How Alex Karp helped turn Palantir into the West’s AI arms dealer

Alex Karp, the chief executive of Palantir, knows his products can be dangerous. Built to extract insights from torrents of data with machine learning and AI, his company’s software is optimizing manufacturing and supply chains but also warfare and kill chains, helping target weapons in Ukraine and across the Middle East. To some, accelerated AI development is an existential threat worthy of a pause. Karp—a self-described socialist who earned his philosophy doctorate in Germany before Palantir chairman Peter Thiel tapped him to run the startup two decades ago—looks at those dangers differently. The way he sees it, not moving fast is the existential risk.

“Our goal as a company is to help the United States and its allies avoid war,” says Karp. “The only way to do that is to project such overwhelming technological and strategic superiority that we scare the daylights out of our adversaries.”

That includes building Maven, the battlespace decision-making system at the heart of the Pentagon’s AI efforts. (Palantir first picked up the contract after Google, prodded by employee protests, walked away from the project in 2018.) Early versions of the software have proved pivotal for surveillance and targeting over Ukraine, and have accelerated the U.S.’s ability to attack Houthi rocket launchers in and around the Red Sea. A targeting operation that required 2,000 people in 2003 now needs just 20.

“The AI revolution is one in which small groups can drastically outperform much larger groups,” Karp says. “Wars are now won on the basis of superior AI and electronic warfare.”

After years in the shadows, where it helped U.S. and foreign governments find terrorists and roadside bombs, Palantir has been expanding its reach on the battlefield and the homeland, too, helping wrangle data and deploy algorithms in all kinds of enterprises, from mining to healthcare, in order to connect dots, root out inefficiencies, and “bring your enterprise back to its principal purpose,” says Karp. According to the company, one large American insurer recently used 78 AI agents on Palantir’s Artificial Intelligence Platform to automate part of their underwriting process, shortening work that once took two weeks to three hours.

Both on and off the battlefield, Karp sees most of AI’s value accruing to those with preexisting scale, capital, and software power. And that bodes well for his company too.

“It’s like back to the future,” he says. “We’re at the forefront of managing large language models and AI, just like we were at the forefront of managing data.”

Palantir’s platform is designed to ground any number of AI models in an enterprise’s particular set of data, logic, and day-to-day operations. This meticulous approach to the “semantics” of an organization was initally developed to integrate piles of often highly sensitive data alongside older machine learning tools. But it also makes large language models useful, says Karp, by providing heightened transparency, security, and guardrails around models and outputs.

“It turns out that the software needed to augment human decision-making is the same software an LLM needs,” he says.

Wall Street—and a diehard online shareholder fanbase—seems to agree: Palantir’s stock price has exploded this year, to the point where Karp’s company—one with fewer than 4,000 employees—now boasts an ear-popping market cap of $174 billion, greater than defense giants like Raytheon or Lockheed Martin. After years fending off “the haters,” it’s vindication, he says. “There should be a reward for the best products in the world.”

Still, the critics haven’t retreated. Palantir’s unapologetic support for Israel’s defense establishment, its contracts with U.S. immigration enforcement agencies under Donald Trump, and Thiel’s politics have even led to unrest within the company. And many ethical and legal questions about AI-driven warfare and surveillance remain open. Karp says Palantir is focused on “not replacing human judgment, but augmenting it,” and he has called for rules to keep humans in the loop in critical domains, as well as for stronger personal data protections. The company has told investors it rejects clients “whose positions or actions we consider inconsistent with our mission to support Western liberal democracy and its strategic allies.” “All sorts of things I’m constantly refusing,” Karp insists, including, he says, pressure to work in China and Russia.

But nearly everywhere else, Palantir has been busy, especially since 2020, when the U.S. and the U.K. used the firm’s software to manage vaccine distribution. Meanwhile, Karp’s unabashed pro-West stance—and his takes on defense spending, wokeness, and the distractions of Silicon Valley—are winning more hearts and minds. His calls to upgrade the U.S. industrial base and make the Pentagon more efficient are certainly popular in the next White House, where even the vice president is a Thiel acolyte. (“I’m optimistic about where America is going,” Karp says of the incoming administration.) And his voluble insistence on defending democracy with software has helped fuel a defense-minded counterrevolution in tech, and a new crop of startups building killer AI.

“What other people view as courage I just view as telling the truth,” says Karp. “And I do think it has shifted how people at least see these things.”

This story is part of AI 20, our monthlong series of profiles spotlighting the most interesting technologists, entrepreneurs, corporate leaders, and creative thinkers shaping the world of artificial intelligence.