Speaking to CNBC on Wednesday, Karp said, “I’m not throwing shade at them, but something has gone completely wrong.” He added, “The basic view among enterprises in this country is I’m going to chillax and waste my time with tokens.”
His remarks come as the cost of running frontier AI models continues to rise, prompting enterprises to shift focus from maximising token usage to prioritising returns on investment and exploring lower-cost open-weight or in-house AI models.
China catching up, businesses turn to in-house AI models
Palantir CEO also cautioned against underestimating China’s rapid progress in AI, saying the country is advancing faster than many in the industry acknowledge.
Karp said, talking to CNBC, that businesses are increasingly moving away from relying on large, general-purpose AI models. Instead, they are building and training smaller, proprietary models tailored to their own needs.
Open-weight models and AI sovereignty gain traction
Karp said he sees open-weight models as a potential solution for CEOs frustrated by AI labs.
As token costs surge, companies are moving to adopt open-weight models, which offer a viable alternative with similar results at a fraction of the price.
Earlier this week, Palantir expanded its partnership with Nvidia to use the chipmaker’s AI software and computing platform to develop custom AI models for US government agencies.
“What aligns me with Nvidia, and I think is what the technical customers want, is control over their compute, their models, their data stack, and their alpha,” Karp told CNBC.
