Palantir CEO Karp blasts AI industry as 'effing insane,' shares jump 9%

Alex Karp accused leading AI labs of overcharging customers and jeopardizing US national security in a blistering CNBC interview.

Palantir Technologies Inc. Chief Executive Officer Alex Karp called the artificial intelligence industry "effing insane" on Wednesday, accusing OpenAI and Anthropic of overcharging enterprise customers while collecting their proprietary data.

"These people are, Sam and Dario — there's nothing more fun than debating Dario in private, so I'm not throwing shade at them — but something has gone completely wrong," Karp said on CNBC's "Squawk Box," referring to OpenAI CEO Sam Altman and Anthropic CEO Dario Amodei. "The basic view among enterprises in this country is I'm going to chillax and waste my time with tokens, I'm gonna get no value, and they're gonna get my IP."

The outburst came as Palantir announced an expanded partnership with Nvidia Corp. to integrate Nvidia's Nemotron AI models into Palantir's Sovereign AI platform, allowing government agencies and enterprises to deploy AI in secure environments while retaining control over model weights and data. Palantir shares surged 9% on the news, adding roughly $9 billion to the company's market capitalization. Karp, worth $12.3 billion according to Forbes, co-founded the company with billionaire Peter Thiel and Stephen Cohen.

The criticism reflects a growing rift between enterprise customers and frontier AI labs over data ownership and pricing. As AI model costs escalate — with training runs now exceeding $4 billion — companies are demanding ownership over the infrastructure powering their systems rather than paying per-token fees that Karp likened to a "wealth tax."

Karp framed the issue as a national security concern, questioning whether the US should rely on Silicon Valley consensus for military AI applications. "Are we really going to outsource the battlefield of this country to the consensus view in Silicon Valley? That is effing insane," he said, adding that warfighters have "serious trust issues" with current AI providers.

The Pentagon has already taken steps to reduce reliance on major AI labs. In March, it designated Anthropic a "supply chain risk" after the company refused to remove restrictions preventing its technology from being used for mass domestic surveillance or fully autonomous weapons. Days earlier, the Pentagon reached a deal with OpenAI that drew criticism from AI policy experts. President Donald Trump issued an executive order in June requiring companies to allow federal oversight of new AI models before public release.

Open-weight models gain traction

Karp's solution centers on open-weight models that give customers control over compute, models, data and their competitive advantage — what he called owning "the means of production." The Palantir-Nvidia partnership directly addresses this demand, offering government clients a secure environment where model weights and proprietary data never leave their control.

This shift is pushing enterprises away from the "tokenmaxxing" mindset — maximizing token usage on frontier models regardless of cost — toward building and training their own proprietary tools. Chinese AI models are accelerating their capabilities, raising concerns that the US frontier labs could face competition from cheaper alternatives. Karp warned the industry should not underestimate the speed of China's AI progress.

Palantir shares, which have more than doubled over the past 12 months, trade at elevated multiples relative to traditional software peers. The Nvidia partnership provides a clear revenue catalyst tied to government AI spending, which is expected to grow as federal agencies seek secure AI deployment. However, Karp's broadside against OpenAI and Anthropic highlights the competitive tension in an industry where enterprise customers are increasingly questioning the value proposition of frontier models. For investors, the key question is whether Palantir's Sovereign AI platform can capture a meaningful share of the government AI market before rivals offer similar secure-deployment solutions.

This article is for informational purposes only and does not constitute investment advice.