Skip to content
TopicTracker
From HackerNewsView original
TranslationTranslation

Palantir's Karp bashes OpenAI, Anthropic token model as completely wrong

Palantir co-founder Alex Karp criticized the business models of AI firms like OpenAI and Anthropic, arguing that charging by the token is fundamentally flawed and does not capture the true value of AI systems. He suggested that pricing should be based on outcomes rather than usage metrics.

Background

- Palantir is a U.S. data analytics company (co-founded by Peter Thiel) best known for its work with intelligence and defense agencies. Its CEO, Alex Karp, is an outspoken public figure. - "Tokens" are how modern AI models (like those from OpenAI and Anthropic) measure and charge for usage — roughly, a token is a chunk of text (about ¾ of a word). Customers pay per token consumed. - Karp is arguing that this token-based pricing model is fundamentally wrong for enterprise or government clients, who need predictable costs and mission-critical reliability, not variable bills based on how much text the AI processes. - This criticism reflects a broader industry debate: AI companies currently charge like utilities (per unit consumed), but big clients want subscription or outcome-based pricing instead. - OpenAI makes the GPT series of models; Anthropic makes the Claude series. Both are leaders in the current AI boom and direct competitors to Palantir's own AI platform offerings.

Related stories