Palantir's Karp bashes OpenAI, Anthropic token model: 'Something has gone completely wrong'

During a recent interview, Alex Karp, CEO of Palantir, expressed his concerns regarding the token pricing model employed by leading AI laboratories, including Anthropic and OpenAI. Speaking on CNBC's "Squawk Box," Karp remarked, "I'm not throwing shade at them, but something has gone completely wrong." He emphasized that as operational costs for AI technologies continue to soar, many enterprises are reassessing their strategies. Karp noted a significant shift in mindset among businesses, moving away from what he termed 'tokenmaxxing'—a focus on maximizing token usage—and towards a more pragmatic approach centered on return on investment. This transformation is leading companies to explore open weight models, which can deliver similar functionalities at lower costs. Moreover, Karp highlighted the rapid advancements being made by Chinese AI models, which are raising alarms about their potential to rival U.S. capabilities. He urged industry stakeholders not to underestimate China's swift progress in AI development. In light of these developments, many businesses are opting to create and train their proprietary AI tools rather than relying on extensive models provided by external labs. In a related announcement, Palantir revealed a strengthened partnership with Nvidia to leverage the latter's AI technologies in crafting bespoke models for U.S. government agencies. Karp believes that open weight models could offer a viable alternative for CEOs seeking greater control over their AI assets, stating, "What is happening among the most technical players is they're saying, 'I want something I own. This is my business.'"