All-In Podcast Ep. 263 – Emergency Pod: Inside the Iran War and the Pentagon's Feud with Anthropic ft. Under Secretary of War Emil Michael

Date: March 6, 2026
Guests: Emil Michael (Under Secretary of War for Research & Engineering)
Hosts: Jason Calacanis (A), Chamath Palihapitiya (D), David Friedberg (C), [David Sacks absent]

Episode Overview

This emergency episode explores two urgent breaking stories:

1. The rapid US-Israeli joint military operation against Iran (“Operation Epic Fury”), the regime change implications, and its ripple effects on global strategy, especially vis-à-vis China.
2. The unprecedented Pentagon conflict with AI company Anthropic, resulting in Anthropic being formally deemed a “supply chain risk” and the broader debate over the moral and technical risks of AI in defense.

Under Secretary of War Emil Michael, a high-level Pentagon technocrat formerly of Team Uber, joins the besties to provide firsthand insights.

1. The Iran War: Operation Epic Fury

Key Developments

• [03:30] The US & Israel launched a major strike on Iran (Operation Epic Fury).
  • Iran's Supreme Leader and 40 senior officials have been killed; ~1,000 total fatalities (03:30-04:00).
  • Six US Army Reserve soldiers died in a Kuwait drone strike.
  • US submarine sank an Iranian ship off Sri Lanka—the first US torpedo kill since WWII.

Administration Objectives

• Trump insists: “This is not a regime change effort ... People of Iran should seize the moment and take their country back.” (03:22)
• Hegseth, Emil’s boss: “Not a ‘regime change war’ ... but the regime sure did change and the world is better off for it.” (03:30)

Discussion: Duration, Strategy & Success Metrics

• Q: Will there be US boots on the ground?
  • Emil Michael:
    • Operation is “weeks, not months”—minimal ground troops, focused on strategic strikes to “disarm the regime,” especially disrupting arms/nuke supply to terror proxies (04:36-05:40).
    • “There's no scenario where we have some protracted boots on the ground—Afghanistan, Iraq, two-like scenario.” (05:33)
• Friedberg: Sees actions in Iran/Venezuela as “leverage building” for critical China negotiations in April, especially since Iran supplies 90% of oil to China.
  • “If there's some transaction with China that gets everyone out of this and puts the US on a strong footing ... it could be a win-win ... A deal with China could be the crowning achievement of this administration, particularly going into the midterms.” (06:03-08:07)
• Chamath:
  • Broadens to global context. By taking out Maduro and Iranian leadership, US created “enormous leverage with China.”
  • Notes China's GDP growth targets are at a 30-year low; high youth unemployment breeds chaos.
  • Predicts, “If your goal is to prevent war with China ... you cut off their oil supplies from Iran and Venezuela—now, all that's off the table.” (08:38-12:20)

Technology & Military Innovation

• No Boots, Precision-First:
  • Emil:
    • Success tied to smart, “unfair” advantages: mature technology, updated rules of engagement (“operational axe”), and battle-hardened leadership (“The people now fighting are generals—they’ve learned a lot of lessons.” (14:08-15:12))
    • “We have a very sophisticated way of doing these things to minimize loss of life and maximize success.” (15:42)

The “Discombobulator” & Technological Leap

• Chamath and Jason probe rumors of a “discombobulator” super-weapon. Emil: “Can’t talk about it.” (15:42-16:06)
• Compared to a decade ago, success now is enabled by:
  • Advanced drones, “drone swarm” tactics (18:05-21:00)
  • Relaxed rules of engagement (“Afghanistan’s rules were insane ... now we’re more pragmatic.” (16:22-18:05))
  • AI as command-and-control; human-in-the-loop where risk is high, full autonomy only for domains like missile defense (“Golden Dome” scenarios) (23:12-24:36, 26:22-26:55)
  • Emergence of Anduril and other “new primes” as disruptors in defense tech.

AI and Automation in Warfare

• Drone deployment at unprecedented scale; AI-guided swarms; autonomous defense (“robot on robot warfare”)—future is “attritable,” low-cost, AI-enabled platforms replacing traditional big-ticket weapons. (18:21-20:53)
• AI error rates and moral dilemmas still critical hurdles; currently, no Skynet risks—US is “nowhere near” ceding life-and-death autonomy to AI (26:22-27:41).

Influence & Israel

• Critique from MAGA right: Is Israel driving US action?
  • Chamath: Dismisses “capture” narrative. Israel’s intelligence is “incredibly capable” & operationally valuable, but US still leads decisions (29:30-30:27).
  • Israeli air defense (Iron Beam) “works,” but true global deterrence awaits space-based lasers—technological progress happening, mostly collaborative with Israel (30:46-33:05).

Economic & Oil Impact

• Oil at $84/barrel; insurance cancelled for Gulf shipping; US steps in to offer war-risk insurance, a potentially lasting shift in global maritime insurance market (38:21-41:19).

2. Pentagon Feud with Anthropic: The Clash over AI, Ethics & National Security

Breaking News: [41:19] The Pentagon formally designates Anthropic (maker of Claude) a “supply chain risk,” cancels $200M contract—the first time ever for a US company.

Why the Conflict?

• Emil Michael lays out the backstory:
  • After Biden’s executive order on AI, only a few model makers (inc. Anthropic) were approved as “winners” (42:40-43:00)
  • Anthropic embedded deeply into DoD’s most sensitive workflows—but restricted their tech:
    • No model use to “plan kinetic strikes,” move satellites, or underwrite scenarios involving harm (“This is the Department of War. This is what we do.” 43:58-44:00)
  • Negotiations with Anthropic dragged on—Anthropic would only grant narrow “exceptions,” not blanket “all lawful uses.” (44:00-45:43)
  • Anthropic also objected to government use of the model for public data aggregation (mass surveillance fears). Attempts to check if their tech was used in classified raid caused deep alarm within DoD: “What if this software went down, some guardrail kicked up, some refusal happened for the next fight—leaving our people at risk?” (46:45-47:11)

The Big Debate: Moral Control Versus Mission Readiness

• Chamath:

  • “What Anthropic showed ... is that they're going to have a political perspective and a set of terms that reflect their philosophy—which can change on a dime ... For the government, you cannot choose to only use one of these things ... it's just a matter of time until some person inside goes on some lunatic moral tirade and jeopardizes your business.” (48:45-50:49)

• Emil:

  • “This is a general substrate of intelligence ... The controlling—whether it has a moral conscience—Anthropic has its own constitution, its own soul. It's not the US Constitution ... A scary thought for Americans.” (50:49-51:51)
  • The final straw: Anthropic’s reluctance to guarantee service in all lawful warfighting contexts, combined with worries about sabotage or last-minute model changes.

Industry Response & Comparisons

• Grok (Elon Musk): “All in” for all lawful uses, both classified/unclassified (52:09-52:28).
• Google (Gemini): “In compliance” but ramping up for classified settings; strong potential due to unique cloud economics (53:05-54:11, 66:29).
• OpenAI: Sam Altman tried to broker peace; willing to step in if Anthropic falters (52:54-53:33).

Why the “Supply Chain Risk” Designation?

• Emil: “I don't want Lockheed Martin using their model to design weapons for me ... If that poison can enter into any part of the defense enterprise ... I can't have that.” (57:38-58:29)

PR, Politics, and Perception

• Dario Amodei (Anthropic) claims the issues are political/retribution for not supporting Trump.
  • Emil: “That's one of the dumbest things I've ever heard. ... Why would I spend three months trying to negotiate with them to get to a simple standard if I would have just said, okay, guys, you're out. Bye.” (67:57-68:47)

The Boardroom Dilemma

• Chamath/Friedberg debate: Is Anthropic sacrificing patriotic duty? Are they simply letting Dario “do his thing” due to astronomical ARR growth? Value of company seen as multiples above even $350B—unless this debacle “blows them up.” (62:39-63:57)

The Future: Model Redundancy, Competition, and New “Primes”

• “As models converge, the DoD intends to remain agnostic—always ready to switch.”
• Defense procurement is shifting: “old primes” (Lockheed, Raytheon) giving way to “new primes” (Anduril, Palantir), but a second layer of disruptors is vital. (75:01-75:22)

3. Geopolitical and Economic Second-Order Effects

• Maritime insurance market could shift to US, ending 400 years of Lloyd’s of London dominance (41:19-41:30).
• Defense “industrial base” is being reshored; strategic capital flowing to domestic capability (71:14-73:20).
• Current munitions are “Cold War vintage”; need to plus up to newer, more advanced weapons as global conflicts deplete stockpiles (73:04-73:50).
• Defense venture capital is 3x last year’s levels; innovation is exploding, but bureaucracy and contracting reform remain hurdles (74:10-77:20).

4. Quick Hits, Notable Quotes, & Memorable Moments

• Chamath on AI/Tech Power:
  • “Anthropic has its own constitution, it has its own soul. It's not the US Constitution.” (50:49-51:51)
• Jason (on regime change):
  • “If you can free people, give him all the Nobels. Like, literally, if you can free people, all of them. Give him every prize.” (36:16-36:24)
• Friedberg on insurance innovation:
  • “Very smart. And ultimately a lot of people are saying this could actually reshore maritime insurance back to the US and create an entirely new insurance industry ... That's cool.” (38:21-41:19)
• Emil (on bureaucracy):
  • “We’re trying to change that ... as little requirements as possible on that side, and then on the contract piece, trying to get to as close to commercial contracts as possible.” (76:15-77:20)
• Friedberg (on defense spending):
  • “I feel like this week was a true reminder of how fortunate we are to have the defense that we have for the United States.” (81:04-81:47)
• Classic All-In banter:
  • Jason trying to assert his "independent" label (34:22–36:30).
  • “Is the discombobulator real?”—ongoing running gag (15:42-16:06).

Key Timestamps

• 03:30: US-Israel joint operation recap, regime change headlines
• 06:03–08:07: Friedberg on grand bargain with China
• 14:08: How “no boots on ground” operations are succeeding
• 18:05–24:36: Drone/autonomous warfare, rules of engagement
• 38:21–41:19: Oil, maritime insurance, economic shocks
• 41:19: Pentagon cancels Anthropic, supply chain risk debate
• 44:00–47:11: The technical & legal saga over Anthropic’s restrictions
• 52:09–54:31: How OpenAI, Grok, and Gemini compare
• 71:14–73:20: Industrial supply chain and “new primes” strategy
• 74:10–77:20: Bureaucracy and contracting reform
• 81:04–81:47: Friedberg’s “thankful for defense” reflection

Summary Takeaways

• Operation Epic Fury marks a major inflection in US military doctrine: Swift, high-tech, limited footprint interventions aimed at depriving adversaries of operational capacity—and, more subtly, shifting the global balance (especially vs. China).
• AI is now an operational and moral fault line: The Pentagon wants “all lawful use”—not model makers’ philosophical guardrails. The Anthropic flare-up is an early salvo in the battle over AI’s place in warfighting.
• Expect a new era of “defense tech primes” and major industrial reshoring: Not just for weapons, but for everything from AI to batteries and minerals.
• The conversation is marked by unusual candor, genuine inside baseball, and a streak of classic All-In irreverence (plus inside jokes about politics, Vegas, and the “discombobulator”).

This episode provides a rare inside view of how 2020s-era defense and tech leadership are integrating high-stakes geopolitics, AI risk, and Washington’s shifting priorities—all, as always, with the All-In crew’s no-filter banter and clarity.