Emil Michael: Iran, Anthropic and the Future of AI at the Pentagon

Episode Summary

Executive Summary: Emil Michael argues the Pentagon must move at wartime speed to keep pace with China and AI’s rapid expansion. He says he cut outdated tech priorities from 14 to 6, elevated applied AI, and exposed dangerous AI contract restrictions that could disrupt military operations. His goal: modernize procurement, diversify vendors, and use AI lawfully without ceding command and control to private companies.

Main Topics: Military modernization and wartime speed (Priority: 5/5): Michael frames the current era as the biggest military buildup in history and says the Department of Defense cannot operate at peacetime speed if it hopes to compete with China and secure supply chains. Reprioritizing technology priorities (Priority: 5/5): Upon taking office, he found 14 stagnant critical technology priorities that were too vague to act on. He reduced them to six, focusing on areas with the greatest impact on combat power and industrial growth. AI adoption inside the department (Priority: 5/5): Applied AI was made the top priority and integrated into the department’s digital strategy. Michael emphasizes using AI to improve enterprise efficiency, intelligence analysis, and warfighting/logistics. Concerns over commercial AI contracts (Priority: 5/5): He describes discovering restrictive AI contract terms that could allow vendors to shut off systems during operations, especially in sensitive military commands. He argues this creates unacceptable operational risk. Democratic oversight and lawful use of AI (Priority: 4/5): Michael argues that AI used by the military must be governed by U.S. law and elected institutions, not a company’s internal values or ‘constitution,’ particularly in the context of command and control. Procurement and industrial-base reform (Priority: 4/5): He says the department is shifting away from cost-plus, vague requirements, and endless change orders toward simpler requirements, fixed-price deals, and faster yes/no decisions to help startups scale. Role of startups and national service (Priority: 3/5): Michael encourages founders to support defense innovation, noting that startups need manufacturing scale and quality systems, while the government needs more patriotic technologists willing to serve.

Key Arguments: The Pentagon inherited outdated, vaguely written tech priorities that were impossible to act on, so reducing them to six created focus and urgency. AI is no longer optional; it is becoming a foundational layer like the internet and must be embedded across enterprise, intelligence, and warfighting functions. The department was dangerously single-threaded on certain AI vendors, creating operational risk if a model could be turned off mid-mission. Commercial AI companies cannot be allowed to impose their own internal values as governing authority over American military command and control. Democratic institutions and law must govern military AI use; if laws are outdated, Congress and regulators should update them rather than companies making unilateral rules. The department must move away from cost-plus procurement and toward firm-fixed-price, clearer requirements, and faster decision-making to accelerate adoption. Startups can win defense business if they learn to scale manufacturing and prove they can pass through real procurement and testing processes. The current geopolitical environment, especially China’s military buildup and supply-chain leverage, requires re-domesticating critical production and building self-reliance.

Data Points: Critical priority areas: 14 reduced to 6 - Michael says the department had 14 outdated critical technology priorities, which he cut to six. Department AI usage: 1.2 million of 3 million personnel - Within 90 days, he says 1.2 million personnel had used some form of AI. Previous AI usage level: 80,000 personnel - He compares current usage with the number before he started. Timeframe for AI adoption increase: 90 days - Period over which department-wide AI usage grew sharply. Personnel in the department: 3 million - Total department workforce referenced when discussing AI adoption. Number of frontier AI companies: 4 - He describes four frontier companies dominating the AI market. Researchers in the AI ecosystem: 1,000 - He notes a large pool of researchers trading among companies. Years tech priorities remained unchanged: Nearly a decade - The 14 priorities had not materially changed for almost ten years.

Pivotal Quotes: "we've got a lot of catching up to do" — Emil Michael: Explaining why the U.S. must shift from peacetime to wartime speed in response to China’s buildup and supply-chain dependencies. "the software, the soul, someone's soul of their model, their constitution, which is not the U.S. Constitution, can't be dictating our command and control environment" — Emil Michael: On why vendors cannot control military use of AI systems through internal model policies or shutdown terms. "I had a holy cow moment" — Emil Michael: Describing his reaction to discovering restrictive and risky AI contract terms embedded in sensitive defense operations.

Implications: Defense AI is moving from experimentation to governance, procurement, and operational risk management. Companies that want to work with the military will need scalable products, reliable support, and alignment with democratic oversight rather than unilateral vendor control.