Palantir Technologies CEO Alex Karp accused OpenAI and Anthropic of overselling token-based AI services that he said often produce little measurable value for large companies. Key Points: Kar
Palantir Technologies CEO Alex Karp accused OpenAI and Anthropic of overselling token-based AI services that he said often produce little measurable value for large companies.
- Karp said enterprises are paying for AI tokens without receiving results that justify the expense.
- He argued that providers may also gain access to proprietary knowledge and competitive insights held by their customers.
- The criticism reflects a wider debate over whether AI pricing should track usage or business outcomes.
Karp made the comments during a CNBC interview, where he challenged the economics behind charging companies according to the number of tokens processed by an AI model.
“Why are they charging for tokens, if it is so valuable?” Karp said, arguing that providers should instead earn a share of the value their systems create for customers.
He also said enterprises risk surrendering proprietary knowledge when they connect internal data and workflows to outside models. Karp described companies as increasingly frustrated after spending heavily on AI tools without seeing corresponding productivity gains.
The criticism directly serves Palantir’s commercial position because the company sells software designed to combine models, data and operational systems inside controlled environments.
Still, his remarks reflect a broader corporate concern about costs, data governance and weak returns from early AI deployments.
Also Read:Bitget Launches Cash Plus To Pay Yield On Idle Stablecoin Balances
Karp’s argument centers on the gap between technical output and business results. Token billing measures how much text or data a model processes, but it does not show whether the work increases revenue, lowers costs or improves decisions.
That distinction matters as corporate buyers become more selective.
Sam Altman has acknowledged that AI spending and efficiency are now central concerns for executives, while companies are testing cheaper models and routing work among several providers.
Karp has proposed outcome-based pricing, under which an AI vendor would receive compensation tied to measurable gains. That approach could reduce spending on unproductive usage, although it would require companies and vendors to agree on how value is calculated.
The dispute follows years of rapid enterprise adoption driven by expectations that general-purpose models would transform routine work. As pilot programs mature, buyers are shifting attention from model capability to ownership, security and documented returns.
Read Next:Stablecoin Market Sheds $10B From May Peak As Crypto Liquidity Contracts
