The Companies Changing Warfare Forever: Palantir & Anduril Execs on Drones, AI & the Future of War

Episode Summary

Executive Summary: Trey Stevens and Sean Sankar argue that U.S. defense is entering a new era where software, autonomy, and reindustrialization must replace the post-Cold War model of slow, centralized procurement. They defend defense tech as ethically necessary, criticize bureaucracy and industrial atrophy, and frame Anduril/Palantir as part of rebuilding deterrence, manufacturing capacity, and American legitimacy.

Main Topics: Origins of Palantir and Anduril (Priority: 5/5): The guests recount how they met through early Palantir demos and how lessons from Palantir informed Anduril’s founding, culture, and go-to-market strategy. War, deterrence, and the ethics of defense tech (Priority: 5/5): They argue war is awful but sometimes unavoidable, and that the moral goal of defense technology is deterrence, precision, and reducing casualties rather than enabling conflict. Decline of the American industrial base (Priority: 5/5): They describe how globalization hollowed out U.S. manufacturing, reduced surge capacity, and weakened readiness, making it hard to replenish munitions or scale production quickly. Defense procurement, monopsony, and innovation (Priority: 5/5): They criticize the government’s monopsony purchasing model, rigid specs, and bureaucratic processes, arguing that innovation requires competition, founder-led heresy, and more autonomy. Autonomy, AI, and human accountability (Priority: 4/5): They support autonomous systems with human accountability and argue AI should improve targeting precision and decision-making while remaining under democratic control. Culture war around defense tech and surveillance (Priority: 4/5): They push back on claims that Palantir enables a surveillance state, saying the tools are lawful, auditable, and often misunderstood by people far removed from military or intelligence service. Future of U.S. power and reindustrialization (Priority: 5/5): They warn that failure to rebuild industrial and defense capacity could lead to Chinese dominance, while success would mean renewed prosperity, stronger institutions, and a healthier middle class.

Key Arguments: Deterrence, not war, is the real objective of defense technology; the goal is to make conflict unthinkable to adversaries. Silicon Valley’s earlier disdain for defense work was a product of the post-Cold War peace dividend and a misunderstanding of geopolitical risk. The U.S. lost critical manufacturing depth during globalization, which now limits its ability to replenish munitions and scale production in wartime. Defense innovation is stifled by monopsony procurement, where the government dictates specs and suppresses competition and experimentation. Anduril’s model is to invest privately in R&D, then sell products, rather than waiting for government requirements to define the roadmap. Autonomous systems can be ethical if humans remain accountable and the systems reduce civilian harm through better precision and discrimination. Claims of a surveillance state are overstated because Palantir does not collect data; it organizes lawful data for decision-making and auditing. The defense sector needs more competition, but capital will likely concentrate into a few winners, similar to venture-backed markets. Political change matters less than leadership; bureaucratic reform depends on people willing to cut through sclerosis and empower innovators. If the U.S. fails to reindustrialize, it risks losing strategic primacy and entering a world where China sets the rules. Data Points: Palantir employee number: 13 - Trey Stevens is described as joining Palantir very early, around employee 13. Palantir early headcount: 25-30 people - The company was still extremely small when Trey joined in 2008. Anduril revenue milestone timing: 22 months - They say Anduril reached $10 million annual revenue in 22 months, faster than Palantir’s early trajectory. Palantir revenue milestone timing: about 5 years - Referenced as the approximate time Palantir took to reach $10 million annually. Anduril valuation: $60 billion - Mentioned as the company reportedly raising money at this valuation. Army contract: $20 billion - Referenced as a recent major Army contract landed by Anduril. Palantir valuation: $400 billion - Used to illustrate how far the company has scaled. Defense spending share to specialists in 1989: 6% - Only 6% of major weapon system spending went to pure-play defense specialists at the end of the Cold War. Defense spending share to specialists today: 86% - Today most major weapon spending goes to defense specialists rather than dual-purpose industrial firms. Defense spending share to dual-purpose companies in 1989: 94% - Most spending then went to companies with broader industrial capacity. Drone production gap vs China: 10,000 to 1 - Cited as a stark readiness gap in drone manufacturing. Shipbuilding capacity disadvantage: 223x - Used to show U.S. weakness relative to China in shipbuilding. Munitions on hand for major China conflict: 8 days - Referenced as current U.S. stockpile readiness in a major conflict scenario. Needed munitions stockpile: 800 days - Presented as the level needed for adequate readiness. Ukraine production/fighting mismatch: 10 years of production and 10 weeks of fighting - Used to argue the stockpile/factory model is insufficient. World War II mobilization timeline: 18 months - Cited as the time needed to ramp industrial production in wartime historically. Modern art movement funding: 7 billion dollars in 2026 dollars - Claimed as Soviet funding for anti-war/peace movements during Vietnam. Snowden-related abuse cases: less than 12 documented cases - Used to argue intelligence data collection was not broadly abused. Fairchild/Apollo integrated circuits share: 96% - In 1968, nearly all integrated circuits sold were sold to Apollo, illustrating government demand as a catalyst. Government-paid R&D cap at Fairchild: 4% - Bob Noyce reportedly kept government-funded R&D below this level.

Pivotal Quotes: "War is bad, categorically bad. That doesn't mean it's always avoidable." — Sean Sankar: On the philosophical question of whether war is good and why defense work exists. "Make it unthinkable to your adversaries that they should ever challenge you." — Trey Stevens: Describing the purpose of Anduril and modern deterrence. "We don't collect data. We don't have any data. It'd be like accusing Excel of being a surveillance tool." — Sean Sankar: Responding to criticism that Palantir enables a surveillance state.

Implications: The conversation frames defense tech as a patriotic, pro-democracy response to real geopolitical risk. For industry, it signals more capital, autonomy, and manufacturing investment; for policy, it argues that deterrence now depends on rebuilding industrial depth and faster procurement.