Palantir CEO Alex Karp recently ignited a debate in the AI industry, labeling it "effing insane" during a CNBC interview. He strongly criticized leading AI firms like OpenAI and Anthropic for charging exorbitant fees, which he called a "wealth tax" on businesses, and for allegedly exploiting customer data to improve their own models. Karp asserted that most CEOs he privately speaks with are "livid" about these practices. He also raised concerns about the U.S. government's reliance on these companies for military and national security technology, questioning the outsourcing of national defense to Silicon Valley's consensus view.
Karp's outspoken comments came as Palantir announced a partnership with Nvidia, aimed at helping the U.S. government more securely deploy advanced AI. He argued that enterprises are tired of paying for "tokens that create no value" and of the potential transfer of their valuable intellectual property. Instead, Karp advocated for enterprises to maintain control over their compute resources, models, data stacks, and "alpha," emphasizing trust and data ownership as critical issues. His perspective aligns with Palantir's strategic positioning as a provider of secure, enterprise-friendly AI solutions that keep sensitive data within the customer's perimeter.
The Palantir CEO highlighted that technical customers desire ownership of their GPUs, data, and models to control their business's "alpha." He believes that current AI model leaders are over-selling their capabilities, leading to a "dis-comfort and loss of trust" among enterprises. Karp stressed the importance of rebuilding trust by providing clear answers on data ownership, caching, and prompt security. His critique suggests that the market needs to re-evaluate the long-term profitability of frontier model providers due to potential pricing compression from model routing, open models, and cheaper inference options. Following his interview, Palantir's shares saw a jump of more than 9%.
