Microsoft, Meta executives warn AI labs risk losing business trust
Palantir Technologies CEO Alex Karp, describing himself as a “madman” during a nearly 20-minute CNBC appearance, ignited debate across the tech industry this month by arguing that enterprise customers are growing skeptical of frontier AI companies such as OpenAI and Anthropic. The interview, in which anchor Becky Quick said “You sound pretty angry,” quickly drew attention on social media and from other technology executives.
Karp said the AI labs are overhyping their capabilities and overcharging for tokens — units of AI consumption — while collecting data from customers that could be used to improve their own models. “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.
Palantir followed the appearance with a dense white paper titled “Institutional Sovereignty in the Age of AI” that laid out 15 steps companies and governments could take to protect themselves from the power and data access of AI labs. The company has a product that sits on top of foundation models and bridges the gap between AI and customers — a position Karp acknowledged gives his company a financial stake in the debate.
David Sacks, a former White House AI czar, amplified Karp’s argument on social media. “Anthropic has launched Claude Science, Claude Security, Claude Legal, and of course Claude Code — each expanding into categories previously served by companies building on top of their models,” Sacks posted. “The pattern is consistent: Watch where value is being created, then move in directly. Dominate the model layer, then use that position to capture the most lucrative verticals.”
Rival tech leaders voiced similar worries in recent weeks. Microsoft CEO Satya Nadella, whose company is a major backer of OpenAI, wrote an essay on the subject and said during a Stanford University appearance that companies must be able to retain the learnings that come from using AI models. “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.
Meta CEO Mark Zuckerberg told Bloomberg News that the pricing from other AI labs “is very extreme and has very high margins” and argued for his company’s ability to offer “frontier or very high-level intelligence at a much more affordable cost.” Meta announced a new version of its own AI model that includes a paid tier this past week.
Neither OpenAI nor Anthropic directly responded to Karp’s criticisms. Both have policies stating that enterprise customer data are not used for training their models. One AI lab insider said: “We would be fools reacting to Karpian theatrics; he does well talking his own book.”
The tension has manifested in other recent conflicts. The Trump administration last month put a hold on Anthropic’s ability to export new powerful models after Amazon CEO Andy Jassy raised concerns privately. Amazon is an investor in Anthropic and has benefited from the AI lab using its chips.
Karp acknowledged his own angle. “This is the voice of American business that is being channeled through me,” he said.