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

Original Media · 1h 09m · English · Podcast · All-In with Chamath, Jason, Sacks & Friedberg

TLDR

Anduril co-founder Trey Stephens and early Palantir employee Sean Sankar discuss the collapse of America's defense industrial base, the rise of new defense tech companies, and the urgent need to rebuild manufacturing capacity before a potential 2027 Taiwan conflict. They argue that deterrence depends on production speed not stockpiles, that Silicon Valley's anti-defense culture was fueled by foreign adversaries, and that abstaining from defense work is itself a moral choice.

For you

PRO members can sign in and request more information — or their own version of this digest, tuned to what they care about.

Sign in to unlock

Surprising Insights

•

Google's revolutionary 'Attention Is All You Need' paper (2017) was motivated by a mere 3% improvement target for Google Translate — showing how banal commercial needs drive breakthrough innovation.

•

The US had 51 major defense contractors in the early 1990s; a Clinton-era 'peace dividend' dinner explicitly encouraged consolidation down to 5-6 primes, creating today's brittle monopsony.

•

80% of active pharmaceutical ingredients for generic US drugs come from China — a strategic vulnerability where a conflict could force choosing between national defense and treating a child's ear infection.

•

Palantir doesn't collect or own any data; it's an analytics layer on customer data with cell-level audit trails, making it 'the most insane platform to try to do something illegal in because you will be caught.'

•

Fully autonomous weapons already exist (e.g., Navy CIWS/SeaWiz) with human accountability baked in — the ethics debate is decades late to systems already deployed.

Key takeaways

•

The US defense industrial base atrophied after the Cold War: 86% of weapons spending now goes to pure-play defense specialists vs 6% in 1989, and only Tesla has built at-scale manufacturing this century.

•

Deterrence comes from factory throughput not stockpiles — Ukraine burned through 10 years of munitions in 10 weeks, revealing the US cannot regenerate inventory fast enough.

•

Anduril's Arsenal One factory in Ohio uses a contract-manufacturer model with modular lines to pivot between products (Fury, Roadrunner, Barracuda) based on demand, avoiding the stinger/javelin restart problem.

•

Venture capital in defense tech follows a power law: most new entrants will fail, and discipline requires climbing down valuation multiples each round rather than chasing peak pricing.

•

The anti-defense culture in Silicon Valley was partly manufactured: the Soviets spent $7B (2026 dollars) funding Vietnam-era protests, and today CCP money flows to groups protesting Palantir and Anduril.

Notable quotes

“War is bad. Categorically bad. That doesn't mean it's always avoidable. And that, you know, there will, there are people who will want to use might to make right, to define a set of rules. And you have to be in a position to protect your people and your interests accordingly.”

▸ Notable quotes · 3:40

“We thought the stockpile was going to deter our adversaries. It was always the factory. It was the ability to generate and regenerate the stockpile.”

▸ Notable quotes · 8:52

“You cannot think of something more banal leading to something more revolutionary.”

▸ Notable quotes · 33:13

“I don't think abstention from participating in the building of technology for national security is a morally neutral decision. You are making a moral decision when you decide to abstain.”

▸ Notable quotes · 50:19

“The Soviets spent $7 billion in 2026 dollars funding the peace movement, the anti-war protests. Now, there's obviously some there is an organic element to it, but this is just dumping gasoline on the fire to sow division and discord.”

▸ Notable quotes · 59:38

Chapters

•

0:00 — Origins: Palantir and the Anduril Connection: Trey Stephens (Anduril co-founder) and Sean Sankar (Palantir employee #13) recount meeting at Palantir, where Trey wore a CIA-cufflinked suit to interview at a t-shirt startup, and how Palantir's early government work seeded Anduril's founding.

•

3:00 — Philosophy of War and Deterrence: Both agree war is categorically bad but sometimes unavoidable; the goal is deterrence through overwhelming capability so adversaries never challenge you. They reject the 'military-industrial complex' profit motive narrative — primes trade at <2x revenue.

•

6:00 — Silicon Valley's Anti-Defense Turn: Post-Cold War peace dividend and globalist mindset made defense taboo in SV. The 2022 Ukraine invasion forced a reckoning. Many protesters held Ukraine flags while opposing defense work — a policy/understanding mismatch.

•

7:00 — Collapse of the US Industrial Base: WWII was won by a dual-use industrial base (Chrysler made ICBMs, General Mills made torpedoes). After 1991, consolidation to 5 primes and offshoring gutted manufacturing. Only Tesla has built new at-scale factories this century.

•

12:00 — Anduril's Arsenal One: Modular Manufacturing: Arsenal One in Columbus, OH (5M sq ft) uses a contract-manufacturer model: modular lines can pivot between Fury, Roadrunner, Barracuda. Avoids the Stinger/Javelin problem where lines and workers vanished after Cold War.

•

16:00 — Business Model: Product vs Cost-Plus: Anduril invests private R&D and sells outcomes (product model), unlike primes who build to government specs (cost-plus). This enables price-performance curves like NVIDIA or SpaceX — impossible under cost-plus.

•

20:00 — Venture Capital and Power Laws in Defense: Defense innovation follows a venture power law. Peanut-buttering capital across many bets fails; concentration on winners is needed. Anduril climbs down valuation multiples each round for discipline heading to IPO.

•

30:00 — Hardware vs Software in Government Sales: Government understands hardware BOMs and margins; software valuation was opaque. Hardware was an easier 'crossing of the Rubicon' for Anduril. Software still struggles with cost-plus procurement.

•

35:00 — Procurement Reform and Consumable Munitions: Current admin reimagines munitions as consumables with exercise/expenditure plans creating sustained demand signals. Shift from rigid requirements to outcome-based buying with program manager autonomy.

•

38:00 — Leadership Over Politics in Reform: Reform depends on heretical leaders (Kelly Johnson, Rickover, Boyd), not party. Bureaucracy entropy grows rules from 7 to 2000 pages without bushwhacking. F-35 spread across 400 districts = political project, not military.

•

45:00 — Autonomous Systems and Ethics: Fully autonomous weapons exist (SeaWiz CIWS). Accountability stays with human commanders. AI improves precision/discrimination — ethically superior to dumb bombs. Abstention from defense tech is a moral choice, not neutral.

•

50:00 — Palantir Surveillance Accusations: Palantir collects no data; it's an analytics layer with cell-level audit trails. Critics confuse patriotism with surveillance. Better tools for law enforcement (traffic cameras, tax compliance) are desirable, not dystopian.

•

55:00 — Cultural Divide and Foreign Influence: Stanford students rarely have military family connections. Vietnam-era schism between academia and defense never healed. Soviets funded anti-war movement ($7B); today CCP funds protests against Palantir/Anduril.

•

60:00 — 2040 Scenarios: Reindustrialization or Vassalage: Wrong path: Chinese century, US vassalage, zero-sum sabotage (e.g., smuggling crop funguses). Right path: massive US reindustrialization, thriving middle class believing children's future > present, restored institutional legitimacy.

Speakers

•

Host — All In Podcast host (likely Jason Calacanis)

•

Trey Stephens — Co-founder, Anduril Industries; former Founders Fund investor

•

Sean Sankar — Early Palantir employee (#13, 2008); Anduril executive

⋮