Palantir CEO Alex Karp said at a company boot camp last week that “AI sovereignty” should shape how companies use artificial intelligence, arguing that economic value should not flow only to “2,500 people sitting in Silicon Valley.” The New York Post reported on August 26 that trust in AI companies remains weak, citing an Anthropic + YouGov poll from Dec. 2025 in which just 15% of Americans said they trust AI companies to make decisions about the technology’s development and use.

Palantir, a Miami-based software company that builds AI platforms for governments and private companies, is promoting an alternative to businesses sending corporate data to frontier labs such as Google DeepMind, OpenAI and Anthropic. Karp has described that dependency as creating “vassal states of the language labs,” while Kevin Kawasaki, Palantir’s Head of Business Development, described the company’s position as: “Your data, your model, not ours, not someone else’s.”

Palantir has pushed the sovereignty message over the last few months through a nine-point manifesto and a boot camp for customers seeking AI tools without giving away data. Lead Palantir Architect Chad Wahlquist said workers using AI, including nurses and factory floor workers, describe a different experience from the fear-focused accounts associated with some AI labs.

Palantir also says it wants to end “token-maxing,” where customers are pushed to use and pay for more tokens (units of text processed by AI systems). Wahlquist said companies such as OpenAI earn more when customers buy more tokens, while Palantir says it is paid on outcomes rather than consumption.

Palantir CEO Alex Karp said at a company boot camp last week that “AI sovereignty” should shape how companies use artificial intelligence, arguing that economic value should not flow only to “2,500 people sitting in Silicon Valley.” The New York Post reported on August 26 that trust in AI companies remains weak, citing an Anthropic + YouGov poll from Dec. 2025 in which just 15% of Americans said they trust AI companies to make decisions about the technology’s development and use.

Palantir, a Miami-based software company that builds AI platforms for governments and private companies, is promoting an alternative to businesses sending corporate data to frontier labs such as Google DeepMind, OpenAI and Anthropic. Karp has described that dependency as creating “vassal states of the language labs,” while Kevin Kawasaki, Palantir’s Head of Business Development, described the company’s position as: “Your data, your model, not ours, not someone else’s.”

Palantir has pushed the sovereignty message over the last few months through a nine-point manifesto and a boot camp for customers seeking AI tools without giving away data. Lead Palantir Architect Chad Wahlquist said workers using AI, including nurses and factory floor workers, describe a different experience from the fear-focused accounts associated with some AI labs.

Palantir also says it wants to end “token-maxing,” where customers are pushed to use and pay for more tokens (units of text processed by AI systems). Wahlquist said companies such as OpenAI earn more when customers buy more tokens, while Palantir says it is paid on outcomes rather than consumption.