlefield fluency. Incremental efforts to increase the use of games across the force have yet to trigger real action at the flag officer level.
In defense of senior leaders, the biggest thing they lack is time. Again, the bureaucracy takes its toll. Senior leaders in corporate America would never allow their days to be as colonized by nonsense and broken administrative processes as the situation that confronts American flag officers daily. It is a disgrace. Our only hope is that in addition to stealing American intellectual property, the Chinese Communist Party also copied the blueprint of this broken system.
Fight to Learn: Competitive Wargames as a Professional Mandate
To escape the current malaise, the first step is simple but radical: Make general officers fight each other. At least quarterly, officers should compete in time-constrained, decision-forcing games linked to major warfighting scenarios and emerging threat capabilities. These games should include a mix of scenarios involving joint task force–level organizations integrating multiple domains in large-scale combat as well as competition games that align multiple instruments of power and even bring in senior officials from the Departments of State, the Treasury, Commerce, and Justice, as well as the intelligence community, consistent with the joint concept for competing.
General officer means just that, a generalist. Hence, infantry division commanders should fight reserve staff officers. J2 intelligence officers should fight J4 logistics officers. Guardians should fight sailors and airman should fight soldiers. While the effort could grow to include small numbers of supporting senior field-grade officers and a mix of scenarios linked to warfighting functions, the core would stay the same: Make them fight. Make general officers hone their decision-making skills and use the games to collect critical data. As stated above, the scenarios should range from campaigns linked to major war plans to competition operations, activities, and investments designed in peacetime to set conditions and deny benefits and impose costs on adversaries.
These are not sprawling, week-long planning events. They are tight, tactical and operational contests—thirty to ninety minutes—where leaders must outthink, outmaneuver, and outfight opponents. Think of it as tactical chess with joint effects and defined by uncertainty and feedback loops. Around each match is a tailored learning environment. Officers receive updated packets on adversary capabilities, emerging doctrine, and new strategic conditions and objectives. They build cognitive fluency through repetition. And unlike most professional military education programs, these contests rank performance. Every officer gets feedback. Every decision gets reviewed. The goal isn’t shame—it’s sharpening. The only unforgivable sin is failing to prepare for the next fight.
From Data to Dominance: Analytics-Driven Training Transformation
Games are great, but data is king in an AI era. Having general officers fight each other provides a unique forum for capturing data on decision-making critical to both training AI agents and transforming training and education.
Every battle generates structured data about each leader: decision timelines, synchronization of fires and effects, doctrinal comprehension, and cognitive load under pressure. Combined across the force, this data becomes a diagnostic engine. Imagine a fusion of the Army’s synthetic training environments and the Marine Corps’s Project Tripoli. Both aim to deliver adaptive, AI-enhanced training—but they need fuel. That fuel is rich wargaming data that can be used to better understand decision-making at the individual and force levels. At the individual level, general officers can receive tailored self-development reading lists based on the results of the game. One general might need a refresher on joint air planning. Another might need a deeper dive into the electromagnetic spectrum and space effects. As the games collect data, they also provide a snapshot of core knowledge gaps across the force. These could be harvested and used to adjust requirements for downstream professional military education courses. This would produce a more dynamic feedback loop for updating training and education requirements than the current process of workshops, councils, and PowerPoint briefings on loosely assessed outcomes-based education
Furthermore, this data-rich environment would allow for better statistical analysis of how general officers performed across core warfighting competencies. It would create data dashboards on performance that told a richer story than narrow, bulletized comments on officer evaluation forms and the buddy network gossip about who has what it takes for the next level. It would also complement procedures implemented at lower-echelon commands like the US Army’s Command Assessment Program.
Competition without consequence is just a game. To change the culture, performance in these wargames must be tied to what matters: command selection, school slating, and promotion boards. Just as physical fitness and weapons qualification shape enlisted progression, warfighting scores should shape officer trajectories. Did a general consistently finish in the top decile for managing joint fires? That should matter for career progression and command opportunities. Did a leader repeatedly fail to demonstrate integration of joint effects under stress? That should raise flags for command. Reward mastery. Identify gaps early. Build a system that says, loud and clear, if you want to lead in war, you have to prove you can fight.
Build the Infrastructure: Invest in Agents and Distributed Combat Labs
To make this vision a reality, the Department of Defense must invest in the infrastructure required to run distributed battles between senior officers, collect the data, and train AI agents linked to refining self-development alongside training and education. That is no small task. Gaming companies and services like Steam offer illustrative cases. These platforms enable millions of users to simultaneously fight, analyzing a wide range of data to improve gameplay. Steam’s top games often have over one million people playing daily, while its library exceeds forty thousand games. This scale offers interesting insights for the US military, which should build its general officer fight club network in a manner that supports future scalability across the force. Furthermore, this network would need to operate at the secret level, which is often more expensive than unclassified compute infrastructure. In other words, the bill is not going to be small.
The ideal architecture includes—like Steam—the ability to play online and offline. Generals could run leader development programs locally at commands and service academies both refining their judgment and teaching their subordinates how to fight. The data would still be harvested, tagged, and used to train a wide range of AI agents. These agents could include mentors on doctrine, domains, and warfighting functions as well as enemy emulators that learn to fight like key adversaries but adapt their approaches to the profile of the each specific general they are matched against. This last dimension would enable a culture of continuous improvement based on a dynamic digital profile for each officer.
These profiles in turn would unlock deeper human-machine integration for the coming age of agentic AI. If the military moves to having clusters of AI agents running analysis in lieu of large staffs, the better these agents understand the commander, the better they are able to pass information, balancing channel output (data flows) with source and coding theory (i.e., tacit knowledge exchange). Working with AI agents should be second nature to every officer, just like running battle drills and more complex joint planning processes. In the future fight, human-machine teaming will define tempo. The side that better integrates agents into planning and rehearsal will hold the cognitive high ground.
The Courage to Compete
Defense leaders say they want a more lethal force. That will require changing the culture and making senior leaders fight each other. The military briefs readiness, so why not the outcome of simulated combat? Unless defense leaders force US officers—especially senior ones—to routinely compete in digital gaming environments that simulate modern conflict, America’s warriors are just managing a bureaucracy, not leading a military.
The historical record is clear: The best militaries embrace continuous rehearsal and analysis of operational dilemmas. Officers fight on maps and in simulations before they ever fire a shot in anger. Today, the US military has the tools to do this at scale. The only thing missing is the will.
It’s time to build the infrastructure, train AI agents, and institutionalize a culture of competition. The US military needs to create a fight club from the top down that’s more than a metaphor. It is time to track performance, tailor learning, and reward mastery. If senior leaders are serious about the future fight, then generals should have to prove they can fight in complex, multidomain battles and campaigns. Not once. Not hypothetically. But routinely, under pressure, against an opponent trying to win.
If warriors don’t practice fighting, they shouldn’t be surprised when they forget how.
Benjamin Jensen is the Frank E. Petersen Chair of Emerging and Disruptive Technology at the Marine Corps University School of Advanced Warfighting and the Director of the Futures Lab at the Center for Strategic and International Studies. He is also an officer in the US Army Reserve. The views expressed are his own and he is open to fighting against any officer in joint warfighting games to facilitate learning.
The views expressed are those of the author(s) and do not reflect the official position of the United States Military Academy, Department of the Army, or Department of Defense.
Image credit: David Poe, Fort Bliss Public Affairs Office