Palantir CEO Alex Karp's critique of OpenAI and Anthropic has sparked a debate among banks about the cost and control of AI models. His 'wealth tax' argument highlights the tension between proprietary and open-source solutions.
TL;DR
- Palantir CEO Alex Karp criticizes OpenAI and Anthropic for charging enterprises a 'wealth tax' by taking proprietary data and workflows.
- Banks are evaluating the trade-offs between proprietary and open-source AI models, focusing on cost, control, and accountability.
What happened
Palantir CEO Alex Karp appeared on CNBC's Squawk Box on July 1, arguing that enterprises pay a 'wealth tax' to AI providers like OpenAI and Anthropic. He claimed that these companies take proprietary data and workflows in addition to charging fees.
Karp's comments have resonated with banks, which are weighing the pros and cons of open-source and closed-source models as AI usage moves from pilots to production.
Why it matters
For banks, the strategic task is to build a tiered AI architecture that assigns the right model to the right job, with the right level of control. This involves evaluating which workloads demand frontier proprietary models and which can be handled by smaller, task-specific models.
The accountability gap is a significant concern. Banks must consider model risk, cybersecurity, data lineage, third-party risk, auditability, explainability, and supervisory expectations when evaluating AI systems.
Key facts
- Alex Karp, CEO of Palantir, criticized OpenAI and Anthropic for charging a 'wealth tax' to enterprises.
- Karp's comments were made during an appearance on CNBC's Squawk Box on July 1.
- Banks are evaluating the trade-offs between proprietary and open-source AI models.
- The accountability gap is a significant concern for banks adopting AI.
- OCC Bulletin 2026-13, issued in April, updated the model-risk management framework but does not cover generative AI and agentic AI.
- Proprietary providers often offer documentation, security reviews, contractual protections, implementation support, and a more established enterprise-control environment.
- Open-source models can create documentation challenges and require investment in computing resources, security controls, model monitoring, MLOps capabilities, and specialized engineering talent.
Context
The debate between proprietary and open-source AI models is not new, but Karp's 'wealth tax' argument has brought new urgency to the question as AI usage moves from pilots to production.
Banks must balance the promise of AI with the need for accountability, risk management, and regulatory compliance.
