Banks & AI: Palantir CEO's OpenAI/Anthropic Critique

Banks should pay close attention to Palantir CEO Alex Karp's recent critique of AI models from companies like OpenAI and Anthropic. Karp states that customers want control over their compute, models, and data. He says they are paying for tokens that create no value, and that proprietary model providers are "stealing the weights and alpha" of their business. This "wealth tax" comment highlights concerns about cost, control, and value. Banks are already weighing open-source versus closed-source models. The move from AI pilots to production makes this question more urgent. Bank executives need to build a tiered AI architecture that assigns the right model to the right job with the right control. They must consider which workloads need frontier proprietary models and which can use smaller, task-specific models. The key is not just cost, but accountability, especially as AI moves into production. Banks evaluate AI systems not only on technical performance but also on model risk, cybersecurity, and auditability. This matters because these are essential conditions for AI adoption in banking.