Palantir Technologies chief executive Alex Karp has told governments they must persuade their leading technology experts working in the United States to return home and directly shape national artificial intelligence policy, warning that the AI race is no place for political or theoretical debate. Speaking on September 3 during a fireside chat in Chapel Hill, North Carolina, with US Commerce Secretary Howard Lutnick, Karp said ministers and regulators lacked the technical depth required to decide which AI models, software stacks and computer chips their countries should adopt while protecting sensitive data and intellectual property.
"You need to get them to come home to your country and sit next to you and explain what regulations, in your terms, what your regulations look like, what should the stack look like, what model should we use, what software we should use, what chips we should buy," Karp said. He argued that the people capable of making those calls were overwhelmingly concentrated in the US, and that national leaders had to issue a direct message to their diaspora technologists: "You are now going to do this for my country, and you're gonna set the policy in my country." His remarks carry particular weight for technology-rich nations such as India, which have produced large pools of executives, engineers and researchers now employed at leading American companies.
ALSO READ | Apple’s First Foldable iPhone Duo Set for Tonight’s Launch with Leaked Color Lineup and Pricing
Karp pushed back against the common practice of announcing large AI funding commitments without addressing who controls the spending. "You don't need money. You need knowledge," he said. "I don't wanna hear how much money you spend. Who's gonna get the money? Like, the money doesn't matter as much as who gets it." He warned that poorly architected AI deployments could effectively transfer a country's proprietary knowledge to foreign model providers, arguing that governments, intelligence agencies and private firms needed application layers that could train models, control data flows and keep sensitive information within protected environments.
Defining AI sovereignty as the ability to control the weights, data and knowledge embedded in artificial intelligence systems rather than exporting them to third parties, Karp said the moment for action had arrived. "This is really game time," he said, cautioning that bad implementations of AI were "a joke" and might even be "migrating your data to someone else," while good ones were not. Countries, he added, would have to map the full AI technology stack — computing capacity, data management, models, software and electricity — and decide which layers could be built domestically and which would continue to depend on American or other foreign suppliers. "You need access to compute, you need access to your own data," he said. "So the stack is compute, data, ability to manage data, ability to control the models, ability to train the models, and ability to compete against certain competitors and adversaries."
Karp reserved some of his sharpest criticism for AI regulations drafted without sufficient technical input, saying companies that simply deployed large language models could find themselves handing valuable information to model providers even while governments believed they were tightening control. "This is not sovereignty. This is anti-sovereignty," he said, adding that advisers promoting such approaches "need to be fired because they are hurting your country." Palantir, which Karp co-founded with Peter Thiel and others in 2003, is headquartered in Denver, Colorado, and builds data-integration and AI software used by US government agencies, allied governments and commercial clients across defence, intelligence, supply-chain and analytics operations.
