Defense Analyst.
AI is compressing the research and synthesis work at the core of defense analysis while creating an entirely new analytical task: evaluating AI itself as a national security variable.
Low risk, high transformation.
Defense analysis depends on judgment, source evaluation, political context, and accountability for conclusions that inform high-stakes decisions. AI cannot replicate those requirements. Demand for the role is rising alongside defense AI budgets and the growing complexity of the threat environment. What is changing is the mix of tasks: AI has compressed the collection and synthesis work that historically defined junior analyst output, and a new analytical category has emerged around AI systems themselves. The entry point into the profession is narrowing for analysts whose value was primarily in document processing. The ceiling for analysts with strong judgment and AI fluency has risen.
3 shifts already visible in the data, in order of magnitude.
Open-source intelligence synthesis has been fundamentally accelerated.
AI tools can now scan, translate, and summarize large volumes of publicly available material across languages and formats faster than a junior analyst could process a single document set. RAND already uses AI to search thousands of research reports to surface relevant material for wargame design and assessment work. The analyst's value is shifting from collection toward interpretation and judgment about what the synthesized material actually means.
The DoD's 2026 AI strategy made AI-first operations a mandate, not a pilot.
The Department of Defense's January 2026 AI strategy requested $13.4 billion for AI systems in FY2026 and directed every service and component to deploy AI across warfighting, intelligence, and enterprise operations under aggressive timelines. For defense analysts, this means AI tools are arriving across their organizations whether or not their specific workflow has been redesigned around them.
Evaluating frontier AI is now a standing analytical problem.
This week, GPT-5.6 Sol was limited to government-approved partners during a national security review, and Anthropic's Mythos 5 was restricted to approved U.S. cyber-defense organizations. The question of what a given model can do, how adversaries might use it, and what access controls are appropriate is now a recurring analytical task. Defense analysts at think tanks and government agencies are being pulled into AI capability assessment alongside their traditional subject-matter work.
What the leaders are doing.
| № | Company | Sector | What they are doing | Year | Source |
|---|---|---|---|---|---|
| 01 | U.S. Department of Defense | Federal Government | The DoD's 2026 AI strategy requested $13.4B for AI in FY2026 and directed AI-first operations across warfighting and intelligence. Maven Smart System, built on Palantir's platform, was formally adopted as a DoD program of record. | 2026 | media.defense.gov ↗ |
| 02 | RAND Corporation | Defense Think Tank | RAND researchers use AI tools to scan thousands of research reports, surface relevant material, and assist in designing wargames. The organization has integrated AI into its research pipeline as a productivity and synthesis tool while maintaining human editorial judgment on all published analysis. | 2025 | rand.org ↗ |
| 03 | Palantir Technologies | Defense Technology | Palantir AIP and Gotham are embedded in U.S. and allied military decision-making. The Army enterprise agreement, potentially worth up to $10 billion over a decade, extends Palantir tools across operations, logistics, and intelligence. | 2026 | military.com ↗ |
What is declining, growing, emerging.
- 01Manual open-source intelligence collection and first-pass document triage
- 02Rote translation and summarization of foreign-language source material
- 03Single-thread scenario production for wargames and tabletop exercises
- 01Evaluating and stress-testing AI-generated intelligence assessments
- 02Structured analytic techniques applied to AI outputs to identify gaps, biases, and failure modes
- 03Cross-domain synthesis combining AI-processed data with human source reporting and political context
- 04Communicating AI capability assessments to non-technical policymakers and commanders
- 01AI capability analysis as a formal analytical discipline: assessing what frontier models can do, at what speed, and with what risk
- 02Adversarial AI assessment: how peer competitors may use or restrict frontier AI systems
- 03Policy analysis of model access controls, export frameworks, and compute governance as national security instruments