
Palantir CEO Alex Karp publicly criticized the token-based pricing models used by Anthropic and OpenAI, saying costs have become unreasonable and enterprises are losing interest. As AI prices rise, businesses are shifting toward cheaper open-weight models or building their own AI tools, and Karp argues they should prioritize owning their data and compute rather than depending on expensive third-party AI labs.
Summaries like this, in your inbox every morning.
Sign up free →What happened
Palantir CEO Alex Karp criticized the token-based pricing used by Anthropic and OpenAI on CNBC, saying "something has gone completely wrong" as AI costs surge. He noted that enterprises are shifting away from what he calls "tokenmaxxing" toward a focus on return on investment, and many are turning to open-weight models (models whose code is publicly available) that can perform similar tasks at a fraction of the price.
Why it matters
As AI model costs rise and new versions become more expensive than earlier ones, businesses are reconsidering how they adopt AI. Some are moving toward building their own proprietary tools or using open-weight models instead of relying on expensive commercial AI labs. Karp argues that enterprises want ownership and control—of their compute, models, and data—rather than dependence on third-party AI providers. Chinese AI models are also advancing rapidly, which adds competitive pressure on U.S. labs.
What to watch
Palantir announced an expanded partnership with Nvidia to build custom AI models for U.S. government agencies. Karp also released a nine-point manifesto on "AI sovereignty" on X, emphasizing the importance of companies maintaining data ownership and rejecting the token-based business model. Palantir's stock climbed 8% on the day of Karp's remarks.
No comments yet. Be the first to share your thoughts!
Log in to join the discussion





Get curated AI news from 200+ sources delivered daily to your inbox. Free to use.
Get Started FreeFree · takes 30 seconds · unsubscribe anytime
1 minute a day. The AI essentials.
200+ sources · Email / LINE / Slack