Palantir chief executive Alex Karp went public with sharp criticism of AI labs, saying enterprise customers are angry about paying for AI tokens that deliver no real value. He made the remarks during a nearly 20-minute appearance on CNBC.
"At every single enterprise I deal with, these people are livid, they're like, 'I am paying for tokens that create no value,'" Karp said during the interview.
Karp described himself as channelling "the voice of American business" rather than expressing personal anger. Anchor Becky Quick told him mid-interview: "You sound pretty angry."
The White Paper and the Core Argument
Palantir followed the TV appearance with a white paper titled "Institutional Sovereignty in the Age of AI." It laid out 15 steps that companies and governments could take to protect their data and decision-making from AI labs like OpenAI and Anthropic.
The paper focused on a concern growing across the tech industry — that AI labs learn from their customers' data and processes, then use that insight to move into the same markets those customers operate in.
Former White House AI adviser David Sacks amplified the argument on social media. He pointed to Anthropic expanding into categories like code, legal and security tools, saying the pattern was consistent: "Watch where value is being created, then move in directly."
Other CEOs Join the Conversation
Karp was not the only major tech figure raising these concerns. Microsoft CEO Satya Nadella wrote an essay on the topic and spoke at Stanford University this month.
"If you're just a consumer of a foundation model, then I'm not sure how you can retain enterprise value, let alone create," Nadella said.
Microsoft has been a major backer of OpenAI but is facing pressure as AI threatens its core software business. Reports that Starbucks is using AI to replace software from Microsoft and IBM briefly hit both companies' shares last week.
Meta CEO Mark Zuckerberg also weighed in, announcing a new paid tier for Meta's AI model. He said competitor pricing has "very high margins" and that Meta sees room to offer frontier AI at lower cost.
Neither OpenAI nor Anthropic responded directly to Karp's criticism. One AI lab insider told the Wall Street Journal: "We would be fools reacting to Karpian theatrics."
Palantir's own product sits between AI foundation models and enterprise customers, which gives Karp a commercial interest in the argument. He acknowledged as much during the interview.
Both OpenAI and Anthropic are expected to go public, with valuations that would place them among the most valuable companies in the world.
