Michael Burry has once again inserted himself into one of the market’s most heated technology debates, this time by arguing that Anthropic is overtaking Palantir in the race for enterprise AI relevance. His now-deleted post may have disappeared quickly, but the reaction did not. Palantir shares fell sharply after the remark, showing that investors remain highly sensitive to any sign that the company’s position in artificial intelligence could be more vulnerable than its supporters claim.
At the center of Burry’s argument is not only valuation or sentiment, but business model. He pointed to Anthropic’s explosive growth and suggested that customers are increasingly gravitating toward tools that are easier to deploy, cheaper to adopt and more intuitive to use. In that framing, the issue is not whether Palantir has impressive clients or deep government ties. It is whether those strengths are enough in a market that may be shifting toward direct relationships with model providers.
The debate matters because Palantir has become one of the stock market’s most visible AI winners. When a prominent contrarian like Burry attacks that narrative, the reaction goes far beyond one deleted social media post. It becomes a test of what investors really believe Palantir is worth and why.
Burry says Anthropic is scaling much faster
Burry’s criticism was built around a sharp comparison. He highlighted Anthropic’s reported jump from a 9 billion dollar annual revenue run rate to 30 billion dollars in just a few months, using that growth to argue that businesses are choosing simpler and more immediately useful AI solutions. The implication was blunt: in a market moving quickly toward direct model access, Palantir may not be the company capturing the biggest new wave of demand.
That comparison is especially provocative because it contrasts speed with history. Burry reportedly pointed out that it took Palantir two decades to reach around 5 billion dollars in revenue, while Anthropic’s rise has been dramatically faster. Even if the two businesses are not directly comparable in every respect, the contrast is powerful enough to raise doubts about whether Palantir’s current valuation is pricing in the right part of the AI stack.
It is also not a new position for Burry. He has been publicly skeptical of Palantir for some time and previously disclosed a bearish options position tied to the stock. This latest attack simply puts more emphasis on a competitor he believes is changing the terms of the contest.
The real argument is about Palantir’s business model
Burry’s deeper thesis is that Palantir is not best understood as a pure software platform with unlimited operating leverage, but as a business that still relies heavily on intensive customer support and embedded engineering work. He has argued that the company’s model depends in part on sending forward deployed engineers into client environments for extended periods, making parts of the business look more like high-end consulting than a simple plug-and-play software sale.
That critique goes directly to the valuation story. If Palantir is fundamentally a more labor-intensive business, then it may deserve a lower multiple than companies whose AI products can be rolled out much faster and with less friction. Anthropic, by contrast, is being framed by Burry as a company selling more direct access to intelligence itself through a simpler application programming interface model.
This is why the debate is so intense. It is not really about whether Palantir is useful. It is about whether its usefulness scales in the way investors expect from a top-tier AI winner.
Palantir and Anthropic are not exact substitutes
Even so, the comparison is not clean. Palantir and Anthropic operate in different parts of the technology stack. Palantir is deeply embedded in operational systems for customers such as the Department of Defense and major health networks, where security, governance, workflow integration and approved data infrastructure matter enormously. Anthropic, on the other hand, provides the reasoning engine through Claude and related enterprise AI tools.
That difference is critical to the bull case on Palantir. Supporters argue that large organizations cannot simply plug an advanced model into sensitive operations without the structured, auditable and highly controlled data environment that Palantir offers. In that view, Anthropic is not eating Palantir’s lunch. It is operating alongside the infrastructure Palantir helps build and govern.
Still, Burry’s warning lands because the market may increasingly reward the companies perceived to own the intelligence layer itself. If enterprises start favoring direct model relationships and broader AI platforms over highly customized integration work, Palantir’s premium could come under pressure even if its products remain strategically important.
The Pentagon dispute added a new complication
The issue became more concrete earlier this year when Anthropic’s dispute with the Pentagon forced federal contractors to unwind some of their use of Claude in government systems. Reuters reported in March that Palantir had to remove Anthropic’s model from parts of its Maven Smart Systems platform after the Defense Department blacklisted Anthropic over a contract conflict tied to safety guardrails. That episode exposed both the overlap and the tension between the two companies.
On one hand, it showed that Anthropic had become important enough inside Palantir-linked federal systems to create disruption when it was removed. On the other, it reinforced the argument that Palantir’s deepest strength may still lie in its government-approved operational role, which can outlast or outmaneuver shifts among underlying model providers.
That is why Wall Street remains divided. Some analysts still see Palantir as a clear winner in the first phase of the AI cycle, especially within government. Others agree that the valuation is now so demanding that even strong execution may not justify much more upside in the near term.
The stock is now a test of what AI investors value most
Burry’s attack has landed because it forces investors to answer a simple but difficult question: in the AI economy, what matters more, the operating system or the reasoning engine. Palantir bulls argue that sophisticated AI cannot function at scale in sensitive environments without the secure data architecture and deployment layer Palantir provides. Burry is betting that the market will increasingly prefer the companies supplying the intelligence itself.
That is not just a disagreement over one stock. It is a wider argument about where the real value of AI will settle over time. Will it belong to those who own the interfaces, the models and the direct enterprise relationships, or to those who make the systems usable, governed and operational in the real world?
For now, the market has not fully decided. But the reaction to Burry’s comments shows that confidence in Palantir is not untouchable. The company still has powerful supporters, real traction and a strong government moat. Yet as Anthropic’s rise accelerates, the burden on Palantir to prove that its role is indispensable, and not merely helpful, is becoming much heavier.
