Palantir’s nine-point manifesto on AI sovereignty stresses retaining data as valuable and warnings against tokenmaxxing.

Palantir expanded its partnership with Nvidia to use Nvidia’s AI tools for building custom models for U.S. government agencies.

In a CNBC interview, he called the AI industry “fèffing insane,” arguing firms overcharge, exploit data, and threaten U.S. national security.

He said AI labs have oversold their models, describing the situation as completely, irresponsibly oversold.

Anthropic reports export controls on Claude Fable 5 and Mythos 5 were lifted by the Commerce Department after national security considerations and government collaboration.

Palantir CEO criticizes OpenAI and Anthropic for a token-based pricing model that delivers little enterprise value and risks exposing proprietary data to providers.

He warns China is rapidly advancing in AI and closing the gap with U.S. labs, underscoring the need not to underestimate this progress.

Some businesses are shifting from large general-purpose models to developing and training their own more cost-efficient models amid cost pressures.

Government scrutiny of AI models continues, with the Pentagon labeling Anthropic a risk in supply chains amid debates over deployments and export controls.

Palantir competes with major AI players for enterprise contracts by offering software that helps governments and corporations deploy AI securely within their own environments, highlighting AI sovereignty.

Karp argues against outsourcing national security and military tech to Silicon Valley, emphasizing control over proprietary data and avoiding over-reliance on external AI providers.

He warns that relying on Silicon Valley perspectives for national security decisions is dangerous and stresses maintaining domestic control.