Lashes out at big AI model giants! Palantir CEO: Their business models have massive problems, "taxing" businesses while ignoring enterprise risks.

Palantir CEO Alex Karp launched a fierce attack on frontier AI labs like OpenAI and Anthropic, directly pointing to fundamental flaws in their business models and warning that enterprise customers are paying ever-increasing token costs for limited value while putting their core data and intellectual property at risk.

On Wednesday, Karp spoke sharply in an interview with CNBC, calling the business model of top AI labs "absolutely insane," and accusing Anthropic and OpenAI of misleading enterprise partners—exaggerating AI risks while selling their most powerful models to global businesses and governments.

He characterized this phenomenon as a "wealth tax" forced on enterprises, essentially handing over their competitive advantages to third parties. When CNBC host Becky Quick said, "You sound angry," Karp responded: "This is the voice of American business, channeled through me."

These remarks reflect growing dissatisfaction among enterprise customers with U.S. AI labs. High costs, questionable return on investment, and mounting regulatory pressure have prompted some large U.S. companies to turn to cheaper Chinese alternatives, putting OpenAI, Anthropic, and others under multiple pressures.

"Something is fundamentally broken"

Karp took direct aim at the token-based pricing model used by Anthropic and OpenAI. "I'm not picking on them, but something is fundamentally broken," he said. "Enterprises in this country generally believe, I'm just burning tokens, wasting resources."

He argued that while AI labs exaggerate risks and aggressively market powerful models, enterprise customers are effectively paying a "wealth tax" to these platforms, transferring their "alpha"—core competitive advantages—to external parties. On national security, his tone was tougher: "Are we really going to outsource the battlefield of this country to the mainstream consensus of Silicon Valley? That's absolutely insane."

Nine-point "AI Sovereignty" manifesto

Karp's public remarks are not an isolated incident. The day before the interview, Palantir posted a nine-point "AI Sovereignty" manifesto on X, systematically laying out its critical stance on the current AI ecosystem.

The manifesto's core logic revolves around "sovereignty": institutions that give up control over their data, model weights, and competitive advantages are handing over their future choices to others who are likely to profit from them. The manifesto states clearly that the "tokenmaxxing" incentive mechanism produces one-shot scripts rather than robust software systems, leading to an "addiction to false progress"—"Those selling tokens refuse to charge by value, and there's a reason for that."

The manifesto also emphasizes that controlling model weights means controlling institutional destiny. "Weights are the essence of knowledge an institution has painstakingly accumulated. Allowing others to control your weights means allowing them to migrate your business alpha into their hands."

A microcosm of the corporate "rebellion"

Karp's criticism is not an isolated case but a concentrated reflection of broader dissatisfaction among enterprise customers with U.S. AI labs. High usage costs, hard-to-quantify actual returns, and an increasingly complex regulatory environment are collectively eroding enterprise trust in mainstream AI platforms.

Some large U.S. companies have begun looking toward Chinese AI alternatives with greater cost advantages, a trend that puts direct commercial pressure on organizations like OpenAI and Anthropic. Palantir's stance is clear: enterprises should treat data as a core asset and AI sovereignty as a prerequisite for institutional survival, rather than entrusting their fate to the "consensus judgment" of external platforms.

Risk Warning and Disclaimer

        Market risk exists, and investment requires caution. This article does not constitute personal investment advice, nor does it consider the specific investment objectives, financial situation, or needs of individual users. Users should consider whether any opinions, views, or conclusions in this article are suitable for their specific circumstances. Investment based on this is at your own risk.