AI in Defense Weekly - 19 November 2025

The government has reopened, the Defense AI market shows no signs of slowing down in spite of broader fears of an AI bubble, and thoughts about how AI can improve the military PME system.

Brian Lampert

and

Ben Van Roo

Nov 19, 2025

The Role of AI in Military Education

A recent essay from Dr. James Lacy at Marine Corps University makes a passionate argument that AI will “hugely enhance professional mliitary education,” do it “with considerably less administrative staff, fewer teachers, and, thus, at significantly less cost per student.” If they fail to do this, they “risk academic oblivion.”

Based on his experience from war college students (professionals with ~20 years of military experience), he recommends:

Transitioning the role of teachers from running seminars to coaching, with AI modules empowering students to learn at their own pace similar to the model used at Alpha School.
Leveraging AI to enhance and augment critical thinking. To quote one of his students: “I am using AI as a strategic amplifier and cognitive multiplier.”
Buy back time for students and professors to conduct research and wargaming activities.

The Ghost in the Code

minor self-promotion, Beyond Visual Range covers Anthropic’s GTG-1002 campaign, as it represents a critical operational shift from AI as a passive informational tool (chat) to an active, autonomous combatant (agentic), fundamentally altering the threat landscape for the Department of Defense. By demonstrating that state-sponsored actors can now utilize agentic workflows to orchestrate 80-90% of a cyber-espionage lifecycle without human intervention, this event signals the arrival of algorithmic warfare where agents autonomously execute complex tactical chains—reconnaissance, lateral movement, and exfiltration—at machine speeds impossible for human operators to match. For the DoD, this necessitates an immediate doctrinal pivot: traditional human-speed defenses are now obsolete against adversaries capable of sub-linear scaling through compute rather than headcount, demanding a counter-strategy that integrates defensive AI agents to match the adversary’s compressed OODA loop and secure the digital battlespace.

The Importance of Open-Weight and Open-Source Models for the DOW

Many of the War Department’s systems are not connected to the cloud, making accessing cloud-hosted frontier lab models challenging. While sometimes this is due to technical debt and antiquated systems, but it’s more practically related to security concerns and the need to operate in a comms degraded or denied environment: the military cannot rely on an assured cloud connection during a conflict with a sophisticated adversary.

This leads developers of military AI solutions to often turn towards open-source models like Meta’s Llama or OpenAI’s gpt-oss. Wired describes how open-source and open-weight models reduce dependance and “a degree of accessibility, control, customizability, and privacy simply not available with closed models.” Incentivizing U.S. model providers to make increasingly powerful offline models available to the War Department will be critical to keep pace with China.

Are We Approaching an (Defense) AI Bubble?

The markets remain concerned about the risk of the AI bubble popping. A Bank of America survey of institutional investors reported that a majority of investors believe AI stocks are already in a bubble and that this represents the biggest risk to the market.

Obviously a “pop” would have negative impacts on Defense AI industry: an ebbing tide lowers all boats. However, a recent estimate suggests a $35B global TAM for military AI by 2035, with $1.8B spent by the DOW in 2024 alone. So while the overall market may be overheated today, the military requirements for AI aren’t slowing down. The real question isn’t wither AI, it’s what companies emerge victorious: AOL or Amazon.

A Canary in the Coal Mine? Palantir is valued at ~$400B, much more than traditional defense heavyweights with larger revenues, with an impressive price-to-earnings ratio of over 390. A recent New York Times article questions “how the stock market can keep supporting (Palantir’s) remarkable share price” and Michael Burry of Big Short fame made waves betting against Palantir (and Nvidia).

The Forum - AI & Defense

Discussion about this post

Ready for more?