Palantir’s Unapologetic AI Bet: Skip the Ethics Debate, Focus on Winning
It’s a crisp, frigid March morning at an undisclosed mid-Atlantic hotel, where Palantir Technologies is holding its annual developer conference. Most attendees—defense contractors, active-duty military officers, and C-suite corporate leaders—arrived unprepared for the sudden cold snap; they’d expected the previous day’s mid-70s warmth to hold. A light cold rain thickens into steady, wet snow, and Palantir staff hand out heavy wool blankets to shivering guests. As attendees trudge between open-air event pavilions, they look more like survivors pulled from a shipwreck than conference guests.
Even so, energy runs high. For this self-selected crowd of supporters, Palantir is delivering on every promise it has made. The company’s stock price is soaring to new records, and the entire gathering hums with the giddy, insular groupthink of a top-tier multilevel marketing convention.
After navigating Palantir’s strict vetting to secure an invitation—made far more challenging by the company’s disapproval of WIRED’s recent coverage of its work—I was eager to peek inside one of Silicon Valley’s most secretive powerhouses. Founded in 2003 by Peter Thiel and his then-obscure former Stanford classmate Alex Karp, Palantir has grown into a core partner in the Pentagon’s AI-powered military transformation. Over the past few years, however, its fastest growth has come from the commercial private sector.
“Our commercial business is growing 120 percent year over year. We’re very proud of the 60 percent growth in government, but they're not even on the same glide slope,” says Palantir CTO Shyam Sankar, who is one of four top Palantir tech executives that serve as lieutenant colonels in the U.S. Army Reserve. (Shortly after this developer conference, the Department of Defense formally designated Palantir as an official Pentagon program of record, locking the company in as a permanent provider of weapons-targeting technology for the U.S. military.)
Generative AI has been the biggest fuel for Palantir’s recent rise, supercharging the hands-on, on-the-ground support the company has long offered clients. Early in its history, Palantir embedded “forward deployed engineers” directly into client organizations, helping them weave the company’s software into their core operations. Large language models allowed Palantir to build far more powerful tools, and today those engineers focus instead on helping customers build their own custom solutions on top of Palantir’s core technology.
“Every time those models got better it seemed like they were tailor-made exactly for us,” says Ted Mabrey, an early Palantir employee who now leads the company’s commercial division. Sankar elaborates on the generative AI shift: “Our whole thesis has been that we’re building Iron Man suits for cognition. We were rate-limited by the number of people, the creativity of the questions, all those sorts of things. And then [with Gen AI] that rate limiter was eliminated, and that changed the rate of growth.”
The morning’s keynote lineup includes a U.S. Navy vice admiral leading the Pentagon’s Project Maven AI battlefield initiative, plus executives from Accenture, GE Aerospace, SAP, and Freedom Mortgage Corporation. That range of speakers mirrors the company’s trajectory from pure defense work to a mixed commercial-defense business model. Over breakfast, I watch a presentation from a 450-person, family-run fashion brand. Jordan Edwards, CEO of Mixology Clothing, says he first found Palantir through an Instagram ad, and the AI-powered system has transformed his business. He uses Palantir’s software to inform inventory buying decisions, and even lets the tool draft and send emails to negotiate supplier prices. For one clothing line he sells, “it drove a 17-point margin swing—from losing $9 a unit to gaining $9 a unit,” he claims. Edwards now describes himself as a “forward deployed CEO,” leaning into Palantir’s signature internal terminology.
Even though Palantir’s fastest growth is in the commercial sector, its core identity remains rooted in defense contracting. During its years-long battle to break into the mainstream defense establishment—at one point, it even sued the U.S. Army to qualify for a major contract—it developed an unwavering focus on delivering measurable results. Palantir’s leadership argues this experience forced the company to adopt a level of rigor that lets it outperform all rivals in the commercial space. One chapter of Sankar’s just-published book, Mobilize: How to Reboot the American Industrial Base and Stop World War III, is titled “The Factory Is the Weapon.” Both Sankar and Karp believe that U.S. industry, especially Silicon Valley, has shown insufficient patriotism when it comes to national defense. They hope Palantir’s example will inspire other corporations to build national defense tools alongside their consumer-facing products.
Karp’s introductory conference remarks emphasized that defense work defines the company, especially now that America is engaged in active global conflict. Dressed unusually in a blazer (“This is to convince my family I have a job,” he jokes), he says that on a normal day, he would be talking to commercial customers about how Palantir can make them wealthier, happier, and help them outcompete their rivals. (He refers to other industry players as “noncompetition” because in his view, none of them rank in Palantir’s class.) But with active combat in Iran, the company’s sole priority right now is supporting U.S. troops. “At Palantir we were built to give our warfighters … an unfair advantage,” he says. “It was, ‘Yeah, we’re going to really fuck our enemies.’ And I take great pride in that.”
Karp claims Palantir’s culture is broad enough to accommodate disparate political views, with just one non-negotiable exception. “The one thing I tell Palantirians is you can be on any side of an issue, but if you’re expecting us not to support warfighters when they’re in battle, you’ve got the wrong company.” Now that the U.S. is at war, he says, “We’re not interested in debating. We are very proud to have our role in having American men and women come home safe. That sometimes means that people on the other side don’t go home.” (The remark came after at least 175 Iranian civilians died when a girls’ school was hit by a missile. The incident is under investigation and Palantir won’t comment on whether its products were involved.) Karp makes clear that if customers don’t share this core stance, they don’t belong as Palantir clients either. “You are engaging in proxy when you are engaging with us,” he says. His remarks were met with loud applause.
Karp never mentions AI company Anthropic by name, but his comments draw a deliberate contrast between Palantir and that firm, which was recently sanctioned by the Pentagon for attempting to set what it considers moral and practical limits on AI use in battle. To Palantir, that kind of limitation is itself immoral. When I mention I’m writing extensively about the broader AI industry to Sankar, he launches into a rant, arguing that the people who invent new technology are almost always the last to understand its real-world impact. The leaders of big AI companies, he says, have holes in their hearts where their moral core should be, and they are trying to fill that void with the promise of AGI. Sankar and Karp make no secret of their dismissal of the utopian, ultra-optimistic AI vision Dario Amodei laid out in his essay “Machines of Loving Grace.”
That is Palantir’s differentiator: an unapologetically jingoistic chip on its shoulder, and a belief that both virtue and business success lie in pushing AI forward to help America win. The company attributes its corporate success directly to this commitment. “There’s a gravity to the defense mission,” says Sankar. “You could ask, would we have ever conceived of [forward] deployed engineering if we didn't feel some sort of moral weight that our software has to fucking work?”
Instead of being a barrier to winning new customers, Mabrey says the company’s controversial reputation works as a useful filter, narrowing the field to those most culturally aligned with Palantir’s values. “We tend to have relatively fewer customers and relatively much deeper relationships with those customers,” he says. “We don’t come in and tell them what to do—and they don’t tell us what to do.”
Palantir also refuses to pass judgment on its government customers when its software is used for questionable purposes. When I ask Sankar why Palantir has continued to work with ICE after the agency’s violent surge of arrests in Minnesota, he says, “The specifics are a tragedy, but the ballot box and the courtrooms work. You have to make a very fundamental call—do you believe in the system or not?”
The snow was still falling when I left the Palantir conference site, returning to the outside world where the company is widely viewed with deep skepticism. Outside the conference bubble, a fierce debate is raging over how AI should and should not be used. Palantir has found energy and wealth by bypassing that entire conversation, instead devoting its complete attention to using AI to win. Loving Grace is for noncompetitors.
Update: 3/23/2026, 1:30 PM EDT: WIRED has updated the article to clarify when the Pentagon designated Palantir as an official program of record.
