Palantir CEO Alex Karp claims AI companies are stealing customers' data while charging them for unproductive tokens

Palantir CEO Alex Karp delivered a blistering critique of leading AI companies like OpenAI and Anthropic during a CNBC interview, accusing them of stealing customer data while charging for unproductiv... Read original →

Karp’s critique exposes a foundational flaw in today’s LLM business model: the inherent tension between enterprise data sovereignty and closed-loop model training. Technically, this accelerates adoption of Sovereign AI OS architectures, pushing NVIDIA to refine on-prem inference chips and forcing token-based billing models to account for ‘unproductive’ usage—potentially reshaping SLAs. Compliance-wise, certifications like ISO27001/17/18 and CMMC Level 2 will become non-negotiable procurement filters, inflating costs for third-party cloud AI. OpenAI and Anthropic may rush to offer air-gapped APIs, but their unit economics will suffer. Over the next 18 months, enterprises will favor platforms with robust Ontology governance; Palantir gains leverage in defense-linked supply chains, while generic-model startups face steeper capital hurdles. This is not just ethics—it’s the opening salvo in AI’s data ownership reckoning.