Palantir CEO says something has gone completely wrong with Anthropic and OpenAI
Palantir CEO Alex Karp has criticised the token-based AI business model used by Anthropic and OpenAI, saying "something has gone completely wrong." He argues that enterprises now want lower AI costs, greater control over their data and customised AI models.
AI has become one of the biggest priorities for businesses, but it is also becoming increasingly expensive. As companies spend more on deploying advanced AI models, many are starting to question whether the current pricing approach makes financial sense. Palantir CEO Alex Karp believes this concern is growing among enterprise customers, and he has openly criticised the token-based business model followed by AI companies such as Anthropic and OpenAI.
Speaking to CNBC's Squawk Box on Wednesday, Karp said the industry's focus on charging customers based on token usage is creating frustration among businesses.
“I’m not throwing shade at them, but something has gone completely wrong,” Karp said. “The basic view among enterprises in this country is I’m going to chillax and waste my time with tokens.”
His remarks come at a time when running advanced AI models is becoming more expensive. As costs continue to rise with every new generation of AI systems, many enterprises are paying closer attention to whether their AI spending delivers meaningful business value instead of simply consuming more tokens.
This has also increased interest in open-weight AI models. These models allow companies to customise AI for their own needs without depending entirely on commercial AI providers. Since they can perform many tasks at a much lower cost, they are becoming an attractive option for organisations trying to reduce AI expenses.
According to Karp, businesses are also looking for greater control over the technology they use. Instead of relying completely on external AI companies, many customers want ownership of their computing infrastructure, AI models and data.
“What aligns me with Nvidia, and I think is what the technical customers want, which is control over their compute, their models, their data stack and their alpha,” Karp said. “They want to know they own the means of production. It’s not being transferred to someone else.”
Palantir has already started moving in this direction. Earlier this week, the company announced an expanded partnership with Nvidia to use the chipmaker's AI technology for building customised AI models for US government agencies. The collaboration focuses on creating AI systems that organisations can tailor for their own work instead of relying on one general-purpose model.
Karp also warned that the United States should not underestimate China's progress in artificial intelligence. He said Chinese companies are improving their AI models quickly, adding to competition in the global AI race.
At the same time, many businesses are choosing to develop their own AI systems instead of depending only on large, general-purpose models. These smaller proprietary models are designed for specific tasks, making them more efficient and less expensive to operate in many cases.
Just before Karp's interview, Palantir published a nine-point manifesto on X centred on what it calls "AI sovereignty." The company criticised the "tokenmaxxing" business model and encouraged organisations to keep ownership of their data rather than handing greater control to AI service providers.
Investors appeared to respond positively to Palantir's strategy. Shares of the AI software company rose 8 per cent on Wednesday. Karp's comments add to the ongoing discussion around the future of enterprise AI, as more companies weigh the benefits of expensive frontier AI models against customised systems that offer greater control and lower operating costs.
AI has become one of the biggest priorities for businesses, but it is also becoming increasingly expensive. As companies spend more on deploying advanced AI models, many are starting to question whether the current pricing approach makes financial sense. Palantir CEO Alex Karp believes this concern is growing among enterprise customers, and he has openly criticised the token-based business model followed by AI companies such as Anthropic and OpenAI.
Speaking to CNBC's Squawk Box on Wednesday, Karp said the industry's focus on charging customers based on token usage is creating frustration among businesses.
“I’m not throwing shade at them, but something has gone completely wrong,” Karp said. “The basic view among enterprises in this country is I’m going to chillax and waste my time with tokens.”
His remarks come at a time when running advanced AI models is becoming more expensive. As costs continue to rise with every new generation of AI systems, many enterprises are paying closer attention to whether their AI spending delivers meaningful business value instead of simply consuming more tokens.
This has also increased interest in open-weight AI models. These models allow companies to customise AI for their own needs without depending entirely on commercial AI providers. Since they can perform many tasks at a much lower cost, they are becoming an attractive option for organisations trying to reduce AI expenses.
According to Karp, businesses are also looking for greater control over the technology they use. Instead of relying completely on external AI companies, many customers want ownership of their computing infrastructure, AI models and data.
“What aligns me with Nvidia, and I think is what the technical customers want, which is control over their compute, their models, their data stack and their alpha,” Karp said. “They want to know they own the means of production. It’s not being transferred to someone else.”
Palantir has already started moving in this direction. Earlier this week, the company announced an expanded partnership with Nvidia to use the chipmaker's AI technology for building customised AI models for US government agencies. The collaboration focuses on creating AI systems that organisations can tailor for their own work instead of relying on one general-purpose model.
Karp also warned that the United States should not underestimate China's progress in artificial intelligence. He said Chinese companies are improving their AI models quickly, adding to competition in the global AI race.
At the same time, many businesses are choosing to develop their own AI systems instead of depending only on large, general-purpose models. These smaller proprietary models are designed for specific tasks, making them more efficient and less expensive to operate in many cases.
Just before Karp's interview, Palantir published a nine-point manifesto on X centred on what it calls "AI sovereignty." The company criticised the "tokenmaxxing" business model and encouraged organisations to keep ownership of their data rather than handing greater control to AI service providers.
Investors appeared to respond positively to Palantir's strategy. Shares of the AI software company rose 8 per cent on Wednesday. Karp's comments add to the ongoing discussion around the future of enterprise AI, as more companies weigh the benefits of expensive frontier AI models against customised systems that offer greater control and lower operating costs.
