Moonshots With Peter Diamandis: Palmer Luckey on Autonomous Weapons and Anduril

SPEAKER_1 (0:01)

When ambition meets urgency, the future starts to look less like a slogan and more like a systems problem. In this episode of Moonshots with Peter Diamandis, Peter sits down with Palmer Luckey, founder of Anduril, to explore autonomous weapons, wildfire response, and defense innovation.

Let's start with something that surprised a lot of people—Anduril's Wildfire XPRIZE win.

What's interesting is that Luckey says this technology has been possible for years. He puts it bluntly: "It is insane that we let fires become conflagrations before we put them out."

Right, but the challenge isn't just spotting a fire. He explains it's about detecting, classifying, rejecting alarms, matching the right asset, and acting fast enough to actually matter. It's a complete systems problem.

That same logic carries into defense. When Diamandis asks about AI risk, Luckey's answer is refreshingly direct: "AI doom is way down the list for me." He argues the real danger isn't AI deciding to kill people—it's bad actors using capable AI to do harmful things. He also thinks people are overreacting without enough real-world perspective. Now, on autonomy, Luckey makes a crucial point. This isn't a brand-new leap. "Autonomy needs to live forward, not back home," he says, pointing to communication delays, jamming, hacking, and the limits of remote control.

What's fascinating is his historical perspective. Autonomous weapons have existed for decades in forms like Aegis, C-RAM, landmines, and traps. The technology is changing, but the underlying idea is old. Building on that point, when Diamandis presses on foundation models and whether a general-purpose defense robot is inevitable, Luckey actually disagrees. In defense, specialization wins. A weapon built for diving, climbing, or tracking a specific target will outperform a generalist system. Unless you're imagining something extreme, like "a humanoid with a fusion reactor," as he puts it. That helps explain why Anduril won't build humanoids right now. Luckey says, "I want Anduril to build things that wouldn't exist otherwise." Humanoids may be useful interfaces for legacy systems, but he wants the company focused on the highest-impact gaps. What do you think about how they work with the Pentagon? Luckey describes Anduril as a product organization that builds first, then sells. That approach has earned trust because they finish on schedule and save money. He recalls the early pitch that Anduril could save taxpayers hundreds of billions. The conversation also widens to John Carmack and the responsibility of wealthy technologists. Luckey argues that capital and credibility should be used to solve hard problems. He says, "It is repulsive to do nothing with one's influence."

The episode closes on VR and entrepreneurship. Luckey praises Meta's new glasses for better ergonomics. And on AI, he's clear: it will speed up business, but it won't replace judgment. "AI does not change the core requirements of leadership or legal responsibility."