As the cost of the deployment of AI by the enterprise continues to rise, the controversy surrounding the large model fee approach is rising. In a recent interview with CNBC, Alexander Karp, CEO Palantir, named OpenAI and Anthropic for criticizing the token billing model used, saying that many corporate executives were already dissatisfied with the rapid rise in related expenditures.
Enterprises value input output more
According to Karp, the patientness of business clients with regard to the pattern of payment in token currency, with continued additional expenditure, is declining. As the price of the new generation model is higher than in the earlier version, the thinking of Enterprise Procurement AI is also changing, with a shift in focus from seeking larger calls to assessing real returns.
In his view, that change was driving some enterprises to reduce their reliance on closed models and to seek lower-cost alternatives. For many agencies with greater technical capacity, it is more attractive to have model systems that capture data and control deployment patterns.
Increased demand for open weight models
Karp stated in an interview that the open weight model was becoming a viable option. Such models could, in part, do work close to head laboratories at lower cost and would be more appropriate for enterprises to customize around their own operations.
He mentioned that a growing number of companies no longer relied on generic models with a wide range of coverage, but began to build and train more efficient self-owned tools. Data attribution and systematic control are becoming more important for enterprises and government agencies.
Paratir expanded cooperation with Nvidia
Palantir announced earlier this week the expansion of cooperation with Nvidia and plans to develop customized models for United States government agencies using the latter ' s AI tool. Karp sees this direction as a path for businesses and government clients to cope with high-cost pressures.
He also stated that the industry should not underestimate China ' s rate of advancement in the development of AI models. As China ' s model capacity continues to grow, so does the competitive pressure on head AI laboratories in the United States.