Palantir CEO Alex Karp said enterprise customers are increasingly unhappy with frontier AI labs, arguing that businesses feel the leading model companies do not understand their needs and are too focused on driving usage of their systems.
Speaking with CNBC on Wednesday, Karp said the frustration extends well beyond ordinary consumers and is widespread among the companies Palantir works with. He said many enterprises believe the labs care more about maximizing token usage than about solving practical business problems.
Karp's comments come at a time when the largest artificial intelligence companies are moving closer to public markets. OpenAI said Monday that it had confidentially filed for an initial public offering, about a week after Anthropic took a similar step. The two firms are among the most closely watched players in the AI industry.
Karp said that although large language models remain important, the real value for businesses will come from how the technology is put to use. In his view, implementation will be the key differentiator over the next several years.
He also said many of Anthropic's publicly discussed projects are operating on Palantir's infrastructure. That claim underscores the software company's role as an enterprise technology partner in the AI ecosystem, even as the model companies draw most of the public attention.
Rising costs have become a growing concern as companies deploy more AI across their operations. Karp's remarks reflect a broader Wall Street debate about whether heavier use of AI will translate into lasting productivity gains or simply higher spending as token consumption increases.
Frontier AI labs, which build the most advanced foundation models, have been under pressure to show that their products can deliver clear returns for corporate customers. Businesses adopting those tools often face questions about cost control, integration, and whether the systems can be tailored to specific workflows.
Karp has often taken a forceful public stance on technology and politics, and his comments on Wednesday also touched on the wider debate around AI's role in society. He said the technology will shape some of the most important political decisions in the United States and warned against treating it as a simple partisan issue.
He described AI as a major transformation with both opportunities and risks, saying the U.S. is uniquely positioned to benefit from it. At the same time, he cautioned that the scale of the shift makes public discussion especially important.
Karp also repeated his support for Anthropic co-founder Dario Amodei, whom he called an important figure in the sector despite their disagreements. That view suggests Palantir sees strategic value in the broader frontier model ecosystem even as Karp criticizes how some of its leading companies engage with enterprise users.
The interview also comes against the backdrop of Karp's increasingly visible political alignment with President Donald Trump's administration, a shift that has drawn scrutiny inside and outside Palantir. The company's communications chief previously described that change as concerning, highlighting tensions around Karp's public profile.
For now, Karp's message was clear: enterprise buyers may still need frontier models, but they are growing impatient with the current economics and the way those systems are being sold. In his telling, the next phase of AI competition will be won not just by better models, but by better execution inside the companies that use them.