Palantir CEO says enterprise clients 'unhappy' with AI labs' tokenmaxxing

Palantir CEO Alex Karp said enterprise customers are "unhappy" with how the frontier AI labs operate.

He told CNBC's Sara Eisen on Wednesday that the discontent goes beyond the general public. "It's not just the man and woman on the street that is unhappy with the frontier labs, it's in private every single enterprise we deal with."

Many customers believe these labs don't understand their businesses and care mainly about "tokenmaxxing," burning through AI tokens to signal productivity, Karp said.

The critique arrives. AI costs are drawing more scrutiny on Wall Street. Accelerating spending on large language models has raised efficiency concerns. Businesses are funneling more of the technology into workloads, Karp noted.

Competition between Anthropic and OpenAI has reached a peak. Karp spoke in that context. OpenAI said it confidentially filed for an initial public offering a week after Anthropic.

Over the next seven years, Karp said he sees more value in AI implementation than in the large language models themselves. The tokenmaxxing critique, in his view, points to a disconnect between AI lab priorities and enterprise needs. He said the tension reflects a broader challenge in AI deployment: turning general-purpose models into production systems that fit specific workflows.

No date has been set for OpenAI's IPO.