a16z

Emil Michael: The Department of War Is Moving Faster Than Silicon Valley on AI | The a16z Show

This conversation with Emil Michael, undersecretary of defense for research and engineering and acting director of the Defense Innovation Unit, was recorded at the a16z American Dynamism Summit in Washington, D.C. Michael walks through how he inherited a department running 14 undefined technology priorities, cut them to six, and made applied AI number one. He also gives the first detailed account of why commercial AI contracts written under the previous administration created a vendor-lock crisis that put active military operations at risk. Timestamps: 0:00—Introduction 2:50—Why Wartime Speed Matters 5:20—Cutting Priorities to Six 7:22—Applied AI Across Defense 9:17—Commercial Models and Lock-In 14:39—Democratic Oversight and Guardrails 19:30—Fixing Procurement Bureaucracy 22:08—What Startups Must Deliver Read the full transcript here: https://www.a16z.news/s/podcast Resources: Follow Emil Michael on X: https://x.com/USWREMichael Follow David Ulevitch on X: https://x.com/davidu Stay Updated: If you enjoyed this episode, be sure to like, subscribe, and share with your friends! Find a16z on X: https://twitter.com/a16z Find a16z on LinkedIn: https://www.linkedin.com/company/a16z Listen to the a16z Show on Spotify: https://open.spotify.com/show/5bC65RDvs3oxnLyqqvkUYX Listen to the a16z Show on Apple Podcasts: https://podcasts.apple.com/us/podcast/a16z-podcast/id842818711 Follow our host: https://x.com/eriktorenberg Please note that the content here is for informational purposes only; should NOT be taken as legal, business, tax, or investment advice or be used to evaluate any investment or security; and is not directed at any investors or potential investors in any a16z fund. a16z and its affiliates may maintain investments in the companies discussed. For more details please see http://a16z.com/disclosures.

EMEmil MichaelguestDUDavid Ulevitchhost

Mar 13, 2026·26m·Watch on YouTube ↗

📖 CHAPTERS

1. 1

   Emil Michael’s path from Silicon Valley to Defense CTO

   Emil Michael explains how a post-acquisition break from tech led him into government through the White House Fellows program, where he worked under Secretary of Defense Robert Gates. That experience created a lasting interest in national security, setting up his return to public service years later.

2. 2

   Why “wartime speed” is necessary now (and how peacetime speed took hold)

   Michael argues the Department of War/Defense cannot operate with the post–Cold War tempo that encouraged consolidation and slower innovation. He frames today’s urgency around China’s historic military buildup and U.S. reliance on fragile, often offshore supply chains.

3. 3

   Supply-chain fragility as a national-security risk (the Skydio example)

   Ulevitch highlights how sanctions and geopolitical dependency can instantly break defense readiness when critical components are sourced abroad. The discussion emphasizes that the defense industrial base includes upstream “precursors,” not just final weapons systems.

4. 4

   Cutting 14 priorities down to 6 to drive execution

   On entering the role, Michael found an outdated, overly broad list of 14 “critical priorities” that had barely changed in a decade. He narrowed them to six to improve focus, clarity, and workforce alignment—putting applied AI at the top.

5. 5

   Applied AI: reorganizing for speed and driving adoption at scale

   Michael describes moving the Chief Digital and AI Office into his organization to accelerate deployment. He claims rapid adoption across the workforce, citing a jump from tens of thousands of users to over a million using AI tools in roughly 90 days.

6. 6

   Where AI helps most: enterprise efficiency, intelligence, and warfighting/logistics

   Michael breaks defense AI into three buckets: back-office productivity, intelligence analysis, and warfighting applications like logistics and planning. He stresses that AI can unlock value from massive, siloed datasets and multiply analyst throughput via anomaly detection and pattern discovery.

7. 7

   Commercial AI models in defense: restrictions, single-vendor risk, and operational exposure

   Reviewing prior contracts, Michael says he discovered extensive use restrictions and a dangerous level of vendor lock-in within sensitive operational commands. He warns that if model behavior or terms-of-service constraints can disable capability mid-operation, it creates unacceptable risk to missions and lives.

8. 8

   The Maduro raid catalyst and the clash between company values and lawful military use

   Michael cites a moment when a vendor questioned whether its software was used during a successful raid, which raised alarms about external influence over lawful operations. He argues that as AI approaches a “substrate” like the internet, private model constitutions cannot supersede democratically authorized command decisions.

9. 9

   Democratic oversight, civil liberties, and guardrails (without self-handicapping)

   The conversation emphasizes that guardrails for defense AI should flow from U.S. law and democratic processes, not private actors. Michael notes existing directives on autonomy and argues adversaries are removing guardrails and stealing models—making unilateral restraint strategically dangerous.

10. 10

    What “good” looks like: multiple frontier partners and never being single-threaded again

    Michael argues the U.S. needs redundancy and choice among frontier AI providers to avoid lock-in and policy shock. He references the post-2018 Project Maven shift—where some firms initially resisted government work but later became strong partners—as a model for today’s AI companies.

11. 11

    Fixing procurement bureaucracy: simpler requirements and firm fixed-price outcomes

    Michael describes a push to dismantle bureaucratic processes that favor slow, cost-plus development and unrealistic requirement lists. He advocates outcome-based requirements, faster cycles, risk-sharing, and fixed-price structures—citing the SpaceX-style incentive model as a blueprint.

12. 12

    What startups must deliver: production scale, real buying signals, and faster yes/no decisions

    Michael advises startups that primes’ key advantage is manufacturing and scale, not inventiveness—so new entrants must build production muscle and quality systems. He also urges founders to look for real procurement signals (tests, buys) and supports a cultural shift toward faster decisions so startups aren’t trapped in endless maybes.

13. 13

    Why serve: rebuilding the culture of government service and patriotism in tech

    In closing, Michael explains his motivation to serve as an immigrant who benefited from the U.S. system and wants his children to see that national strength requires sacrifice and builders. He calls for more technologists to view public service as an honored profession again, akin to historic moments like the Manhattan Project.